The Human Mutation K237_V238del in a Putative Lipid Binding Motif within the V-ATPase a2 Isoform Suggests a Molecular Mechanism Underlying Cutis Laxa

Vacuolar ATPases (V-ATPases), proton pumps composed of 16 subunits, are necessary for a variety of cellular functions. Subunit “a” has four isoforms, a1–a4, each with a distinct cellular location. We identified a phosphoinositide (PIP) interaction motif, KXnK(R)IK(R), conserved in all four isoforms, and hypothesize that a/PIP interactions regulate V-ATPase recruitment/retention to different organelles. Among the four isoforms, a2 is enriched on Golgi with a2 mutations in the PIP motif resulting in cutis laxa. We hypothesize that the hydrophilic N-terminal (NT) domain of a2 contains a lipid-binding domain, and mutations in this domain prevent interaction with Golgi-enriched PIPs, resulting in cutis laxa. We recreated the cutis laxa-causing mutation K237_V238del, and a double mutation in the PIP-binding motif, K237A/V238A. Circular dichroism confirmed that there were no protein structure alterations. Pull-down assays with PIP-enriched liposomes revealed that wildtype a2NT preferentially binds phosphatidylinositol 4-phosphate (PI(4)P), while mutants decreased binding to PI(4)P. In HEK293 cells, wildtype a2NT was localized to Golgi and co-purified with microsomal membranes. Mutants reduced Golgi localization and membrane association. Rapamycin depletion of PI(4)P diminished a2NT-Golgi localization. We conclude that a2NT is sufficient for Golgi retention, suggesting the lipid-binding motif is involved in V-ATPase targeting and/or retention. Mutational analyses suggest a molecular mechanism underlying how a2 mutations result in cutis laxa.

The 90kDa a-subunit contains a cytosolic N-terminal domain and a membrane-bound C-terminal domain consisting of eight transmembrane helices (TMs) [10,38,39].In Saccharomyces cerevisiae, the sorting signal for V-ATPases within the N-terminal half of the a-subunit homolog, Stv1p, is well characterized [10,11].The sorting information, W 83 KY, present in the Stv1p isoform dictates the retention of Stv1p-containing V-ATPases within the Golgi network [10].However, little is known about the targeting signals for the mammalian Golgi-specific a2 isoform.Evidence in yeast suggests the interaction of the a subunits with membrane phosphoinositides can account for the membrane retention of the V-ATPases at specific locations [40,41].
Phosphoinositides (PIPs) are generated by reversible phosphorylation of the precursor phosphatidylinositol (PI) at the inositol headgroup, mediated by organelle-specific kinases and phosphatases [42][43][44].Organelle-specific distribution and composition of the seven different PIPs play important roles in membrane protein trafficking [42,45].Spatial or temporal enrichment of PIPs at microdomains within organelle membranes is involved in regulating the activity of membrane-bound ion channels and transporters [45,46].The cytosolic N-terminal domain of the yeast a isoform, Vph1p, was recruited to the vacuolar membrane when the level of endosome/vacuole-specific PI(3,5)P 2 was elevated [41].In contrast, Stv1p was shown to bind directly to a Golgi-specific PI(4)P, and the interaction was attributed to its cytosolic N-terminal domain [40].
In previous work, we showed that the plasma membrane-specific a4 isoform preferentially binds to PI(4,5)P 2 and that the a4NT-PI(4,5)P 2 interaction is responsible for membrane retention [47].We proposed a putative binding domain within the N-terminal half of the a subunit, containing a conserved basic motif (K/R)X(K/R)(K/R).In the present study, we found that the a2 isoform interacts in vitro with PI(4)P, a Golgi-enriched PIP, and show that PI(4)P at the Golgi helps to retain the cytosolic N-terminal domain of a2 at the Golgi membrane.Additionally, we show that the cutis laxa causing mutation K237_V238del within the critical a2 isoform K 237 VKK 240 binding motif not only reduced interactions with PI(4)P but also disrupted protein membrane retention, suggesting the molecular mechanism underlying the disease.

Mutations within the Putative Binding Motif reduced Interaction of a2 with PI(4)P-Enriched Liposomes In Vitro
In our previous work, we proposed a putative lipid binding domain located at the distal lobe of the N-terminal half of a4.We further suggested that a conserved basic motif, (K/R)X(K/R)(K/R), is critical for the interaction with the acidic headgroup of PIPs [47].In a2, we propose this critical lipid binding motif is K 237 VKK 240 (Figure 1).Using mutagenesis to verify the basic motif, we generated two mutations in the a2 N-terminal domain (a2NT), K237A/V238A and K237_V238del, both within the putative binding motif.The K237_V238del is a mutation found within humans that results in cutis laxa [48].
We first assessed whether either mutation affected protein folding using circular dichroism (CD) spectroscopy.WT a2NT contains a mainly helical structure, exhibited by two negative minima at 222 nm and 208 nm.The CD spectra of the mutants and WT, all in aqueous buffer, overlap at the characteristic wavelengths, indicating that the mutations did not alter protein structure (Figure 2A).To mimic the membrane environment and determine whether the presence of micelles enhances the helicity of the proteins, we added detergent with an SDS-to-peptide ratio of 370:1.The spectra of WT and mutants behaved similarly in the presence of SDS micelles with increased negative ellipticity at 208 nm and 222 nm, suggesting the membrane-bound behavior of the proteins (Figure 2B).Variation in the signal of the positive peak at 190-200 nm is likely due to different trace amounts of imidazole in the buffer.We first assessed whether either mutation affected protein folding using circular dichroism (CD) spectroscopy.WT a2NT contains a mainly helical structure, exhibited by two negative minima at 222 nm and 208 nm.The CD spectra of the mutants and WT, all in aqueous buffer, overlap at the characteristic wavelengths, indicating that the mutations did not alter protein structure (Figure 2A).To mimic the membrane environment and determine whether the presence of micelles enhances the helicity of the proteins, we added detergent with an SDS-to-peptide ratio of 370:1.The spectra of WT and mutants behaved similarly in the presence of SDS micelles with increased negative ellipticity at 208 nm and 222 nm, suggesting the membrane-bound behavior of the proteins (Figure 2B).Variation in the signal of the positive peak at 190-200 nm is likely due to different trace amounts of imidazole in the buffer.
(A) (B)  We first assessed whether either mutation affected protein folding using circular dichroism (CD) spectroscopy.WT a2NT contains a mainly helical structure, exhibited by two negative minima at 222 nm and 208 nm.The CD spectra of the mutants and WT, all in aqueous buffer, overlap at the characteristic wavelengths, indicating that the mutations did not alter protein structure (Figure 2A).To mimic the membrane environment and determine whether the presence of micelles enhances the helicity of the proteins, we added detergent with an SDS-to-peptide ratio of 370:1.The spectra of WT and mutants behaved similarly in the presence of SDS micelles with increased negative ellipticity at 208 nm and 222 nm, suggesting the membrane-bound behavior of the proteins (Figure 2B).Variation in the signal of the positive peak at 190-200 nm is likely due to different trace amounts of imidazole in the buffer.
(A) (B) We tested in vitro PIP interactions of wildtype and mutants with PIP-enriched liposomes using a liposome pull-down assay.Wildtype and mutants were incubated with PolyPIPosome liposomes enriched with different PIPs, including the Golgi-specific PI(4)P, and its derivatives PI(3,4)P 2 ,PI(4,5)P 2 and PI(3,4,5)P 3 (labelled as PIP3).The proteinliposome complexes are collected via high-speed centrifugation and resolved by Western blot.WT a2NT showed significantly higher binding to PI(4)P compared to PI(3,4)P 2 and PIP3 (Figure ??A,B).While both mutations visually appear to reduce association with all polyPIPosome liposomes (Figure ??B left panel), only the liposomes enriched with PI(4)P resulted in a significant reduction (Figure ??B right panel).The binding differences between WT and mutants to I(3,4)P 2 , PI(4,5)P 2 , and PIP3 were not statistically significant (Figure ??C).This result is consistent with the fact that the a2 isoform is localized in Golgi and that PI(4)P is primarily found in Golgi.
WT and mutants to I(3,4)P2, PI(4,5)P2, and PIP3 were not statistically significant (Figure 3C).This result is consistent with the fact that the a2 isoform is localized in Golgi and that PI(4)P is primarily found in Golgi.
These data suggest that the basic motif K 237 VKK 240 in a2 is required for PI(4)P interaction, and mutations within this motif negatively impact the interaction.This further suggests that disruption of the a2-PI(4)P interaction could be the molecular mechanism underlying the disease-causing mutation, K237_V238del.

Mutations K237A/V238A and K237_V238del Reduce a2NT Golgi Localization
To assess cytosolic a2NT membrane association, we expressed FLAG-tagged These data suggest that the basic motif K 237 VKK 240 in a2 is required for PI(4)P interaction, and mutations within this motif negatively impact the interaction.This further suggests that disruption of the a2-PI(4)P interaction could be the molecular mechanism underlying the disease-causing mutation, K237_V238del.

Mutations K237A/V238A and K237_V238del Reduce a2NT Golgi Localization
To assess cytosolic a2NT membrane association, we expressed FLAG-tagged wildtype a2NT and mutants a2NT K237A/V238A and K237_V238del in HEK293 cells and performed subcellular fractionation.Although the a2NT.FLAG lacks a transmembrane domain, it was still detected in microsomal fractions, suggesting that the membrane-bound C-terminal domain is not essential for membrane retention and that the N-terminal domain is sufficient to bring the protein to the membrane (Figure 4A).There was a significant decrease in the amount of both mutants in the microsomal fraction compared to the wildtype (Figure 4B).These results indicate that both mutations disrupt PI(4)P binding in vitro and reduce membrane retention in vivo.Immunofluorescence microscopy was used to visualize the localization of wildtype and mutant a2NT in the HEK293 cells.Wildtype a2NT.FLAG (red) were enriched at the Golgi, visualized with the Golgi specific marker, Tgn38-CFP (Figure 5A, top row).As a2NT was recruited to the Golgi in the absence of a membrane-bound C-terminal domain, this suggests that Golgi sorting information lies within the cytosolic N-terminal half.In contrast, there was a significant reduction in the Golgi localization of a2.K237A/V238A (Figure 5A, middle row) and a2.K237_V238del (Figure 5A, bottom row) (Figure 5B).This result aligns with the diminished membrane retention of the two mutants in microsomal fractions, supporting the hypothesis that the PI(4)P binding motif in a2 is, in part, responsible for Golgi membrane targeting/retention.Immunofluorescence microscopy was used to visualize the localization of wildtype and mutant a2NT in the HEK293 cells.Wildtype a2NT.FLAG (red) were enriched at the Golgi, visualized with the Golgi specific marker, Tgn38-CFP (Figure 5A, top row).As a2NT was recruited to the Golgi in the absence of a membrane-bound C-terminal domain, this suggests that Golgi sorting information lies within the cytosolic N-terminal half.In contrast, there was a significant reduction in the Golgi localization of a2.K237A/V238A (Figure 5A, middle row) and a2.K237_V238del (Figure 5A, bottom row) (Figure 5B).This result aligns with the diminished membrane retention of the two mutants in microsomal fractions, supporting the hypothesis that the PI(4)P binding motif in a2 is, in part, responsible for Golgi membrane targeting/retention.

Depletion of Golgi PI(4)P Impairs a2NT Recruitment to Golgi
We next tested whether the depletion of Golgi PI(4)P impairs a2NT Golgi recruitment.Sac1 is a PI(4)P phosphatase converting PI(4)P to PI [49,50].We recruited Sac1 phosphatase to Golgi using the rapamycin-induced dimerization method [51-53].Sac1 phosphatase coupled to FK506 binding protein FKBP (Sac1-FKBP) and Golgi membrane anchor Tgn38 coupled to FKBP-rapamycin binding domain FRB (Tgn38-FRB-CFP) were dimerized by the addition of rapamycin (Figure 6A).Sac1-PJ phosphatase is recruited to the Golgi (Figure 6B), where it converts PI(4)P to PI, resulting in the depletion of the Golgi pool of PI(4)P.

Discussion
The cytosolic N-terminal domain of the a subunit serves as a connector for V1 and Vo assembly [54] and is an important target for multiple V-ATPase regulators [15,17,20,21].Experiments with chimeras of the two yeast orthologs, Vph1p and Stv1p, showed that the aNT contains information for both the localization and regulation of V-ATPase by reversible assembly [13].Phosphoinositides regulate transmembrane channels and transporters [42,45,46].Interactions between a subunit isoforms and different PIPs may impact both functional regulation and localization, which, in turn, could account for the differences in functional destinations among a subunit isoforms.

Discussion
The cytosolic N-terminal domain of the a subunit serves as a connector for V 1 and V o assembly [54] and is an important target for multiple V-ATPase regulators [15,17,20,21].Experiments with chimeras of the two yeast orthologs, Vph1p and Stv1p, showed that the aNT contains information for both the localization and regulation of V-ATPase by reversible assembly [13].Phosphoinositides regulate transmembrane channels and transporters [42,45,46].Interactions between a subunit isoforms and different PIPs may impact both functional regulation and localization, which, in turn, could account for the differences in functional destinations among a subunit isoforms.
Previously, we proposed a putative lipid binding domain in the a subunit and the possible involvement of this domain in V-ATPase regulation [47].Sequence alignment of the four mammalian V-ATPase a subunit isoforms with the two yeast orthologs revealed a conserved basic motif KX n K(R)IK(R) required for the PIP's interaction in the N-terminal domain of a4.Here, we provide evidence that the Golgi-specific a2 isoform directly interacts with PI(4)P, a Golgi-enriched PIP.Our results with a2 mutations, K237A/V238A and K237_V238del, indicate that mutations within the binding motif (K 237 VKK 240 ) compromised the membrane association and Golgi localization, further suggesting that this basic motif is essential for PI(4)P binding.The depletion of Golgi PI(4)P with rapamycin had similar effects on a2 localization, indicating that a2-PI(4)P interaction is important for the Golgi localization of a2-containing V-ATPases.A study identified another PI(4)P interaction sequence K 221 WY within the vicinity of the putative binding domain, and the mutation of the K221 residue compromised PI(4)P binding [55].PIPs are expressed as lipid rafts within the membrane [56,57] so that multiple basic residues are exposed to the membrane, which would strengthen protein-lipid interactions.We hypothesize that the K 221 WY sequence could help to strengthen the PIP/protein interaction or potentially define the PIP's specificity of the a2 isoform.
PIP binding is often associated with protein conformational changes [58][59][60][61].Structural analyses of yeast V-ATPases suggest that there are conformational changes between Vph1pNT in holoenzyme V 1 -V o and in free V o , as well as movement of the N-terminal domain in different states of the active enzyme [6,62].Our putative lipid binding motif is within the distal lobe of the aNT, which rotates between the different V-TPase active states.Similar to the stabilization of K + channel Kir2.2 upon binding to PI(4,5)P 2 [59], it is possible that binding to the Golgi PI(4)P traps the a2 isoform at the Golgi membrane as well as in a conformational state that promotes V-ATPase assembly and/or activity.Studies in yeast indicate that mutations resulting in the loss of PI(4)P binding compromise Stv1p-containing V-ATPases' function, resulting in growth defects at alkaline pH [40].Structural studies of conformational changes induced by PIP binding can provide mechanistic insights for the functional role of this interaction.Nevertheless, aNT is the target of multiple V-ATPase regulators [15,17,20,41,63], and PIP interaction is only one mode of V-ATPase regulation at specific membranes.
The a2 K237_V238del mutation has been identified in patients with cutis laxa [48].Here, we show that this mutation disrupted PI(4)P interaction and compromised Golgi localization.The characterization of conserved residues implicated in diseases has been successfully used to determine functional domains and has informed the discovery of novel therapeutic targets [64][65][66].Understanding a-PIP interactions and their impact on V-ATPase localization and regulation could similarly inform the development of a therapeutic control of V-ATPase subpopulations, enabling the inhibition, specifically, of V-ATPases involved in osteoporosis [7,21], and metastatic cancer [67,68].
The N-terminal domain of human a2 (amino acid 1-400) with a 6X His tag at the carboxyl end was expressed in E.coli Rosetta (DE3) via pET32a plasmid and purified as described in a previous study [47].

Immunofluorescence
In the rapamycin treatment experiment, cells were treated with 100 nM of rapamycin for 15 min at room temperature before fixing.Images were acquired with a confocal microscope (Leica Confocal SP8, Germany) using a 63x oil objective.Colocalizations were measured with Mander's coefficient M1 [69].Antibodies used were as follows: mouse anti-DDDDK tag (Abcam ab18230), goat anti-mouse IgG Alexa Fluor TM 647 (Invitrogen A21235), and DAPI stain (Invitrogen D1306).Institutional Review Board Statement: The HEK293 cell line was obtained from ATCC.Human a2NT plasmid constructs were cloned in the lab as described above.
Informed Consent Statement: Not applicable.

14 Figure 1 .
Figure 1.a2NT model generated by Phyre2.0 with the critical basic K 237 VKK 240 motif in the distal lobe highlighted in green.

Figure 1 .
Figure 1.a2NT model generated by Phyre2.0 with the critical basic K 237 VKK 240 motif in the distal lobe highlighted in green.

Figure 1 .
Figure 1.a2NT model generated by Phyre2.0 with the critical basic K 237 VKK 240 motif in the distal lobe highlighted in green.

Figure 3 .
Figure 3. Mutations within the putative K 237 VKK 240 binding motif reduced interaction of a2NT with PI(4)P-enriched liposomes in vitro.(A) Liposome pull-down assay with PolyPIPosomes (Echelon) enriched with the indicated PIPs and HIS-tagged a2WT.In total, 20 µg of protein was incubated for 1 h at room temperature with 20 µL of 1 mM PolyPIPosomes containing 5% of the indicated PIPs in binding buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.05% Nonidet P-40).Additionally, 5 µg of purified HIS-tagged a2WT was used as the input for loading control.(Right) Quantification was performed by measuring the intensity ratio of protein pulled down with the liposomes relative to input (n = 3).(B) Liposome pull-down assay of WT and mutant proteins with PolyPIPosomes (Echelon) enriched with indicated PIPs (PI(4)P, PI(4,5)P2, PI(3,4)P2, and PIP3.(Right) Quantification by intensity ratio of WT and mutants pulled down with PI(4)P-enriched liposomes with respect to input.(C) Quantification by intensity ratio of WT and mutants pulled down with liposomes enriched with PI(4,5)P2, PI(3,4)P2, and PIP3 with respect to input.n = 3 for all figures.Error bars indicate ± S.D. Statistical significance was analyzed by one-way ANOVA with Dunnett's multiple comparisons test comparing mutants to WT. * indicates p < 0.05.** indicates p < 0.01, *** indicates p < 0.001.

Figure 3 .
Figure 3. Mutations within the putative K 237 VKK 240 binding motif reduced interaction of a2NT with PI(4)P-enriched liposomes in vitro.(A) Liposome pull-down assay with PolyPIPosomes (Echelon) enriched with the indicated PIPs and HIS-tagged a2WT.In total, 20 µg of protein was incubated for 1 h at room temperature with 20 µL of 1 mM PolyPIPosomes containing 5% of the indicated PIPs in binding buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.05% Nonidet P-40).Additionally, 5 µg of purified HIS-tagged a2WT was used as the input for loading control.(Right) Quantification was performed by measuring the intensity ratio of protein pulled down with the liposomes relative to input (n = 3).(B) Liposome pull-down assay of WT and mutant proteins with PolyPIPosomes (Echelon) enriched with indicated PIPs (PI(4)P, PI(4,5)P 2 , PI(3,4)P 2 , and PIP3.(Right) Quantification Int. J. Mol.Sci.2024, 25, x FOR PEER REVIEW 6

Figure 5 .
Figure 5. Mutations affect a2NT recruitment to Golgi.White arrows indicate the presence of a2NT at the Golgi.(A) Plasmids containing FLAG-tagged a2NT wildtype (WT), mutants K237A/V238A and K237_V238del were co-transfected with Tgn38-CFP in HEK293 cells.Cells were fixed, permeabilized, and stained for FLAG-tagged proteins (red) and DAPI (blue).(B) Quantification.A minimum of 30 cells from each cell line were measured for the intensity of the red signal at the vicinity of the Golgi membrane (green).Data represent mean value ± SEM from three independent experiments.Statistical significance was analyzed by one-way ANOVA with Dunnett's multiple comparisons test comparing mutants to WT. **** indicates p < 0.0001, scare bar=10 µm.

Figure 5 .
Figure 5. Mutations affect a2NT recruitment to Golgi.White arrows indicate the presence of a2NT at the Golgi.(A) Plasmids containing FLAG-tagged a2NT wildtype (WT), mutants K237A/V238A and K237_V238del were co-transfected with Tgn38-CFP in HEK293 cells.Cells were fixed, permeabilized, and stained for FLAG-tagged proteins (red) and DAPI (blue).(B) Quantification.A minimum of 30 cells from each cell line were measured for the intensity of the red signal at the vicinity of the Golgi membrane (green).Data represent mean value ± SEM from three independent experiments.Statistical significance was analyzed by one-way ANOVA with Dunnett's multiple comparisons test comparing mutants to WT. **** indicates p < 0.0001, scare bar = 10 µm.

Figure 6 .
Figure 6.Rapamycin recruits Sac1-FKBP to the Golgi in the presence Tgn38-FRB.(A) Schematic illustration of rapamycin-induced dimerization.Sac1 phosphatase coupled to FK506 binding protein FKBP (Sac1-FKBP) and Golgi membrane anchor Tgn38 coupled to FKBP-rapamycin binding domain FRB (Tgn38-FRB-CFP) were dimerized by the addition of rapamycin.(B) Sac1-FKBP (magenta) recruitment to Golgi (white arrow), labeled by Tgn38 (green), upon treatment with 100 nM of rapamycin for 15 min at room temperature before fixing.Quantification: A minimum of 30 cells from each cell line.Data represent mean value ± SEM from three independent experiments.A paired t-test was run to analyze the significance in mean difference.* indicates p < 0.05, scare bar=10 µm.

Figure 6 .Figure 7 .
Figure 6.Rapamycin recruits Sac1-FKBP to the Golgi in the presence Tgn38-FRB.(A) Schematic illustration of rapamycin-induced dimerization.Sac1 phosphatase coupled to FK506 binding protein FKBP (Sac1-FKBP) and Golgi membrane anchor Tgn38 coupled to FKBP-rapamycin binding domain FRB (Tgn38-FRB-CFP) were dimerized by the addition of rapamycin.(B) Sac1-FKBP (magenta) recruitment to Golgi (white arrow), labeled by Tgn38 (green), upon treatment with 100 nM of rapamycin for 15 min at room temperature before fixing.Quantification: A minimum of 30 cells from each cell line.Data represent mean value SEM from three independent experiments.A paired t-test was run to analyze the significance in mean difference.* indicates p < 0.05, scare bar = 10 µm.HEK293 cells were co-transfected with plasmids containing a2NT.FLAG, Sac1-PJ-FKBP, and Tgn38-FRB-CFP.a2NT co-localized with Tgn38-CFP, indicating recruitment to the Golgi (Figure7A, top row).Additionally, 100 nM rapamycin decreased the a2NT intensity at the

Figure 7 .
Figure 7. Depletion of Golgi PI(4)P impairs a2NT recruitment to Golgi.White arrows indicate the localization of a2NT at Golgi (A) Plasmids containing FLAG-tagged a2NT wildtype (WT) were co-transfected with Tgn38-FRB-CFP (green) and Sac1-FKBP in HEK293 cells.Cells were treated with 100 nM of rapamycin for 15 min at 30 h post-transfection.Cells were then fixed, permeabilized, and stained for FLAG-tagged proteins (red) and DAPI (blue).(B) Quantification was performed with a minimum of 30 cells from each cell line measured for the intensity of the red signal in the vicinity of Golgi (green).Data represent mean value SEM from three independent experiments.A paired t-test was run to analyze the significance in mean difference.*** indicates p < 0.001.scare bar = 10 µm.

4. 5 .
Statistical Analysis GraphPad Prism 9.4.1 software was used for statistical analysis and statistical graph production.One-way ANOVA followed by Dunnett's multiple comparison test or Student's t-test were used as indicated in figure legends.In figures, asterisks are used as follows: * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001.The experimental results are expressed as the mean ± SEM.Author Contributions: A.C. performed all experiments except those represented in Figure 2, which were performed by M.G. and Y.Y. prepared constructs used in the study and provided technical expertise.C.M.D. helped with the initial design of CD experiments and provided expertise in the data interpretation of the experiments.M.F.M. supervised the project, participated in data interpretation, and revised the draft of the manuscript.All authors have read and agreed to the published version of the manuscript.Funding: This work was supported by Operating Grants PJT-14850 from the Canadian Institutes of Health Research and RGPIN-2022-05169 from the Natural Science and Engineering Research Council of Canada to M.F.M, and by Operating Grant #376666 from the Canadian Institutes of Health Research to C.M.D.