Introduction

Distal renal tubular acidosis (dRTA) is a disorder characterized by the inability of the collecting duct system to secrete acids during metabolic acidosis¹. In addition to hyperchloremic metabolic acidosis, patients with this disease can present with hypokalemia, kidney stones, urinary sodium waste and difficulty to thrive. Expression of pathogenic variants in the ATP6V0A4, ATP6V1B1, FOXI1, WDR72 and SLC4A1 genes are the usual genetic aetiologies^2–4. The SLC4A1 gene encodes the anion exchanger 1 (AE1) protein, which is an electroneutral chloride/bicarbonate exchanger⁵. It exists in two forms: a 911 amino acid erythroid isoform known to interact with erythroid cytoskeletal proteins and participate in red cell respiration and integrity, and a 65 amino acid (NH2-terminal) truncated isoform primarily found in the basolateral membrane of renal type A intercalated cells (A-IC)^6,7 and podocytes⁸. This isoform participates in bicarbonate reabsorption and through its physical and functional interaction with the cytosolic carbonic anhydrase II and apical H⁺-ATPase, it supports apical proton export and urine acidification⁹.

Pathogenic variants in the SLC4A1 gene can result in either red cell defects (such as Southeast Asian ovalocytosis^10,11 and hereditary spherocytosis¹²), renal cell defects (dRTA¹⁰) or both in patients with homozygous (Band 3 Coimbra and Band 3 Courcouronnes^13,14) or compound heterozygous variants^15,16. Renal SLC4A1 disease-causing variants have only been found in the transmembrane domain - where it could impact protein structure and its transport function- or in the short carboxyl domain – where it possibly affects protein-protein interactions. The pathophysiology of dominant or recessive SLC4A1 related dRTA (hereafter named dRTA) has originally been linked with the mis-trafficking defect of mutant kAE1 protein^10,17. However, recent in vivo studies in mice and humans have revealed a complex pathophysiology where a loss of kAE1-expressing intercalated cells, the intracellular relocation of the H⁺-ATPase and accumulation of autophagy marker p62 and ubiquitin-positive material in the remaining type-A intercalated cells was suggested to be the principal cause of the disease¹⁸.

The highly active transport properties of collecting duct cells require the maintenance of cellular energy and homeostasis. The autophagy-mediated turnover of damaged organelles is necessary for protecting collecting duct cells as in most renal cells¹⁹. The chloride/bicarbonate exchange function of kAE1 in A-ICs confers a pivotal role in pH homeostasis and thus is a major contributor to cellular homeostasis. kAE1 protein interacts with several proteins such as integrin-like kinase (ILK), adaptor-related protein complex 1, 3 & 4 (AP-1, AP-3 & AP-4 mu1A), transmembrane protein 139 (TMEM139), kinesin family member 3B (KIF3B), and clathrin among others^7,20–22 that support protein stability and trafficking. It also interacts with homeostatic proteins including the glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH)²³ and the antioxidant enzyme peroxiredoxin 6 (PRDX 6)²⁴, which play major roles in cellular energy metabolism and oxidative stress response, respectively.

In this study, we report the characterization of newly identified dRTA variations and provide evidence of abnormal autophagy and degradative pathways in cells and kidneys from mice expressing dRTA mutant kAE1 proteins.

Concise Methods

Ethics Approval

This study was conducted in accordance with all national and institutional animal care guidelines and approved by the University of Alberta’s Animal Care and Use Committee (AUP #1277).

Newly identified SLC4A1 variations from dRTA patients

The patient carrying the R295H mutation was a boy carrying the variation in the homozygous state, whose genetic diagnosis was made at the age of 5, following growth retardation of -2 SD for both weight and height, with bicarbonate at 18 mmol/L, potassium at 2.8 mmol/L, chloridemia at 94 mmol/L, calcemia at 2.55 mmol/L, and a urinary pH of 7.5. He also had a history of a pyeloureteral junction syndrome that was surgically managed at the age of 3. The R295H dRTA variation is a nonsense homozygous substitution characterized by a replacement of guanine (G) on position 884 by adenine (A) in the coding sequence. It has an allelic frequency of 0.14 % in the European population.

For the Y413H variant, the patient was a female diagnosed at 1 month old, with an urinary pH of 7.5, evidence of nephrocalcinosis and failure to thrive. The S525F variant has been previously reported²⁵, but in brief, the patient was a 13 year old female with plasma pH of 7.25, plasma bicarbonate at 15.3 mmol/L who also presented with polyurethral junction syndrome, nephrocalcinosis and nephrolithiasis since childhood. The Y413H and S525F dRTA variations are nonsense heterozygous substitutions characterized by a replacement of thymine (T) at position 1237 by cytosine (C) and a replacement of cytosine (C) at position 1574 by thymine (T), respectively. The R589H dRTA variation has been previously described¹⁸. No follow up data were available for all patients.

Mice

Transgenic mice carrying the R607H knockin (KI) mutation (murine equivalent to human R589H mutation) were previously described¹⁸. Homozygous mice used throughout the study display incomplete dRTA with alkaline urine without metabolic acidosis at baseline as previously reported¹⁸. Homozygous mice or wild-type (WT) littermates were fed a standard rodent chow (Picolab® Rodent Diet 20 # 5053, LabDiet, ST. Louis, MO, USA) or for Figure 5 (H-K), a salt-depleted diet with acid challenge as previously reported²⁶ with adequate and constant water supply, and maintained on a 12-hour light and dark cycle throughout their lifespan.

Statistical analysis

All the experiments were independently repeated a minimum of three times. Experimental results were analyzed using the GraphPad Prism software and are summarized as mean ± SEM. All statistical comparisons were made using unpaired Student’s t test or one/two-way ANOVA followed by a post-hoc test as indicated in figure legends. A P-value less than 0.05 was considered statistically significant. All datasets were assessed for normality, and outliers identified by Prism were excluded.

Detailed Material and Methods are listed in the Supplementary Material

Results

The dRTA kAE1 variants traffic to the basolateral membrane but have reduced transport activity in mIMCD3 cells

We first characterized three newly identified dRTA mutations and compared them with kAE1 WT or previously characterized kAE1 R589H mutant. Figure 1A depicts the alpha fold predicted structure of kAE1 showing amino acids mutated in each kAE1 mutant. kAE1 R295H is a recessively inherited substitution in the N-terminal cytosolic domain of the protein. kAE1 Y413H is a dominantly inherited substitution in transmembrane domain (TM) 1 of the core domain. In the gate domain, S525F and R589H substitutions occur in TM 5 and TM 6, respectively. We generated the newly identified dRTA variant cDNAs and expressed them or kAE1 WT in mIMCD3 cells to assess protein abundance (Figure 1B), localization (Figure 1C, D), transport activity (Figure 1E) and lifespan (Figure 1F). As seen on Figure 1B, both kAE1 R295H and S525F variants display 2 typical bands similar to kAE1 WT. The top band encompasses kAE1 proteins carrying a complex oligosaccharide that have reached the Golgi and beyond, while the bottom band corresponds to high mannose-carrying kAE1 proteins located in the endoplasmic reticulum. However, kAE1 Y413H mutant bands intensity was overall weaker than WT and displayed a predominant single band aligned with high mannose-carrying kAE1 proteins. These results indicate that the 3 newly described dRTA mutants are expressed in mIMCD3 cells. We next localized these mutants by immunofluorescence in mIMCD3 cells. While kAE1 WT is targeted to the plasma membrane, the kAE1 R295H, Y413H and S525F variants partially colocalize with endoplasmic reticulum marker calnexin (Figure 1C). Yet, both kAE1 WT and mutants appropriately co-localized with basolateral membrane marker beta-catenin in polarized mIMCD3 cells (Figure 1D). However, when the transport function was assayed on confluent cells, both kAE1 Y413H and S525F had significantly lower transport activity than WT (Figure 1E) and were not significantly different from un-induced WT-Dox cells (not shown on the graph). Finally, we observed that degradation of the kAE1 Y413H mutant begins 6 hours post-synthesis while kAE1 WT abundance remained stable for 24 hours (Figure 1F). Given this premature degradation, we did not perform further assays on cells expressing the Y413H protein. Overall, these results indicate that except for kAE1 Y413H mutant, the other mutants are expressed and traffic to the basolateral membrane of polarized mIMCD3 cells, similar to the previously published kAE1 R589H mutant.

Figure 1.

Expression of the kAE1 R295H, S525F and R589H dRTA mutants alter autophagy processes in mIMCD3 cells and whole kidney lysates

In mice expressing kAE1 R607H, a striking reduction in type-A intercalated cells was noted and in the remaining cells, autophagy marker p62 and ubiquitin accumulated in these abnormally enlarged cells¹⁸. We therefore investigated the autophagy machinery in dRTA mutant mIMCD3 cells and in R607H knockin (KI) mice. We first examined the ratio of autophagosome marker LC3BII protein relative to the total intensities of LC3BI and LC3BII as well as p62 levels. These experiments were performed at steady state, upon autophagy induction by starvation, or after autophagy inhibition by Bafilomycin (Baf) A1 (Figure 2A-C). Figure 2A – I show a consistent increase in the ratio of LC3B II to total LC3B (I + II) in the mutant cells at steady state (panel D), with starvation (panel F) and with Baf A1 (panel H) except for the kAE1 R295H mutant which was not significantly different from WT at steady state only. This suggests an altered autophagy process in the mutant cells, in agreement with preliminary findings from Mumtaz and colleagues¹⁸. There was no significant difference in p62 abundance in mutant cells compared to WT at steady state and with starvation (Figure 2E-G). However, with Baf A1, R589H mutant cells had significantly lower p62 abundance compared to WT (Figure 2I). To confirm these findings, we next assessed the abundance of these markers in whole kidney lysates from the kAE1 R607H KI mice. We observed that the total LC3B abundance was significantly higher in homozygous KI mice (KI/KI) compared to WT, with no difference in p62 (Figure 2J-L). Overall, these results suggest an abnormal autophagy in mutant mIMCD3 cells and in kidneys of R607H KI mice.

Figure 2.

dRTA kAE1 mutant expressing cells have more alkaline intracellular pH than WT and altered autophagy flux

Given these preliminary findings of abnormal autophagy in the dRTA mutant cells, we next assessed the efficiency of the autophagy machinery in WT or dRTA mutant-expressing cells using the eGFP-RFP-LC3 construct (Figure 3A). The green fluorescent protein fused to LC3 is quenched in the acidic environment of the autolysosome, while both eGFP and RFP fluoresce in vesicles in the neutral lumen of the autophagosome^27,28. Focusing on cells expressing either kAE1 S525F or R589H, we quantified the number of eGFP+ (green), RFP+ (red) and double eGFP+/RFP+ (yellow) vesicles in the kAE1 mutant or WT transfected live cells (Figure 3B-D). kAE1 S525F mutant cells have significantly more autophagosomes (yellow) than WT counterparts (Figure 3B), and both kAE1 S525F and R589H have significantly more autolysosomes (red) than WT (Figure 3C). Figure 3D shows that both mutants, particularly the kAE1 S525F mutant cells had significantly more autophagosomes than autolysosomes. This finding suggests an upregulation of autophagy and inhibition in the downstream steps of autophagy that involves the fusion of the autophagosome with the lysosome²⁹. As autolysosomes require luminal v-H⁺-ATPase -dependent acidification to efficiently clear cell debris ³⁰ and as kAE1 is a base transporter whose function is linked to the v-H⁺-ATPase, we next wondered whether the steady-state cytosolic pH of mIMCD3 cells expressing kAE1 WT or mutant was different. Using BCECF-AM, we observed that the steady state intracellular pH (pHi) of kAE1 mutant cells was more alkaline than WT cells (except for kAE1 R295H cell pHi which has a similar trend but is not significantly different from WT) (Figure 3E), in agreement with transport activity (Figure 1F). Given the lack of difference in phenotype between the kAE1 R295H and kAE1 WT mIMCD3 cells, we focused the subsequent experiments on kAE1 S525F and R589H mutant cells. As a higher intracellular pH could contribute to alterations in the autophagy machinery³¹, we chemically acidified the cytosolic pH of cells expressing kAE1 mutant or WT and examined autophagy and lysosomal markers (Figure 3 F-H). We first determined that incubation of mIMCD3 cells in 0.033 μM nigericin for 2 hours acidified cytosolic pH to 6.9 without causing cell death (Supplementary Figure 1). We observed that chemically reducing pHi to 6.9 in mutant expressing cells reduced the ratio of LC3B II to total LC3B (Figure 3I) and the abundance of lysosomal-associated membrane protein 1 (LAMP1) in R589H cells (Figure 3 J) to levels similar to WT cells at steady state. These findings suggest that abnormal autophagy in the mutant cells may be caused by their alkaline pHi, resulting from a reduced anion exchange activity of the mutant kAE1 protein.

Figure 3.

mIMCD3 cells expressing dRTA kAE1 mutants and R607H KI kidney tissues have abnormal lysosome number and size

The accumulation of autophagosomes and autolysosomes as seen above may suggest one or a combination of the following: an inability of autophagosomes to fuse with lysosomes and/ or a defect in lysosomal degradative activity in the mutant cells^19,32. We first examined the lysosomal degradative activity by assessing lysosomal protease Cathepsin B activity using Magic Red staining (Figure 4A-B). In agreement with increased RFP+ vesicles (Figure 3D), the kAE1 S525F mutant cells had a significantly higher number of Magic Red positive vesicles than WT whereas the kAE1 R589H mutant cells had significantly larger Magic Red positive vesicles, suggesting an accumulation of undigested material³³. To validate this finding, we performed immunostaining and quantified LAMP1 positive staining in ß1 ATPase positive cells (a marker of ICs) in WT and R607H KI mouse kidney sections. The KI mice showed significantly more and larger LAMP1 positive vesicles compared to WT mice in both cortex and medulla (Figure 4 C-F). We probed further into the lysosomal activity by quantifying lysosomal protease Cathepsin D (immature, intermediate and mature) protein abundance by immunoblot in isolated primary murine A-ICs. Although the abundance of immature and intermediate cathepsin D did not differ between genotypes, the KI mice showed a significantly decreased abundance of mature cathepsin D (Supplementary Figure 2). Thus, in line with in vitro findings, A-IC from homozygous R607H KI mice display relatively more and larger lysosomes with reduced active protease abundance than WT littermates, suggesting a lysosomal defect in the dRTA kAE1 mutant cells.

Figure 4.

Figure 5.

dRTA kAE1 mutant cells have lower ATP production rate and abnormal mitochondrial content

Lysosomal degradation is highly dependent on a low luminal pH generated in part by the vacuolar-type H⁺-ATPase³⁴ whose activity depends on ATP hydrolysis. We therefore analysed the ATP production rate in mIMCD3 cells, specifically glycolysis and oxidative phosphorylation. We measured the oxygen consumption rate (OCR) (Figure 5A) and extracellular acidification rate (ECAR) in empty vector transfected, kAE1 WT or mutant cells (Figure 5B). Both kAE1 S525F and R589H mutant cells had a lower ATP production rate compared to WT (Figure 5C). More specifically, the R589H mutant cells had a lower mitochondrial ATP production rate (Figure 5D) whereas the kAE1 S525F mutant cells exhibited a lower glycolytic ATP production rate (Figure 5E). With the mitochondria being the major ATP-producing organelles³⁵, we assessed mitochondrial content by immunostaining of translocase of the outer membrane 20 (TOM20) both in vitro and in vivo. Both kAE1 S525F and R589H mutant cells have higher mitochondrial content compared to WT as determined by total overall intensity of TOM20 positive puncta (Figure 5F&G). In line with this result, although a decreased fluorescence intensity was observed in the cortex, there was a significantly higher TOM20 fluorescence intensity in medullary kidneys of homozygous R607H KI mutant mice compared to WT littermates (Figure 5H-K).

Discussion

In this study, we characterized three newly identified dRTA-causing kAE1 variations. Combining in vivo and in vitro studies, we demonstrated that reduced transport activity of the kAE1 mutants correlated with increased cytosolic pH, reduced ATP synthesis, attenuated downstream autophagic pathways and lysosomal dysfunction, pertaining to the fusion of autophagosomes and lysosomes and/ or lysosomal degradative activity.

In line with previous observations¹⁸, the kAE1 R295H, Y413H and S525F mutants were intracellularly retained in non-polarised mIMCD3 cells but properly localized to the basolateral membrane after polarization. When expressed in Madin-Darby canine kidney (MDCK) cells, other dRTA-causing kAE1 mutants such as dRTA R602H, G701D, V488M, deltaV850 variants also exhibited a plasma membrane trafficking defect^10,36,37. In contrast, the kAE1 R589H mutant was correctly targeted to the plasma membrane¹⁸. Functionally, cells expressing the Y413H and S525F mutants exhibit about 40 % reduction in chloride/bicarbonate exchange activity compared to kAE1 WT, similar to the previously characterized recessive G701D mutant but unlike the R295H mutant³⁸. Therefore, the mechanism causing dRTA remains unclear in the case of the newly identified R295H variant. These findings add to the growing list of SLC4A1 gene variations causing dRTA.

We next investigated the roots of the altered autophagy briefly reported in R607H KI mice¹⁸ using mIMCD3 cells and whole kidney lysates. In kidneys of the KI mice, a decrease in A-IC, accumulation of p62 and ubiquitin and enlarged remaining A-ICs, prompted us to investigate autophagy pathways in dRTA mutant cells and in homozygous R607H KI mice^18,38. During the autophagy process, LC3B I (a marker for autophagosomes) is converted to lipidated LC3B II³⁹ and p62 aggregates to facilitate the degradation of ubiquitinated proteins within the autophagosome complex⁴⁰. In mIMCD3 cells, LC3B lipidation was increased in the kAE1 R295H, S525F and R589H mutant cells compared to WT, an increase that persisted with both Baf A1 treatment and starvation. Although opposite effects were expected under inhibition or induction of autophagy, such similar effect has been previously described. In the proximal tubule of obese mice, LC3B accumulation indicating a stagnated autophagy flux, was observed with both chloroquine treatment and 24 hour starvation⁴¹. Similarly, in NRK-52E cells, a disruption of the autophagy machinery by LC3B II accumulation in high cadmium-stressed cells occurred under either Baf A1 or rapamycin (an autophagy inducer) treatment⁴². Although LC3B II is elevated during both increased autophagy flux and disrupted autophagy, we did not observe significant accumulation of p62 in mIMCD3 cells expressing dRTA mutants. p62 is specifically a marker of autophagy-mediated protein clearance^43–45. Therefore, p62 accumulation in R607H KI mouse kidney sections strongly suggests a compromised autophagy-mediated clearance while the increased LC3B lipidation without significant changes in p62 in mIMCD3 cells points towards either an increased autophagic flux and/or a disrupted autophagy.

To obtain a clearer picture of the precise autophagic pathway altered in the dRTA mutants, we probed further into the different stages of autophagy and autophagy flux. We noted an accumulation of autophagosomes and autolysosomes in the S525F and R589H mutant cells. This was recapitulated in the R607H KI mice which showed significantly more and bigger LAMP 1 positive vesicles in both kidney cortex and medulla, suggesting a blockage in autophagy flux in both dRTA mutant cells and KI mice. Such blockage has been implicated in the pathophysiology of several diseases. In lysosomal storage disease, lysosome accumulation in proximal tubule cells is a key component in the pathways mediating epithelial dysfunction¹⁹. In this study, restoring autophagy flux attenuated disease progression. Another study in SK-N-SH, RT4-D6P2T and HeLa cells implicated autophagosome and lysosome accumulation in cellular toxicity associated with neurodegenerative diseases³². In agreement with altered lysosomal function, we also noted a greater abundance and size of active cathepsin B lysosomal protease vesicles in mIMCD3 cells. Increased cathepsin B activity affects lysosomal biogenesis, autophagy initiation and cellular homeostasis⁴⁶. In the renal context, cathepsin B knockout mice demonstrated a higher resistance and quicker recovery from glomerular damage⁴⁷, suggesting that cathepsin B accumulation may be detrimental to the cells. Increased cathepsin B abundance in the dRTA mutant cells also correlates with the accumulation of lysosomes. Overall, these findings suggest that the pathogenesis of dRTA in our models involves an inhibition of autophagy flux at the downstream steps involving autophagosome-lysosome fusion and lysosomal protein clearance ^29,33,48,49.

The question remained as to how these dRTA variants altered autophagy. The SLC4A1-3 gene family that includes AE1 are regulators of intracellular pH in different cell types^50–52. The reduced anion exchange activity in mIMCD3 cells expressing the R589H and S525F variants expectedly correlated with an increased pHi compared to WT counterparts. We wondered whether this was the mechanism explaining blockade in autophagy flux since pHi variations impact autophagy^{31,34,53–55}. Previous studies reported more perinuclear localization of lysosomes and autophagosome-lysosome fusion in cells with an increased pHi^53,54. Although not examined in our study, an increase in intracellular pH due to starvation decreased the levels of lysosomal kinesin superfamily member KIF2 and ADP-ribosylation factor-like 8B (ARL8), which are responsible for redistributing lysosomes to the cell periphery. This reduction subsequently inhibited mTORC1, resulting in increased autophagosome synthesis and autophagosome-lysosome fusion⁵³.

While an alkaline cytosolic pH partially explains the impairment in autophagy flux, a close relationship also exists between intracellular pH and cellular energy dynamics and metabolic stress, all of which are key regulators of autophagy⁵⁶. Therefore, it was plausible that kAE1 variant-induced autophagy dysregulation occurs through a signalling pathway akin to that of energy deprivation-induced autophagy. This hypothesis is supported by our findings of reduced ATP production rate in dRTA kAE1 S525F and R589H mutant cells compared to kAE1 WT cells. We also found that both kAE1 S525F and R589H mutant cells have higher mitochondrial content compared to kAE1 WT cells. The increased mitochondrial content coupled with low ATP point towards improper mitochondrial function in dRTA variant cells. Low ATP levels as seen in dRTA mutant cells may impair autophagy as was shown in human RPE cells⁵⁷. ATP reduction in RPE cells led to complex mitochondrial changes such as structural disorganization, enzyme activity decline, and oxidative damage to mitochondrial components and DNA⁵⁷. Similarly, in pancreatic islet cells, an alkaline pHi led to increased uptake of phosphate by mitochondria, accelerating the production of superoxide, promoting mitochondrial permeability transition, and inducing translational attenuation due to endoplasmic reticulum stress, ultimately impairing insulin secretion⁵⁸. Overall, our data support that expression of the kAE1 variants increases pHi which alters mitochondrial function and leads to reduced cellular energy levels that eventually attenuates energy-dependent autophagic pathways including autophagosome-lysosome fusion and lysosomal protein clearance.

In light of these observations, we postulated that correcting the alkaline pHi of dRTA mutant-expressing mIMCD3 cells will alleviate this blockage in autophagy flux. We observed that a chemically engineered pHi of 6.9⁵⁹ reduced LC3B II accumulation and LAMP1 abundance in mIMCD3 mutant cells to expression levels similar to that of WT cells at baseline. This suggests that a chemically reduced pHi facilitated protein clearance in the two dRTA mutant cells^31,53 as noted in other studies. In one such study, treatment of SH-SY5Y cells with FCCP and nigericin also acidified intracellular pH and triggered autophagy and mitophagy³¹. Similarly, acid loading in proximal tubular cells under chronic metabolic acidosis showed increased autophagic flux and mitophagy⁶⁰. These findings are in line with our results and establish a link between altered autophagy flux and the alkaline pHi of dRTA variant cells.

In conclusion, our study established a strong relationship between the expression of defective kAE1 proteins, reduced mitochondrial activity, decreased autophagy and impaired protein degradative flux. Whether this abnormal degradative pathway explains the premature loss of A-IC will need to be elucidated in further studies.

Acknowledgements

We thank Kristina MacNaughton, Jared Bouchard, Kiera Smith and Hilmar Strickfaden for excellent technical assistance. Imaging experiments were performed at the University of Alberta Faculty of Medicine & Dentistry Cell Imaging Core, RRID:SCR_019200, which receives financial support from the Faculty of Medicine & Dentistry, the University Hospital Foundation, Striving for Pandemic Preparedness – The Alberta Research Consortium, and Canada Foundation for Innovation (CFI) awards to contributing investigators. Services were provided by the University of Alberta Faculty of Medicine & Dentistry Workshop, RRID:SCR_019181, which receives financial support from the Faculty of Medicine & Dentistry.

Additional information

Funding

This study was funded by the Canadian Institutes of Health Research (PJT#168871) and the Kidney Foundation of Canada (2020KHRG-666615) to E.C, by a grant from the Deutsche Forschungsgemeinschaft to M.J.S. (IRTG 1830) and the Canadian Institutes of Health Research (PS#165816) to N.T. G.E. received a Graduate Student Engagement Scholarship, a Faculty of Medicine and Dentistry Delnor Scholarship and a Faculty of Medicine and Dentistry 75^th Anniversary award. M.R. received a Sir Frederick Banting and Dr. Charles Best Canada Graduate Scholarship-Master’s (CGS-M) from the Canadian Institutes of Health Research; Walter H. Johns Graduate Fellowship; a University of Alberta Faculty of Medicine and Dentistry/Alberta Health Services Graduate Student Recruitment Studentship (GSRS) and an Alberta Graduate Excellence Scholarship (AGES). A.K.M.S.U. received an NSERC CREATE graduate studentship. F.C. was supported by a Discovery Grant to E.C. from the Natural Sciences and Engineering Research Council (RGPIN-2017-06432), and was awarded a Graduate Recruitment scholarship from the University of Alberta. S.M.A.H. received a PhD scholarship from the DAAD (Deutscher Akademischer Austauschdienst).

Additional files

Supplementary Material and Figures