A genetic variant in SLC30A2 causes breast dysfunction during lactation by inducing ER stress, oxidative stress and epithelial barrier defects

Abstract

SLC30A2 encodes a zinc (Zn) transporter (ZnT2) that imports Zn into vesicles in highly-specialized secretory cells. Numerous mutations and non-synonymous variants in ZnT2 have been reported in humans and in breastfeeding women; ZnT2 variants are associated with abnormally low milk Zn levels and can lead to severe infantile Zn deficiency. However, ZnT2-null mice have profound defects in mammary epithelial cell (MEC) polarity and vesicle secretion, indicating that normal ZnT2 function is critical for MEC function. Here we report that women who harbor a common ZnT2 variant (T²⁸⁸S) present with elevated levels of several oxidative and endoplasmic reticulum (ER) stress markers in their breast milk. Functional studies in vitro suggest that substitution of threonine for serine at amino acid 288 leads to hyperphosphorylation retaining ZnT2 in the ER and lysosomes, increasing ER and lysosomal Zn accumulation, ER stress, the generation of reactive oxygen species, and STAT3 activation. These changes were associated with decreased abundance of zona occludens-1 and increased tight junction permeability. This study confirms that ZnT2 is important for normal breast function in women during lactation, and suggests that women who harbor defective variants in ZnT2 may be at-risk for poor lactation performance.

Figure 1

Markers of breast dysfunction and oxidative stress in breast milk from women harboring wild-type ZnT2 (T²⁸⁸) or the ZnT2 variant (S²⁸⁸). (a) Measurement of lactoferrin concentration in breast milk from women harboring two wild-type ZnT2 alleles (T²⁸⁸) and women harboring the ZnT2 variant (S²⁸⁸). Milk lactoferrin concentration was measured by ELISA. Data represent mean milk lactoferrin concentration (g/L) ± SD from n = 5 samples/genotype. (b) Evaluation of MMP-2 activity in breast milk from women harboring T²⁸⁸ or S²⁸⁸. MMP-2 activity was determined by gelatin zymography (arrow); data represent mean gelatin lysis area (clear bands) relative to T²⁸⁸ ± SD from n = 4–5 samples/genotype. Cropped gel is displayed and full-length gel can be found in Supplementary Fig. S2a. (c) Representative immunoblots of oxidative stress markers (4-HNE, mucin-4 and endoplasmin) in a fixed volume (5 µL) of breast milk from women harboring T²⁸⁸ or S²⁸⁸. A replicate gel was stained with Coomassie Blue as a loading control (bottom panel). Cropped blots are displayed and full-length blots can be found in Supplementary Fig. S2b–d. Quantification of relative protein abundance of (d) 4-HNE, (e) mucin-4 and (f) endoplasmin. Data represent mean signal intensity normalized to T²⁸⁸ ± SD from n = 4–5 samples/genotype; p < 0.05*, p < 0.01**.

Figure 2

Ectopic expression of the S²⁸⁸ variant of ZnT2 is retained in the ER, accumulates Zn in ER, and induces ER stress. (a) Representative confocal images of ER-ZAPCY1 (green) and ZnT2-HA (red) in MECs transfected to express wild-type ZnT2 (T²⁸⁸) and the ZnT2 variant (S²⁸⁸). Merged images (yellow) illustrate co-localization of ER-ZAPCY1 and ZnT2. Nuclei were counterstained with DAPI (blue). Note robust co-localization of ER-ZAPCY1 and ZnT2-HA in MECs expressing S²⁸⁸ (Pearson’s coefficient = 0.87), indicating ER localization of S²⁸⁸ compared with MECs expressing T²⁸⁸ (Pearson’s coefficient = 0.34; scale bar, 25 µm). (b) Representative FRET analysis demonstrating the changes in FRET ratio (R) of ER-ZAPCY1 in MECs expressing T²⁸⁸ or S²⁸⁸ treated with TPEN (100 µM) and thapsigargin (2 µM; R_min) followed by zinc pyrithione (100 µM; R_max), n = 10–14 cells/genotype, from four independent experiments. (c) Representative pseudocolored FRET signal images of ER-ZAPCY1 in MECs expressing T²⁸⁸ or S²⁸⁸ at rest (Basal), after TPEN (100 µM) + thapsigargin (2 µM; TPEN) treatment, in each case followed by zinc pyrithione (100 µM; Zn) treatment (scale bar, 10 µm). (d) Quantification of basal FRET ratio in MECs expressing T²⁸⁸ or S²⁸⁸. Data represent mean FRET ratio at basal levels ± SEM, n = 10–14 cells/genotype, from four independent experiments; p < 0.05*. (e) Representative immunoblot of endoplasmin (Endo) in total lysates from MECs expressing T²⁸⁸ or S²⁸⁸ treated with Zn. β-actin served as a loading control. Dotted lines indicate spliced sections obtained from a single blot; representative samples (n = 2/group) were selected for publication. Spliced blots are displayed and full-length blots can be found in Supplementary Fig. S3a,b. (f) Quantification of endoplasmin expression. Data represent mean endoplasmin expression normalized to β-actin ± SD, n = 6 samples/genotype, from three independent experiments; p < 0.05*.

Figure 3

Cells expressing the S²⁸⁸ variant of ZnT2 have increased oxidative stress, lysosomal activity and STAT3 activation. (a) Assessment of reactive oxygen species (ROS) level in untransfected MECs (Control) or MECs transfected to express wild-type ZnT2 (T²⁸⁸) or the ZnT2 variant (S²⁸⁸). Cells treated with H₂O₂ (100 µM) were used as a positive control. Data represent mean DCF-HA fluorescence/µg of protein ± SD, from n = 6 samples/group; the experiment was repeated three times. Means with different letters are significantly different, p < 0.01. (b) Representative confocal images of FluoZin-3 (green) and Lysotracker Red (red) in MECs expressing T²⁸⁸ or S²⁸⁸. Merged images (yellow) illustrate co-localized FluoZin-3 and Lysotracker Red. Note greater Lysotracker Red fluorescence in MECs expressing T²⁸⁸ compared with MECs expressing S²⁸⁸ (scale bar, 20 µm). (c) Representative immunoblots of p-STAT3 and total STAT3 in cell lysates from MECs expressing T²⁸⁸, S²⁸⁸, or mock-transfected (Mock) cells. β-actin served as a loading control. Cropped blots are displayed and full-length blots can be found in Supplementary Fig. S4a–c. (d) Quantification of STAT3 activation. Data represent mean p-STAT3/total STAT3 ± SD from n = 6 samples/genotype, from three independent experiments. Means with different letters are significantly different, p < 0.05.

Figure 4

Barrier function is disrupted in MECs expressing the S²⁸⁸ variant of ZnT2. (a) Representative confocal images of E-cadherin (green) and ZnT2-HA (red) in MECs transfected to express wild-type ZnT2 (T²⁸⁸) or the ZnT2 variant (S²⁸⁸). Nuclei were counterstained with DAPI (blue; scale bar, 10 µm). (b) Representative immunoblot of zonula occludens-1 (ZO-1) in total cell lysates from MECs expressing T²⁸⁸ or S²⁸⁸. Ponceau staining served as a loading control. Cropped blots are displayed and full-length blots can be found in Supplementary Fig. S5a,b. (c) Quantification of ZO-1 expression. Data represent mean ZO-1 signal intensity ± SD from n = 3 samples/genotype; p < 0.05*. (d) Assessment of barrier function in cells expressing T²⁸⁸ and S²⁸⁸. Data represent mean FITC-dextran fluorescence (arbitrary units) ± SD, from n = 3 samples/genotype; the experiment was repeated two times. Mock represents mock-transfected MECs. Means with different letters are significantly different, p < 0.05.

Figure 5

Substitution of serine for threonine at position 288 (S²⁸⁸) in ZnT2 leads to ZnT2 hyperphosphorylation. (a) Graphical representation of potential phosphorylation sites (serine, threonine and tyrosine) in wild-type ZnT2 (T²⁸⁸; left) and ZnT2 variant with a threonine to serine substitution (S²⁸⁸; right) as inferred from NetPhos 2.0. Green line represents potential phosphorylated serine residues; blue line represents potential phosphorylated threonine residues; pink line represents potential phosphorylated tyrosine residues; red horizontal line indicates threshold for modification potential; score indicates predicted phosphorylation potential score. (b) Representative immunoblot of phosphorylated serine in immunoprecipitates (IP) from MECs expressing T²⁸⁸ or S²⁸⁸. HA was used as normalization and input control. Cropped blots are displayed and full-length blots can be found in Supplementary Fig. S6a,b. (c) Quantification of serine phosphorylation. Data represent mean p-serine/HA ratio ± SD, n = 6 samples/genotype, from two independent experiments; p < 0.05*.

Figure 6

Model comparing MEC functions of wild-type ZnT2 (T²⁸⁸) and the ZnT2 variant (S²⁸⁸) during lactation. (a) Optimal lactation is achieved through tight regulation of milk secretion, MEC polarity and barrier integrity. During lactation, wild-type ZnT2 (T²⁸⁸) imports zinc into secretory vesicles in MECs, which is critical for secretory differentiation and secretory activation. (b) However, a common hyperphosphorylated ZnT2 variant (S²⁸⁸) is retained in the ER and lysosomes, leading to increased ER and lysosomal Zn accumulation, ER and oxidative stress, defects in tight junction formation and paracellular barrier formation, resulting in sodium leakage into milk.