Mitochondrial transmembrane proteins face a solubility-stability tradeoff.

(A) Non-mitochondrial proteins are co-translationally inserted via the SRP–Sec61 pathway with accessory Oxa1-like insertases (e.g. EMC/GET/TMCO) — bypassing solubility constraints (1, 2). Nuclear-encoded IMM proteins are post-translationally targeted (3), remaining vulnerable to mistargeting (4) and aggregation in the cytoplasm or during mitochondrial import (5). Once successfully imported (6), nuDNA-encoded IMM proteins are inserted into the IMM via TIM machinery and can be assisted by OXA1L (7). mtDNA-encoded IMM proteins, by contrast, are co-translationally inserted via OXA1L directly from the matrix (8), but face hydrophobicity-driven assembly and proteostasis constraints (5). (B) Bacterial (E. coli) proteins are co-translationally inserted into the cell membrane via the SRP– Sec pathway with an Oxa1-like insertase (YidC), bypassing aqueous exposure and proteostasis constraints. (C) Schematic illustrating transmembrane (red) and soluble (blue) helix categories analyzed from TM proteins (left) and soluble proteins (right). (D) Computational pipeline for structural alignment of evolutionarily related TMHs between E. coli and H. sapiens. Created with BioRender.

Evolutionary comparisons show aliphatic-to-threonine substitutions in mitochondrial transmembrane proteins

(A, B) Hydrophobicity scores of structurally aligned TMH pairs from evolutionarily related E. coli and H. sapiens nuDNA-encoded and mtDNA-encoded IMM proteins, respectively. (C) Mean amino acid composition of TMHs from evolutionarily related H. sapiens nuDNA-encoded and mtDNA-encoded IMM proteins, expressed as fold change relative to paired E. coli TMHs and normalized to the corresponding fold change in non-mitochondrial TM proteins. Red indicates enrichment, blue indicates depletion. (D, E) Aliphatic residue fractions of structurally aligned TMH pairs from evolutionarily related E. coli and H. sapiens nuDNA-encoded and mtDNA-encoded IMM proteins, respectively. (F, G) Threonine fractions of structurally aligned TMH pairs from evolutionarily related E. coli and H. sapiens nuDNA-encoded and mtDNA-encoded IMM proteins, respectively. Lines connect paired helices; bars represent mean ± SD. p-values from Wilcoxon signed-rank test. (H) Relative substitution matrix for aligned TMHs from evolutionarily related E. coli and H. sapiens IMM TM proteins, normalized to substitutions from non-IMM TMHs. Asterisks denote statistical significance from Fisher’s exact test. (I, J) Positional distribution of ALIVF fraction (I) and threonine fraction (J) along UniProtKB-annotated TMHs from H. sapiens mitochondrial and non-mitochondrial TM proteins and their respective evolutionarily related E. coli proteins, aligned agnostically to membrane orientation. Values on figure denote odds ratio/significance for core vs. peripheral enrichment from Fisher’s exact test (✱p<0.05, ✱✱p<0.01, ✱✱✱p<0.001).

Threonine enrichment distinguishes human mitochondrial proteins

(A–C) Hydrophobicity scores (A), ALIVF fraction (B), and threonine fraction (C) of human TMHs identified from TM protein AlphaFold2 structures, grouped by membrane compartment. Each point represents one TM protein; bars represent mean ± SD. p-values from one-way ANOVA with Tukey’s post-hoc test (all pairwise comparisons shown in Figure 3—figure supplement 1). (D) Amino acid residue frequency (%) in human TMHs identified from TM protein AlphaFold structures, grouped by membrane compartment. Cell values represent mean residue frequency; color scale reflects fold-change relative to the mean of non-mitochondrial compartments (red, enriched; blue, depleted; all compartments shown separately in Figure 3—figure supplement 1). (E, F) Positional distribution of ALIVF fraction (E) and threonine fraction (F) along TMHs from human TM proteins, aligned agnostically to membrane orientation, grouped by membrane compartment. Values on figure denote odds ratio/significance for core vs. peripheral enrichment from Fisher’s exact test (✱p<0.05, ✱✱p<0.01, ✱✱✱p<0.001). Created with BioRender.

ATS is found in mitochondrial gene transfers

(A) Top, mtDNA reduction across eukaryotes reflects differential retention of mitochondrial genes after endosymbiotic gene transfer. Bottom, mtDNA-encoded IMM proteins are synthesized and inserted locally in the matrix (1). After endosymbiotic gene transfer to the nuclear genome (2), the same IMM protein becomes nuDNA-encoded (3) and requires mitochondrial targeting (4), exposing hydrophobic TMHs to aggregation and mistargeting. (B) Computational pipeline for identifying and analyzing mtDNA-to-nuDNA gene transfer events across eukaryotic phyla. (C, D) Hydrophobicity scores (C) and predicted aggregation propensity (D) of structurally aligned TMH pairs from mtDNA- and nuDNA-encoded versions of the same IMM protein across eukaryotes. (E) Amino acid residue frequency (%) in TMHs of mtDNA- and nuDNA-encoded IMM proteins across eukaryotes. Cell values represent mean residue frequency; color scale reflects fold-change relative to mtDNA-encoded proteins (red, enriched; blue, depleted). (F) ALIVF fraction and (G) threonine fraction of structurally aligned TMH pairs from mtDNA- and nuDNA-encoded versions of the same IMM protein across eukaryotes. Lines connect paired helices; bars represent mean ± SD. p-values from Wilcoxon signed-rank test.

ATS is convergently acquired in independent ATP6 gene transfer events

(A) Phylogenetic tree of ATP6 sequences from apicomplexans, chlorophytes, and metazoans. Blue indicates nuclear-encoded ATP6; red indicates mitochondrially encoded ATP6. Numbers at nodes indicate bootstrap support values. (B) Structural alignment of mlATP6 (blue) and ppATP6 (red) reveals conservation of proton channel architecture following nuclear gene transfer. Yellow indicates threonine residues in mlATP6. RMSD = 1.790 Å. (C) Whole-protein hydrophobicity profiles of mlATP6 (blue) and ppATP6 (red). Colored bars above and below indicate predicted transmembrane helices for ppATP6 and mlATP6, respectively. Yellow circles indicate threonine residues in mlATP6 TMHs. p<0.0001, Wilcoxon rank-sum test comparing TMH hydrophobicity scores between mlATP6 and ppATP6.

ATS is required for mlATP6 targeting.

(A) Schematic of the three EGFP-tagged ATP6 constructs: WT-hATP6 (human MT-ATP6 with mlATP6 MTS appended), WT-mlATP6, and STA-mlATP6 (Thr-to-aliphatic substitutions reversing ATS). Yellow lines, threonine residues in WT-mlATP6; red lines, Thr-to-aliphatic substitutions in STA-mlATP6. (B) Whole-protein hydrophobicity profiles of WT-hATP6 (red), WT-mlATP6 (solid blue), and STA-mlATP6 (dashed blue). Yellow circles mark threonine positions in WT-mlATP6. Bars above and below indicate TMH regions; p-values denote hydrophobicity comparisons between constructs. (C–E) Confocal imaging of HEK293T cells expressing WT-hATP6 (C), WT-mlATP6 (D), or STA-mlATP6 (E) shown as EGFP (green), Mito-BFP (magenta), and merged channel. Scale bars, 10 µm. (F) Scoring of EGFP localization across cells as Non-Mito, Partial, or Mito for each construct. Data are mean ± SD from three independent experiments. Representative images of each scoring category (Non-Mito, Partial, Mito; left to right) are shown above. (G) Manders’ colocalization coefficient (M1; fraction of EGFP overlapping with Mito-BFP) quantifying mitochondrial localization for each construct. Data are mean ± SD from three independent experiments; p-values from repeated-measures one-way ANOVA with Greenhouse-Geisser correction and Tukey’s multiple comparisons test. Created with BioRender.

ATS supports hATP6 targeting.

(A) Computational pipeline for designing progressively less hydrophobic hATP6 constructs. (B) Schematics of the hATP6-WT, -T1, -T2, and -T3 constructs (MTS, blue; ATP6, white; EGFP, green); vertical lines mark aliphatic-to-threonine substitution positions, colored by construct (T1, yellow; T2, cyan; T3, magenta). Below, Kyte-Doolittle hydrophobicity profiles of each construct with threonine positions indicated. p-values from pairwise Wilcoxon rank-sum tests comparing sequential constructs. (C) Representative confocal images of HEK293T cells stably expressing the indicated hATP6 constructs (EGFP, green; MitoTracker Red, red; nuclei, blue), shown as individual and merged channels. Scale bars, 10 µm. (D) Scoring of EGFP localization across cells as Non-Mito, Partial, or Mito for each construct. (E) Manders’ colocalization coefficient (M1; fraction of EGFP overlapping with MitoTracker Red) for each construct. (F) Fraction of DAPI (nuclear) signal overlapping with EGFP for each construct, reflecting non-mitochondrial EGFP distribution. Data are mean ± SEM from three independent experiments; p-values from one-way ANOVA with Tukey’s post hoc test. Created with BioRender.

Threonine remains membrane-compatible despite polar character.

(A) Schematic of simulation system, with UCP1 in dark blue ribbons, and POPC membrane in red sticks. The figure was made using PyMOL 3.1.6.1. (B) Canonical χ1 rotamers of threonine. Newman projections along the Cα–Cβ bond. Red, Oγ; blue, backbone N; dashed arc, χ1 dihedral. At −60° and +60°, the hydroxyl oxygen is positioned to form a hydrogen bond with the backbone carbonyl; the 180° rotamer precludes this interaction. (C, D) χ1 rotamer occupancy of threonine residues by structural context. Mean probability (±SD) of occupying the ±60° or 180° χ1 rotamer for threonines in UCP1 and ANT1, grouped by context: membrane-interior helix (red, n=6), aqueous helix (blue, n=8), and aqueous loop (cyan, n=6). Dots represent individual residues. (E) Rotamer transition free energy as a function of OH–water hydrogen bonding. ΔG(±60°→180°) for individual threonine residues in UCP1 and ANT1 plotted against their mean number of hydroxyl–water hydrogen bonds. Red and blue shading denote the membrane-interior and aqueous environments, respectively. (F) Model for threonine’s role in ATS: its hydroxyl hydrogen bonds to water while the protein is unfolded during cytoplasmic transport, and to the i−4 backbone carbonyl once folded into transmembrane helices — keeping threonine compatible with both mitochondrial targeting and transmembrane helix folding. (G) Membrane-to-water transfer free energy versus helical propensity for all amino acids. ΔG(membrane → water) plotted against ΔG(coil → helix) for each amino acid, colored by side-chain chemistry: nonpolar (red), charged and amides (blue), OH-groups (orange), and other (black). Schematics depict the membrane-to-water and coil-to-helix equilibria. Proline is absent in the published helicity scale due to its extremely unfavorable free energy of helix formation and was omitted from this figure. Created with BioRender.

Structural alignment enables comparison of divergent proteins.

(A) Needleman-Wunsch sequence alignment of E. coli ZNTA and human ATP7A. (B, C) PyMOL structural alignment (RMSD = 2.373 Å) shown in two orientations. Transmembrane helices are colored consistently across all panels; lighter and darker shades denote E. coli and human sequences respectively.

ATS is specific to mitochondrial TMHs over evolution

(A–C) Hydrophobicity score (A), threonine fraction (B), and serine fraction (C) of structurally aligned TMH pairs from evolutionarily related E. coli and H. sapiens non-IMM TM proteins. (D–F) Hydrophobicity score (D), aliphatic residue fraction (E), and threonine fraction (F) of structurally aligned SolH pairs from evolutionarily related E. coli and H. sapiens non-mitochondrial, nuDNA-encoded IMM, and mtDNA-encoded IMM TM proteins. Lines connect paired helices; bars represent mean ± SD. p-values from Wilcoxon signed-rank test.

Mitochondrial soluble proteins do not show ATS relative to proteobacterial counterparts.

(A) Non-mitochondrial soluble proteins are synthesized and targeted to various compartments (1, 2), bypassing solubility constraints. Nuclear-encoded mitochondrial soluble proteins are post-translationally imported into the matrix (3, 4) and released as soluble proteins (5). Unlike inner mitochondrial membrane TM proteins, neither class of soluble proteins faces hydrophobicity-driven solubility constraints. (B) E. coli soluble proteins are translated and folded in the cytoplasm, bypassing hydrophobicity-driven solubility constraints. (C) Hydrophobicity score of structurally aligned soluble helix pairs from evolutionarily related E. coli and H. sapiens non-mitochondrial and mitochondrial soluble proteins. (D) Mean amino acid composition of helices from evolutionarily related H. sapiens mitochondrial soluble proteins, expressed as fold change relative to paired E. coli helices and normalized to the corresponding fold change in non-mitochondrial soluble proteins. Red indicates enrichment, blue indicates depletion. (E) Related to (C), threonine fraction of structurally aligned soluble helix pairs from evolutionarily related E. coli and H. sapiens non-mitochondrial and mitochondrial soluble proteins. Lines connect paired helices; bars represent mean ± SD. p-values from Wilcoxon signed-rank test. Created with BioRender.

Mitochondrial, but not chloroplast thylakoid, membrane proteins are threonine-enriched relative to bacterial counterparts.

(A–C) Hydrophobicity score (A), ALIVF fraction (B), and threonine fraction (C) of all UniProtKB-annotated TMHs from E. coli and H. sapiens IMM-targeted TM proteins. (D–F) Hydrophobicity score (D), ALIVF fraction (E), and threonine fraction (F) of all UniProtKB-annotated TMHs from R. rickettsii and H. sapiens IMM-targeted TM proteins. (G–I) Hydrophobicity score (G), ALIVF fraction (H), and threonine fraction (I) of all UniProtKB-annotated TMHs from Synechocystis sp. and A. thaliana chloroplast thylakoid membrane proteins. Bars represent mean ± SD. p-values from a Mann–Whitney U test.

Substitution matrices for soluble helix controls show no ATS enrichment.

(A) Relative substitution matrix for aligned SolHs from evolutionarily related E. coli and H. sapiens IMM TM proteins, normalized to substitutions from non-IMM SolHs. (B) Relative substitution matrix for aligned helices from evolutionarily related E. coli and H. sapiens soluble proteins, normalized to substitutions from non-mitochondrial soluble proteins. Asterisks denote statistical significance from Fisher’s exact test (*p<0.05, **p<0.01, ***p<0.001).

Structurally annotated mitochondrial TMHs have lower hydrophobicity and higher threonine content.

(A–C) Hydrophobicity scores (A), ALIVF fraction (B), and threonine fraction (C) of human TMHs identified from TM protein AlphaFold2 structures, grouped by membrane compartment, with all pairwise comparisons shown. Each point represents one TM protein; bars represent mean ± SD. p-values from one-way ANOVA with Tukey’s post-hoc test. (D) Amino acid residue frequency (%) in human TMHs identified from TM protein AlphaFold structures, grouped by membrane compartment. Cell values represent mean residue frequency; color scale reflects fold-change relative to the mean of non-mitochondrial compartments (red, enriched; blue, depleted).

UniProtKB-annotated mitochondrial transmembrane helices have lower hydrophobicity and higher threonine.

(A–C) Hydrophobicity scores (A), ALIVF fraction (B), and threonine fraction (C) of human TMHs identified from TM protein UniProtKB annotations, grouped by membrane compartment. Each point represents one TM protein; bars represent mean ± SD. p-values from one-way ANOVA with Tukey’s post-hoc test. (D) Amino acid residue frequency (%) in human TMHs identified from TM protein UniProtKB annotations, grouped by membrane compartment. Cell values represent mean residue frequency; color scale reflects fold-change relative to the mean of non-mitochondrial compartments (red, enriched; blue, depleted).

Threonine is enriched in the core of mitochondrial TMHs.

(A) Positional distribution of HKRDE fraction along TMHs from human TM proteins, aligned agnostically to membrane orientation, grouped by membrane compartment. (B, C) Positional distribution of ALIVF fraction (B) and threonine fraction (C) along a separate set of TMHs oriented by structurally informed annotation, grouped by membrane compartment. Values on figure denote odds ratio/significance for core vs. peripheral enrichment from Fisher’s exact test (✱p<0.05, ✱✱p<0.01, ✱✱✱p<0.001).

Soluble helices of nuclear-encoded TM proteins have similar hydrophobicity and composition.

(A–C) Hydrophobicity scores (A), ALIVF fraction (B), and threonine fraction (C) of human soluble helices identified from TM protein AlphaFold2 structures, grouped by membrane compartment, with all pairwise comparisons shown. Each point represents one TM protein; bars represent mean ± SD. p-values from one-way ANOVA with Tukey’s post-hoc test. (D) Amino acid residue frequency (%) in human soluble helices identified from TM protein AlphaFold structures, grouped by membrane compartment. Cell values represent mean residue frequency; color scale reflects fold-change relative to the mean of non-mitochondrial compartments (red, enriched; blue, depleted).

Soluble mitochondrial proteins do not show decreased hydrophobicity or altered composition.

(A–C) Hydrophobicity scores (A), ALIVF fraction (B), and threonine fraction (C) of human soluble helices identified from soluble protein AlphaFold structures, grouped by compartment. Each point represents one soluble protein; bars represent mean ± SD. p-values from one-way ANOVA with Tukey’s post-hoc test. (D) Amino acid residue frequency (%) in human soluble helices identified from soluble protein AlphaFold2 structures, grouped by compartment. Cell values represent mean residue frequency; color scale reflects fold-change relative to the mean of non-mitochondrial compartments (red, enriched; blue, depleted).

Threonine is enriched in transmembrane over soluble helices in IMM proteins.

Threonine fraction of concatenated transmembrane versus soluble helices within each TM protein. (A–D) nuDNA-encoded non-mitochondrial TM proteins (E, F) nuDNA-encoded mitochondrial TM proteins (G) mtDNA-encoded IMM TM proteins (H) Threonine fraction of individual, unconcatenated helices from mtDNA-encoded IMM TM proteins Lines connect paired transmembrane and soluble helices from the same protein; bars represent mean ± SD. p-values from Wilcoxon signed-rank test (A–G) and Mann–Whitney test (H).

Serine preference over threonine is reversed in IMM transmembrane helices.

Threonine versus serine residue fraction in helices from TM proteins. Top panels (A–H) show transmembrane helices; bottom panels (I–P) show soluble helices. Lines connect paired transmembrane and soluble helices from the same protein; bars represent mean ± SD. p-values from Wilcoxon signed-rank tests and Mann–Whitney tests for unconcatenated helices.

Targeting and insertion constraints shape eukaryotic TMH hydrophobicity and threonine content.

(A–C) TMH hydrophobicity scores for metazoan (A), plant (B), and jakobid (C) proteins. (D–F) Threonine fractions within TMHs for the corresponding metazoan (D), plant (E), and jakobid (F) proteins. Proteins were grouped into non-mitochondrial, nuclear-encoded inner mitochondrial membrane, and mitochondrially encoded IMM classes; plant proteins additionally included nuclear-encoded and chloroplast genome-encoded chloroplast thylakoid membrane (CTM) proteins. Statistical significance was assessed using one-way ANOVA with multiple-comparison testing; adjusted p-values are shown above comparisons. Sample sizes are indicated below each group.

ATS is TMH-specific during mitochondrial gene transfers.

(A) Hydrophobicity scores of structurally aligned soluble helix pairs from mtDNA- and nuDNA-encoded versions of the same IMM protein across eukaryotes. (B) Hydrophobicity scores of structurally aligned soluble helix pairs from mtDNA- and nuDNA-encoded versions of the same soluble mitochondrial protein across eukaryotes. (C) Predicted aggregation propensity of the above soluble proteins. (D) Amino acid residue frequency (%) in SolHs of mtDNA- and nuDNA-encoded soluble proteins across eukaryotes. Cell values represent mean residue frequency; color scale reflects fold-change relative to mtDNA-encoded proteins (red, enriched; blue, depleted). (E–H) ALIVF fraction (E, F) and threonine fraction (G, H) of structurally aligned soluble helix pairs from mtDNA- and nuDNA-encoded versions of the same IMM transmembrane protein (E, G) and soluble protein (F, H) across eukaryotes. Lines connect paired helices; bars represent mean ± SD. p-values from Wilcoxon signed-rank test. (I) Ratio of mean TMH threonine content in nuDNA- to mtDNA-encoded TM proteins, calculated separately for organisms retaining mitochondrial gene encoding and those having transferred the gene to the nuclear genome. Lines connect paired organisms; bars represent mean ± SD. p-value from Wilcoxon signed-rank test. (J, K) Positional distribution of ALIVF fraction (J) and threonine fraction (K) along UniProtKB-annotated TMHs from mtDNA- and nuDNA-encoded IMM proteins across eukaryotes aligned agnostically to membrane orientation. Values on figure denote odds ratio/significance for core vs. peripheral enrichment from Fisher’s exact test (✱p<0.05, ✱✱p<0.01, ✱✱✱p<0.001).

Nuclear-encoded ATP6s possess cleavable N-terminal MTSs.

(A–C) TargetP 2.0 prediction plots for nuclear-encoded ATP6 sequences. The y-axis shows predicted probability of mitochondrial targeting sequence (MTS) identity at each residue position. The predicted cleavage site (CS) is indicated. (A) mlATP6, (B) crATP6, (C) tgATP6. Predicted MTS likelihoods: mlATP6 = 0.99, crATP6 = 0.56, tgATP6 = 0.49.

MT-ATP9 underwent independent nuclear gene transfers.

Phylogenetic tree of MT-ATP9 sequences. Blue indicates nuclear-encoded ATP9; red indicates mitochondrially encoded ATP9. Numbers at nodes indicate bootstrap support values.

Nuclear-encoded ATP9s possess cleavable N-terminal MTSs.

(A–D) TargetP 2.0 prediction plots for nuclear-encoded ATP9 sequences. The y-axis shows predicted probability of mitochondrial targeting sequence (MTS) identity at each residue position. The predicted cleavage site (CS) is indicated. (A) tgATP9 (B) aqATP9 (C) crATP9 (D) paATP9; Predicted MTS likelihoods: tgATP9 = 0.94, aqATP9 = 0.96, crATP9 = 0.59, paATP9 = 0.99.

Sequence and structural alignment of ATP6s shows ATS.

(A) ClustalW sequence alignment of ppATP6 and mlATP6. Red and blue highlights indicate UniProtKB-annotated transmembrane helices in ppATP6 and mlATP6, respectively. Yellow indicates threonine residues in mlATP6 TMHs. (B) Structural alignment of crATP6 (blue) and baATP6 (red) reveals conservation of proton channel architecture following nuclear gene transfer. Yellow indicates threonine residues in crATP6. RMSD = 0.868 Å. The putative MTS and TMH1 were omitted from the structural visualization as they do not form part of the proton channel architecture. (C) Whole-protein hydrophobicity profiles of crATP6 (blue) and baATP6 (red). Colored bars above and below indicate predicted transmembrane helices for baATP6 and crATP6, respectively. Yellow circles indicate threonine residues in crATP6 TMHs. p<0.0001, Wilcoxon rank-sum test comparing TMH hydrophobicity scores between crATP6 and baATP6. (D) ClustalW sequence alignment of baATP6 and crATP6. Red and blue highlights indicate UniProtKB-annotated transmembrane helices in baATP6 and crATP6, respectively. Yellow indicates threonine residues in crATP6 TMHs. (E) Structural alignment of tgATP6 (blue) and jbATP6 (red) reveals conservation of proton channel architecture following nuclear gene transfer. Yellow indicates threonine residues in tgATP6. RMSD = 10.109 Å. TMHs 1–3 of jbATP6 were omitted from the structural visualization as they do not form part of the proton channel architecture. (F) Whole-protein hydrophobicity profiles of tgATP6 (blue) and jbATP6 (red). Colored bars above and below indicate predicted transmembrane helices for jbATP6 and tgATP6, respectively. Yellow circles indicate threonine residues in tgATP6 TMHs. p<0.0001, Wilcoxon rank-sum test comparing TMH hydrophobicity scores between tgATP6 and jbATP6. (G) ClustalW sequence alignment of jbATP6 and tgATP6. Red and blue highlights indicate UniProtKB-annotated transmembrane helices in jbATP6 and tgATP6, respectively. Yellow indicates threonine residues in tgATP6 TMHs.

mlATP6 localizes to mitochondria across cell types.

(A) HEK293T (human embryonic kidney) cells expressing mlATP6-V5, detected by anti-V5 immunofluorescence (green) and Mito-BFP (magenta). Merge shown at right. (B) Neuro2A (mouse neuroblastoma) cells expressing mlATP6-Neptune (red) and Mito-BFP (green). Merge shown at right. (C) C2C12 (mouse skeletal muscle) cells expressing mlATP6-Neptune (red) and Mito-BFP (green). Merge shown at right. (D) JHH7 (human hepatocellular carcinoma) cells expressing mlATP6-EGFP (green) and Mito-mCherry (red). Merge shown at right. (E) HEK293T cells expressing mlATP6-EGFP (green) and Mito-BFP (magenta). Merge shown at right. (F) HEK293T cells expressing STA-mlATP6-EGFP (green), stained for the ER marker calnexin (anti-calnexin, red), with Mito-BFP (blue). Merge shown at right. Scale bar, 10 µm.

Strong MTSs alone are insufficient for hATP6 targeting.

(A–E) Representative confocal images of HEK293T cells expressing the indicated MTS-hATP6 constructs (EGFP, green) and Mito-BFP (magenta). Scale bars, 10 µm. (F) Scoring of EGFP localization across cells as Non-Mito, Partial, or Mito for each construct. Data are mean ± SD from three independent experiments. Representative images of each scoring category (Non-Mito, Partial, Mito; left to right) are shown above. (G) Manders’ colocalization coefficient (M1; fraction of EGFP overlapping with Mito-BFP) quantifying mitochondrial localization for each construct. Data are mean ± SEM from three independent experiments; p-values from repeated-measures one-way ANOVA with Tukey’s multiple comparisons test.

ATS substitutions preserve hATP6 structure.

(A) AlphaFold2-predicted structure of hATP6-WT (red) with residues substituted in each ATS construct highlighted: T1 substitutions (yellow), T2 substitutions (cyan), T3 substitutions (magenta). (B) Structural alignment of independently predicted AlphaFold2 structures for hATP6-WT (red), T1 (yellow), T2 (cyan), and T3 (magenta), each colored as a whole structure. Predicted MTS and loops removed for clarity. (C) Predicted aggregation propensity scores for hATP6-WT, -T1, -T2, and -T3. Bars show mean ± SD. p-values from one-way ANOVA with Tukey’s post hoc test.

Targeting tracks with hydrophobicity, not expression.

Representative confocal images of HEK293T cells stably expressing hATP6-WT, -T1, -T2, or -T3 (left to right; EGFP, green), acquired under identical confocal settings. Lower-expressing cells were selected for WT and T1; higher-expressing cells were selected for T2 and T3. Scale bars, 10 µm.

Threonine hydroxyl solvation by structural and solvent context.

Mean number of water molecules within 3.3 Å of the threonine sidechain hydroxyl, for threonines classified as intrahelical/buried, intrahelical/aqueous, or loop/aqueous, averaged across all simulations. Error bars denote SD.