Main

The recommendation by pediatricians that mothers breast-feed (1) is based on the unique superiority of human milk and on the nutritional, neurodevelopmental, immunologic, and psychologic advantages it confers. Yet our understanding of the elaborate machinery responsible for milk synthesis and secretion and the manner of its regulation in health and disease is rudimentary. Advances in knowledge of the cellular physiology of the mammary gland may provide avenues to influence the quantity and quality of milk produced by nursing mothers and increase the rate and average duration of breast-feeding.

The most common explanation for premature cessation of breast-feeding is the mother's perception that her milk production is inadequate (2, 3). Because lactose is the major carbohydrate and osmotic constituent of human milk, the volume of milk produced is a function of the rate of lactose synthesis. Lactose synthesis takes place within the Golgi (4) and is catalyzed by lactose synthetase, a complex of galactosyltransferase and the mammary gland-specific protein α-lactalbumin. The substrates for lactose synthesis are UDP-galactose and free glucose. Direct measurement of intracellular glucose concentration demonstrated that glucose transport into the mammary epithelial cell may be rate-limiting for lactose synthesis (5). The mammary epithelial cell not only must transport glucose from the blood across the basal membrane into the cell but also must deliver this glucose to the Golgi. The requirement for free glucose within the Golgi is unique to the lactating mammary gland. A Golgi-specific glucose carrier protein accounting for Golgi glucose uptake during lactation was proposed (6) 5 y before the cloning of GLUT1 (7), the first member of the family of facilitated diffusion glucose transporter isoforms.

GLUT1 is the only isoform of the facilitated diffusion family of glucose transporters known to be expressed in mammary gland. GLUT1 is targeted to the plasma membrane in most cell types. No glucose transporter isoform has been proven to target to the Golgi, nor is there evidence of any other mediator of glucose transport into Golgi. Five years after the discovery of GLUT1, its potential role in lactation was examined for the first time. Subcellular fractionation and Western blotting of d-10 lactating rat mammary glands suggested that during lactation, GLUT1 may also be found in the Golgi (8), but no other time points were examined. Because this study relied exclusively upon subcellular fractionation, the possibility that Golgi fractions were contaminated with other organelles cannot be excluded. In other studies (9, 10) in d-10–12 lactating rat mammary gland relying only upon microscopy, high and polarized expression of plasma membrane GLUT1 during lactation was seen, and total cellular GLUT1 content declined after weaning pups for 24 h (9). However, Golgi targeting of GLUT1 was not observed. These contradictory results demonstrated to us the need for a comprehensive multifaceted approach to examine whether or not GLUT1 is a candidate Golgi glucose transporter during lactation.

Therefore, this study examined whether changes in the amount, activity, and subcellular targeting of GLUT1 in lactating mouse mammary gland are regulated in a manner consistent with a possible role for GLUT1 in supplying free glucose to Golgi as a substrate for lactose synthesis. Specifically, we hypothesized that during lactation, the amount of GLUT1 would increase and a transition from plasma membrane targeting to Golgi targeting would occur. Furthermore, we tested the hypothesis that under conditions of forced weaning, abruptly curtailing milk synthesis, Golgi targeting of GLUT1 would diminish. To provide multiple lines of evidence, methods included subcellular fractionation and density gradient centrifugation, epifluorescent and confocal immunofluorescent microscopy, and immunohistochemistry.

METHODS

Antisera and reagents.

GLUT1 antibody was a well-characterized highly specific rabbit polyclonal antiserum to human GLUT1 raised against synthetic peptide made up of the 16 C-terminal amino acids, a kind gift of Dr. M. Mueckler (Washington University School of Medicine, St. Louis, MO). This antibody was affinity purified using the same synthetic peptide bound to thiopropyl sepharose (Pharmacia, Piscataway, NJ) (11) before use for immunoblotting or immunocytochemistry. Mouse MAb to rat β-COP were obtained from Sigma Chemical Co. (St. Louis, MO). Fluorescein-labeled goat anti-rabbit antibodies and Texas-red sheep anti-mouse antibodies were from ICN (Aurora, OH). Reagents were from Sigma Chemical Co. unless otherwise specified.

Animals.

Nulliparous female CD-1 mice (Harlan Sprague-Dawley, Indianapolis, IN) were mated, and, at parturition (d 0 of lactation), the litters were adjusted to 10 pups. Animals were fed on Purina lab chow (Ralston Purina, St. Louis, MO) and had access to water ad libitum and a daily photoperiod of 12 h. Experiments were carried out in duplicate or triplicate using different animals, and representative results are shown. Studies were reviewed and approved by the Animal Studies Committees of the Washington University School of Medicine and the Baylor College of Medicine.

Subcellular fractionation and density gradient centrifugation.

One mammary gland preparation per animal was prepared as follows. Mammary glands were removed and rinsed twice with ice-cold PBS and once with sucrose solution (0.25 M sucrose, 10 mM triethanolamine, 10 mM acetic acid, pH 7.8), resuspended in a small volume of homogenization buffer (PBS, 1 mM EDTA), and homogenized with five strokes in a tight-fitting Dounce homogenizer. After centrifugation at 3000 ×g for 10 min at 4°C, the supernatant was centrifuged at 17 000 ×g for 10 min at 4°C. This supernatant was centrifuged at 100 000 ×g for 30 min at 4°C. In certain experiments, the 17 000-g pellet was resuspended and subjected to density gradient centrifugation in a self-generating iodixanol density gradient (10–37%) at 180 000 ×g for 3 h at 4°C. Fractions were collected from the top (Labconco Auto-densi Flow, Kansas City, MO), and the lowest density fraction (1.05–1.08 g/cm3) was analyzed. Alkaline phosphatase (12) and galactosyltransferase (13) were assayed as described.

Western blotting.

Samples were prepared and subjected to standard SDS-PAGE on 10% gels as previously described (14). Proteins were immobilized on nitrocellulose by wet transfer. Primary antibody was the peptide affinity purified GLUT1 antibody described above (1 μg/mL in 5% nonfat dry milk in PBS). Horseradish peroxidase–linked donkey anti-rabbit antibody (Amersham, Piscataway, NJ) served as secondary antibody, and signal was detected using ECL-Plus (Amersham). Quantitative differences in signal strength were measured using a laser densitometer and ImageQuant software (Molecular Dynamics, Sunnyvale, CA).

Immunohistochemical staining.

Mammary tissue was fixed in 10% neutral buffered formalin supplemented with zinc chloride (Anatech, Ltd., Battle Creek, MI) and processed for paraffin sections by using standard techniques. Tissues were embedded in paraffin and sectioned at 4 μm on a rotary microtome (Leitz 1512), collected on standard glass microscope slides, and stained with hematoxylin and eosin by using a routine Harris hematoxylin solution and alcoholic eosin counterstain. Sections were affixed to capillary gap glass microscope slides (Ventana Medical Systems, Inc., Tucson, AZ), dried at 60°C for 1 h, and deparaffinized. The tissue sections were incubated for 20 min in a 1:10 citrate buffer solution (Dako Corporation, Carpinteria, CA) in a steam environment to enhance antigen availability and then incubated in the same solution for an additional 20 min at room temperature. The sections were rinsed in PBS/T to enhance flow in the capillary gap; PBS/T was used to rinse the sections between all steps of the protocol. All reagents used in the immunohistochemical procedure were made in PBS/T supplemented with 0.5% crystalline grade BSA (Sigma Chemical Co.) as a protein carrier. GLUT1 immunohistochemistry was performed using a Tek-Mate 500 automated system (Ventana Medical Systems, Inc.). Sections were incubated in a 1:75 solution of normal goat serum (Vector Laboratories, Burlingame, CA) for 20 min at room temperature. The serum was removed, anti-GLUT1 antibody was applied at a concentration of 2.5 μg/mL, and the sections were incubated in a humid chamber at room temperature overnight. Negative control sections were incubated in diluent rather than the GLUT1 antibody. After incubation, sections were rinsed and a biotin-conjugated goat anti-rabbit IgG (Vector Laboratories) was applied at 2.25 μg/mL for 45 min at room temperature. This was followed by endogenous peroxidase exhaustion using 3% H2O2 in absolute methanol, three changes of 5 min each. The sections were treated with a peroxidase-tagged avidin-biotin complex (Vector Laboratories) for 45 min at room temperature. Antigenic sites were visualized using diaminobenzidine enhanced with 1% nickel chloride as the chromogen (Sigma Chemical Co. Chemicals, St. Louis, MO). The tissue sections were counterstained with eosin, then processed through ascending grades of ethyl alcohol and xylene, and mounted on coverslips by use of a synthetic mountant.

Immunofluorescent staining.

Sections were prepared as described above. After incubation with primary antibody against GLUT1, sections were rinsed and an FITC-conjugated antibody directed against rabbit IgG (Dako Corporation) was applied at a concentration of 1:30 for 45 min in the dark. Sections were washed well, mounted in a nonfluorescent aqueous medium, and viewed with a Zeiss Axiophot epifluorescent microscope at a wavelength of 490 nm. The images are shown as equivalent exposures acquired by a Cohu (San Diego, CA) 4910 uncooled charge-coupled device camera; no enhancement or intensification was performed. For confocal microscopy slides, GLUT1 antibody was used as primary antibody as described above, and β-COP antibody was also used as primary antibody at 1:80. Then, fluorescein-labeled goat anti-rabbit antibodies and Texas-red sheep anti-mouse antibodies were both applied as described above. Also, for confocal microscopy, preimmune GLUT1 serum rather than diluent was used as a negative control. A Molecular Dynamics Multiprobe 2010 inverted confocal laser scanning microscope was used.

RESULTS

Developmental changes in GLUT1 expression.

GLUT1 expression in the mouse mammary gland (Fig. 1) was studied in virgins, in late pregnancy (d 20), on the day of delivery (d 0), at midlactation (d 8), at the peak of lactation (d 18), and then at different time points during weaning (d 21, 23, and 29). Expression gradually rose from an extremely low level in virgins, increased on the day of delivery to a peak during lactation, and then declined to very low levels as weaning progressed (Fig. 1).

Figure 1
figure 1

GLUT1 induction during pregnancy and lactation. Homogenate fractions from mouse mammary gland were subjected to SDS-PAGE, Western blotting, enhanced chemiluminescence (ECL), and laser densitometry as described in “Methods.” Samples containing 10 μg of total cellular protein were run on the same gel, and ECL exposure time was 1 min. Results are expressed relative to peak GLUT1 expression, which was observed 8 d after delivery. GLUT1 rises from very low levels in virgin gland during pregnancy, increases further during lactation, then rapidly declines upon weaning.

Full size image

To demonstrate whether there are changes in the subcellular targeting of GLUT1 as it is induced during pregnancy and lactation, subcellular fractionation was used to prepare a 17 000-g pellet enriched in Golgi and a 100 000-g pellet enriched in plasma membrane. Enrichment was demonstrated by assays of marker enzymes. Relative to total cellular levels of each marker, the activity of galactosyltransferase, a Golgi marker, was enriched 3.5- to 4.3-fold in the Golgi-enriched fraction, whereas alkaline phosphatase, a marker of plasma membrane, was enriched 4.5- to 8.6-fold in the plasma membrane-enriched fraction. Virgins demonstrated predominance of GLUT1 in the plasma membrane-enriched fraction, whereas mammary gland differentiation during pregnancy and lactation was associated with an increase in GLUT1 targeting to the Golgi-enriched fractions (Fig. 2). Quantitation by laser densitometry showed that GLUT1 was preferentially targeted to the Golgi-enriched fraction, as indicated by 3.9-fold enrichment on d 8 (Fig. 2) compared with an enrichment of 1.9-fold in the plasma membrane-enriched fraction. As expected, there was no significant targeting of GLUT1 to the 3000-g pellet, which is enriched in nuclei, or to the 100 000-g supernatant, which is enriched in cytosol. Immunohistochemistry, a method well suited for evaluation of plasma membrane staining, demonstrated labeling of plasma membrane in mammary gland of virgin, pregnant, and lactating mice (Fig. 3). The increase in GLUT1 expression during pregnancy is partially accounted for by the proliferation of mammary epithelial cells. Staining for GLUT1 was not observed during weaning, consistent with results of Western blotting. Virgin gland demonstrated a predominance of fat cells, but the mammary epithelial cells showed significant plasma membrane targeting to both the basolateral and apical membrane. In contrast, during lactation, there was intense staining of the basolateral membrane and no staining of the apical plasma membrane, indicating a polarization of membrane targeting. Immunofluorescent microscopy, which is suitable for evaluation of intracellular as well as plasma membrane targeting, confirmed nonpolarized targeting of GLUT1 in the virgin gland and an increase in expression during pregnancy and lactation (Fig. 4). In the lactating gland, intracellular staining and basolateral plasma membrane staining were seen. No specific staining was seen in mammary gland during weaning. To further define the subcellular targeting of GLUT1 during lactation, confocal immunofluorescent microscopy, which detects signal in a single plane and is more suitable for definition of intracellular targeting, was used to evaluate d-18 lactating mammary gland stained for GLUT1 and β-COP, a Golgi marker (15). GLUT1 was distributed in a predominantly perinuclear punctate pattern with some vesicular and plasma membrane staining as well (Fig. 5D). β-COP was similarly distributed but with a somewhat more diffuse pattern (Fig. 5E). The yellow signal in Figure 5F results from merging red and green signals in Figure 5, D and E , and the abundant yellow signal indicates a high degree of colocalization of GLUT1 with the Golgi marker β-COP. Figure 5, A – C , demonstrates the insignificance of nonspecific staining.

Figure 2
figure 2

Golgi targeting of GLUT1 during lactation. Subcellular fractionation of mammary gland was carried out as described in “Methods.”Lane 1, homogenate;lane 2, 3000-g (nuclear) pellet;lane 3, 3000-g supernatant;lane 4, 17 000-g (Golgi-enriched) pellet;lane 5, 100 000-g (plasma membrane-enriched) pellet;lane 6, 100 000-g supernatant (cytosol). Samples contained 60 μg of protein except for the virgin samples, which contained 25 μg. ECL exposure times were adjusted as needed to assess relative targeting to Golgi-enriched and plasma membrane-enriched fractions while avoiding saturation of signal from any one fraction and ranged from overnight for the virgin samples to 30 s for the samples from d 8. Total GLUT1 expression at these time points was directly compared in Figure 1. GLUT1 targeting shifts from predominance in the plasma membrane-enriched fraction in the virgin, to approximately equivalent plasma membrane and Golgi targeting in late pregnancy and on the day of delivery, to predominance in the Golgi-enriched fraction by d 8.

Full size image
Figure 3
figure 3

Basolateral plasma membrane targeting of GLUT1 during lactation. (A) virgin, (B) pregnant d 20, (C) lactating (d 18 after delivery), and (D) weaning (d 21 after delivery). Bar, 60 μm. Immunocytochemistry was carried out as described in “Methods.” Positive staining for GLUT1 is indicated by brown. Control slides showed no signal. The virgin gland is predominantly fat, but nonpolarized plasma membrane targeting of GLUT1 is seen in islands of mammary epithelial cells. During lactation, GLUT1 staining is intense but is observed only in basolateral plasma membrane and not in apical plasma membrane.

Full size image
Figure 4
figure 4

Intracellular targeting of GLUT1 during lactation. (A) virgin, (B) pregnant d 20, (C) lactating (d 18 after delivery), and (D) weaning (d 21 after delivery). Bar, 60 μm. Immunofluorescent staining and microscopy were carried out as described in “Methods.” Control slides showed a level of signal equivalent to that seen in weaning gland. Nonpolarized plasma membrane targeting of GLUT1 in virgin gland and polarized targeting of GLUT1 in the lactating gland are seen. In addition, strong intracellular signal is observed in the lactating but not in the virgin gland. The weaning gland shows only nonspecific staining. Strong signal is observed in red blood cells due to autofluorescence.

Full size image
Figure 5
figure 5

Colocalization of GLUT1 with the Golgi marker β-COP during lactation. (A) control with GLUT1 preimmune serum, (B) control with secondary antibody only, (C) control signals merged, (D) GLUT1 staining, (E) β-COP staining, and (F) GLUT1 and β-COP signals merged with yellow indicating colocalization. Immunofluorescent staining and microscopy were carried out as described in “Methods.”Bar, 1 μm. The high degree of colocalization of GLUT1 with the Golgi marker β-COP (F) demonstrates that GLUT1 is also targeted to Golgi. Control panels A, B, and C demonstrate that the staining shown in D, E, and F is specific.

Full size image

Reversible changes in GLUT1 content and subcellular targetingduring forced weaning.

The removal of nursing 18-d-old pups from their dams resulted in changes in the amount and subcellular targeting of mammary gland GLUT1 within 3 h (Fig. 6). A decline in total GLUT1 content and a change from predominance of GLUT1 targeting in the Golgi-enriched fraction to predominance in the plasma membrane-enriched fraction were observed. Iodixanol density gradient centrifugation of the 17 000-g pellet was used to provide further enrichment. Because Golgi membranes have the lowest buoyant density of any subcellular membrane fraction, the lowest density fraction, corresponding to a density of 1.05–1.08 g/cm3, was analyzed. A fall in Golgi enrichment of GLUT1 occurred within 3 h, and there was no Golgi enrichment of GLUT1 by 5 h (Fig. 7). However, when pups were then returned to the dam for 5 h, Golgi enrichment of GLUT1 was again observed. When pups were returned for 15 h, Golgi enrichment of GLUT1 was fully restored.

Figure 6
figure 6

Decline in GLUT1 expression and in Golgi targeting of GLUT1 during forced weaning. Eighteen-day-old pups were removed from lactating dams for the specified time. Subcellular fractionation of mammary gland was carried out as described in “Methods.”Lane 1, homogenate;lane 2, 3000-g (nuclear) pellet;lane 3, 3000-g supernatant;lane 4, 17 000-g (Golgi-enriched) pellet;lane 5, 100 000-g (plasma membrane-enriched) pellet;lane 6, 100 000-g supernatant (cytosol). Samples contained 60 μg of protein, and ECL exposure times were identical. Results shown are representative and at any time point varied 20% or less. Declines in the amount and Golgi targeting of GLUT1 are apparent 3 h after weaning and are pronounced by 5 h after weaning.

Full size image
Figure 7
figure 7

Reversibility of the effect of forced weaning on GLUT1 targeting. Pups were removed from lactating dams for the specified time. In certain experiments, indicated in crosshatch, pups were returned to their dams after a 5-h weaning period. Subcellular fractionation and iodixanol density gradient centrifugation of mammary gland were carried out as described in “Methods.” Representative results are expressed as enrichment of the purified Golgi fraction relative to total cellular levels of GLUT1. Golgi targeting of GLUT1 was lost by 5 h and was not observed after 10 or 20 h of weaning. However, when pups were returned to the dam after a 5-h absence, Golgi targeting of GLUT1 was observed within the next 5 h and was fully restored 15 h after lactation was resumed.

Full size image

DISCUSSION

The production of an adequate volume of milk by the nursing mother is a prerequisite for successful lactation. Because lactose is the major osmotic constituent of human milk, the synthesis of lactose determines the volume of milk produced. Because changes in lactose synthetase activity do not correlate with changes in milk production (16), the process may be regulated at the level of substrate availability. The lactating mammary gland is unique in its requirement for the transport of free glucose across the Golgi membrane. GLUT1 is the only member of the facilitated diffusion glucose transporter family expressed in mammary gland. Neither GLUT1 nor any other isoform of the glucose transporter family is considered a Golgi resident, although a six-amino acid portion of GLUT4 does confer targeting to the trans-Golgi network and insulin-sensitive translocation to the plasma membrane in fat and muscle cells (17).

Several investigators have studied GLUT1 expression in the lactating mammary gland with conflicting results described above. The purpose of the experiments reported here was to systematically test whether the developmental regulation of the amount and subcellular trafficking of GLUT1 in mouse mammary gland indicates an important role for GLUT1 in the provision of substrate for lactose synthesis. In contrast with previous studies, independent methods were used, and multiple time points during the normal developmental cycle and the forced weaning-refeeding cycle were examined.

Importantly, consistent results from independent methods, subcellular fractionation followed by density gradient centrifugation, and epi- and confocal immunofluorescent microscopy indicate that during lactation, GLUT1 is targeted not to the plasma membrane as it is in most cells but to the Golgi. This resolves the contradictory previous studies and leads to the conclusion that GLUT1 is diverted from normal sorting pathways to the Golgi. Preferential targeting of GLUT1 to basolateral compared with apical membrane during lactation is consistent with the need of the mammary epithelial cell to take up glucose from the blood. More work is needed to demonstrate that GLUT1 actually controls the provision of glucose to the Golgi for lactose synthesis. The results do not exclude the previously suggested (8) possibility that a novel transporter, yet to be identified, resides in the Golgi.

The results suggest a tissue- and developmental stage-specific Golgi targeting mechanism for GLUT1. The structural determinants of targeting of proteins to Golgi are controversial. The transmembrane-spanning domain or specific amino acid motifs within it have seemed important (18). However, no common retention signal is apparent on examination of a large number of cloned glycosyltransferases (19). Other proposed explanations include the formation of nonmobile protein oligomers or the influence of the high cholesterol content of Golgi membranes on mobility of resident Golgi enzymes (20). Because our general understanding of the determinants of protein targeting to Golgi is limited, it is not surprising that the molecular basis of the hormonally regulated Golgi targeting of GLUT1 is unclear. Further study may reveal mechanisms regarding GLUT1 targeting that are relevant to other Golgi proteins as well. Hormonal regulation of GLUT1 subcellular targeting suggests a flexibility of the Golgi targeting machinery that has not previously been appreciated.

The reversible nature of the targeting of GLUT1 to Golgi and the time course over which changes in GLUT1 targeting are observed suggest a dynamic process that requires frequent suckling to maintain substrate supply for lactose synthesis. Future work will explore the relevance of this mechanism to the phenomenon of decreased production of human milk when nursing intervals are prolonged.