In vitro mammary epithelial cell models typically fail to form a consistently tight barrier that can effectively separate blood from milk. Our hypothesis was that mammary epithelial barrier function would be affected by changes in luminal ion concentration and inflammatory cytokines. Bovine mammary epithelial (BME-UV cell line) cells were grown to confluence on permeable supports with a standard basolateral medium and either high-electrolyte (H-elec) or low-electrolyte (L-elec) apical medium for 14 days. Apical media were changed to/from H-elec medium at predetermined times prior to assay. Transepithelial electrical resistance (Rte) was highest in monolayers continuously exposed to apical L-elec. A time-dependent decline in Rte began within 24 h of H-elec medium exposure. Change from H-elec medium to L-elec medium time-dependently increased Rte. Permeation by FITC-conjugated dextran was elevated across monolayers exposed to H-elec, suggesting compromise of a paracellular pathway. Significant alteration in occludin distribution was evident, concomitant with the changes in Rte, although total occludin was unchanged. Neither substitution of Na+ with N-methyl-d-glucosamine (NMDG+) nor pharmacological inhibition of transcellular Na+ transport pathways abrogated the effects of apical H-elec medium on Rte. Tumor necrosis factor alpha, but not interleukin-1β nor interleukin-6, in the apical compartment caused a significant decrease in Rte within 8 h. These results indicate that mammary epithelium is a dynamic barrier whose cell-cell contacts are acutely modulated by cytokines and luminal electrolyte environment. Results not only demonstrate that BME-UV cells are a model system representative of mammary epithelium but also provide critical information that can be applied to other mammary model systems to improve their physiological relevance.
- transepithelial electrical resistance
- apical cation concentration
- paracellular permeability
- inflammatory cytokines
mammary epithelium is responsible for both secretion of milk and the barrier to separate milk from blood and therefore has unique pathophysiological challenges that are difficult to address. When the barrier function of this secretory epithelium is compromised, permanent damage to the gland can occur. Change in milk electrical conductivity, related to ionic composition, is one of the earliest indicators of pathological changes in epithelial function in the bovine mammary gland (9, 31). The mammary epithelial barrier breaks down during gland inflammation, milk production is reduced, and the lack of separation of the milk compartment from the blood compartment is associated with a cascade of events that can cause permanent damage to the gland (16, 17, 24). However, any causal relationship between the change in epithelial barrier function and the change in milk ion content remains incompletely defined. Current tenets suggest that breakdown of the epithelial barrier allows electrolyte-rich plasma to enter the milk compartment, thus causing the changes in milk electrical conductivity associated with clinical or preclinical mastitis. An alternative explanation is that a pathologically induced elevation in electrolyte concentration contributes to the breakdown of mammary epithelial integrity.
Inflammatory mediators are recruited to the mammary gland under pathological conditions, although direct effects of cytokines on mammary epithelium have not been fully delineated. During mammary infection, cytokines interleukin-1 (IL-1β), interleukin-6 (IL-6), and tumor necrosis factor alpha (TNF-α) are elevated in the milk within 6–8 h following bacterial inoculation of the gland (1, 25, 27, 44). It has been speculated, from work in other tissues, that these chemicals signal changes in milk composition by directly or indirectly affecting the epithelium (4, 13, 14, 19, 30). It is possible that cytokine and chemokine recruitment may directly affect the barrier function of the epithelium. Thus, it is critical to determine if these cytokines directly affect the epithelium and to determine which aspects of epithelial function are affected.
Mammary epithelial barrier function has been difficult to quantify (reviewed in Refs. 20 and 32). A previous report from this laboratory suggested that luminal composition, which is dependent on epithelial ion transport, could affect the epithelial barrier integrity as measured by transepithelial electrical resistance (Rte) (29). Nonetheless, little is known regarding the time course over which this barrier breakdown occurs, whether a paracellular pathway is affected, or the target proteins that are involved in increased epithelial permeability. Examining the effects of changes in ionic composition on the epithelia is a critical step to understanding the progression of the breakdown of cell-cell contacts, as well as illuminating subsequent options for treatment of mastitis.
Many in vitro mammary epithelial cell lines are currently being used to study the dynamics of the gland and to examine a variety of pathologies, including breast cancer (11, 15, 26, 28, 37). In most cases mammary cancer cell lines of epithelial origin are being grown on solid supports, which precludes full membrane polarization and the opportunity to assess physiological function. Also, these systems have typically utilized the same high-electrolyte medium bathing both the apical and basolateral aspects of the epithelium. Attempts are being made to grow mammary epithelial cells on permeable supports (8). However, Schmidt et al. (29) provide the only publication to date in which transformed mammary epithelial cells were cultured on permeable supports in asymmetrical media that approximates the in vivo environment and indicates a culture-dependent change in barrier function.
The primary goal of this study was to test mammary epithelial integrity following changes in apical ionic composition (to simulate changes in milk electrical conductivity) using an in vitro bovine mammary epithelial model. This attempt to determine whether luminal electrolytes affect epithelial integrity is the first to concomitantly measure Rte, permeation by uncharged particles, and the distribution of proteins associated with tight junctions directly at the level of the epithelium in an in vitro system. A secondary goal for this study was to identify and characterize effects of inflammatory cytokines (TNF-α, IL-1β, IL-6) on epithelial barrier function in this in vitro model. We hypothesized that exposure to increased luminal ionic concentration or inflammatory cytokines would rapidly decrease epithelial barrier function. The results indicate that mammary epithelial tight junctions are dynamic structures that can be rapidly modified by TNF-α and by changes in the fluid electrolyte composition in the milk compartment. More importantly, the cytokine effects on mammary epithelial barrier function pale compared with the rapid and significant compromise of barrier function during exposure to elevated electrolytes in the luminal compartment. The decrease in barrier function is due, at least in part, to changes in localization of the tight junction protein occludin. This information brings into focus the need to reevaluate current paradigms in progression of mammary pathology.
MATERIALS AND METHODS
Cell culture media.
An immortalized bovine mammary epithelial cell line, BME-UV, was used for the present studies, and two media were used for cell culture. The medium used to bathe the basolateral aspect of the monolayers was described along with the BME-UV cells (41). This medium contains little lactose, has concentrations of electrolytes that closely mirror serum, and in the context of the current experiments will be termed “high electrolyte” (H-elec) medium to differentiate it from the low-electrolyte medium (L-elec) that is closer in electrolyte composition to milk (29), containing less than half the Na+ and Cl− and including 160 mM lactose (see Table 1 for analysis). The L-elec medium in the current studies is similar in composition to the alternative medium described by Schmidt et al. (29). Osmolality was 290 ± 5 mosmol/kgH2O in both H-elec and L-elec media, pH was adjusted to 7.3, and included an HCO3−/CO2 buffer system.
Cell culture protocols.
Stock cultures of BME-UV cells (provided by Jeff Smith, University of Vermont) were grown in H-elec medium on 25-cm2 plastic culture flasks (Cellstar; Frickenhausen, Germany) to a maximum of 65–75% confluency. Cells were dissociated for passage using a solution containing 0.25% (vol/vol) trypsin and 2.65 mM disodium ethylenediaminetetraacetic acid (EDTA) in phosphate-buffered saline for cell culture (PBScc; composition in mM: 140 NaCl, 2 KCl, 1.5 KH2PO4, 15 Na2HPO4). Dissociated and dispersed cells were seeded on permeable tissue culture supports (Snapwell Clear; Corning-Costar, Acton, MA) as described previously (29). Cells were maintained in culture on permeable supports for 15 days prior to assay, during which time growth to a confluent, polarized, electrically tight monolayer has been documented (29). H-elec medium bathed the basolateral aspect of the cells throughout all experiments. The apical aspect of the cells was exposed to either H-elec or L-elec medium as described below. Cells were incubated at 37°C in a humidified atmosphere containing 5% CO2. Media on both the apical and basolateral aspects of the cells were refreshed daily, with a final change occurring 4–8 h prior to experimental analysis.
Two sets of experiments were conducted to test for effects of changing the apical medium from, or to, H-elec medium. In both experiments, Snapwell trays were seeded with BME-UV cells in symmetrical H-elec medium. Twenty-four hours postseeding is considered “day 1” in culture. In the first experiment, the apical medium of five Snapwells in each tray was changed to L-elec medium on day 1 in culture. Beginning on day 4, 7, 10, and 13, successive monolayers were apically exposed to H-elec medium for the remaining duration of the culture period (Fig. 1A). One monolayer was exposed to H-elec apical medium throughout the entire culture period. In the second experiment, the H-elec apical medium was maintained for 1, 4, 7, 10, and 13 days before changing to apical L-elec medium on successive monolayers (Fig. 1B). In all cases, monolayers were evaluated for barrier integrity on postseeding day 15. At this time, monolayers were also fixed for assessment of tight junction-associated proteins by immunocytochemistry or harvested for Western blot analysis as described below.
Transepithelial electrical resistance.
Epithelial cell monolayers were mounted in modified Ussing flux chambers (model DCV9; Navicyte, San Diego, CA) in symmetrical Ringer solution (composition in mM: 120 NaCl, 25 NaHCO3, 3.3 KH2PO4, 0.8 K2HPO4, 1.2 MgCl2, 1.2 CaCl2) maintained at 37°C and continually circulated with an air-lift system using 95% O2/5% CO2. After open circuit transepithelial electrical potential was recorded, each monolayer was clamped to 0 mV with a multichannel voltage-clamp apparatus (model 558C; University of Iowa, Department of Bioengineering, Iowa City, IA). At 100-s intervals the monolayers were exposed to a 5-s, 0.5-mV bipolar pulse. Voltage and current measurements were made continuously and data were acquired digitally at 1 Hz, using a Macintosh computer (Apple Computer, Cuppertino, CA), an MP100A-CE interface (BIOPAC Systems, Santa Barbara, CA), and Aqknowledge software (ver. 3.2.6; BIOPAC Systems). Ohm's law (resistance = Δ voltage/Δ current) was applied to the resulting current deflections to determine Rte.
Transepithelial dextran permeation.
Movement of large, uncharged molecules across epithelial monolayers was assessed with FITC-conjugated dextran. Methods were similar to those used by Broughman et al. (3). Briefly, BME-UV monolayers were grown on Snapwell permeable supports with changes from or to H-elec apical medium on the days postseeding as indicated. Monolayers were washed with PBScc and then bathed in Ringer solution, both apically and basolaterally. Two different sizes of FITC-conjugated dextran, 9.5 kDa and 77 kDa (Sigma-Aldrich, St. Louis, MO), were utilized to assess epithelial permeability. FITC-conjugated dextran was introduced into the apical compartment of all wells at the outset of the assay. Monolayers were incubated at 37°C in a humidified atmosphere containing 5% CO2 for 60 min after which samples of basolateral solution were obtained for analysis. Filter supports lacking cells were included to verify that the culture support did not provide a significant barrier to dextran permeation and cell monolayers exposed to Ringer solution in the absence of dextran were employed to establish that no components of the assay system contributed to the fluorescence measurements. Fluorescence in the basolateral compartment was quantified in a Fluoroskan Ascent FL plate reader (Labsystems, Helsinki, Finland). Known dilutions of each dextran stock were included in each 96-well assay plate to construct a standard curve after background subtraction.
After exposure to the indicated culture conditions, BME-UV cell monolayers were washed in PBScc and fixed in 4% paraformaldehyde (Fisher Scientific International, Hampton, NH), and stored less than 24 h in phosphate buffered saline for cytochemistry (PBS; composition in mM; 150 NaCl, 5 KH2PO4, 15 K2HPO4, pH 7.2). Monolayers were permeabilized with 0.1% Triton X-100 in PBS, blocked for 1 h with goat serum (Gibco-BRL, Rockville, MD; 5% vol/vol in PBS). Primary antibodies to zonula occludens-1 (rat anti-ZO-1; Chemicon, Temecula, CA), and occludin (rabbit anti-occludin; Zymed, Carlsbad, CA), diluted 1:200 with 1% goat serum in PBS were incubated with the monolayers for 2 h at room temperature. Cells were washed 3 × 10 min in PBS and subsequently incubated for 2 h at room temperature with rhodamine-conjugated goat anti-rat (for ZO-1; Vector Laboratories, Burlingame, CA) or Alexa 488-conjugated goat anti-rabbit (for occludin; Molecular Probes, Eugene OR) secondary antibody. Once again, cells were washed 3 × 10 min in PBS. Cells were further incubated with 10 μM TO-PRO3 (Invitrogen, Eugene, OR) for 1 h and cells were washed briefly with PBS immediately prior to mounting on slides with Fluormount G (SouthernBiotech, Birmingham, AL). Images were assessed by both standard (Leica Microsystems AG, Wetzlar, Germany), and confocal (Carl Zeiss MicroImaging, Thornwood, NY) microscopy using appropriate filters for each fluorophore. Identical settings were used to obtain and process images derived from paired monolayers. Negative controls included protocols with the primary antibody omitted, the secondary antibody omitted, or, when available, peptide-preadsorbed primary antibody.
Western blot analysis.
BME-UV cell lysates were prepared in the 1.13 cm2 Snapwell culture substrates using Phosphosafe lysis buffer (Novagen, San Diego, CA) including 1% protease inhibitor cocktail (Sigma-Aldrich) and maintained in a frost-free freezer at −20°C until assay. Total protein content was determined using a bicinchoninic assay (Pierce, Rockford, IL) and 20 μg of total protein was loaded in each well of a 10–20% SDS-PAGE prepoured gel (Bio-Rad, Hercules, CA) for electrophoresis. Gels were run at 160 V for 45 min, then transferred to Millipore PVF-Immobilon membranes for 8.5 h at 90 mV. Transfer was confirmed by staining gels with Gelcode Blue (Pierce). Membranes were blocked in SuperBlock Blocking Buffer (Pierce) and probed with rat anti-ZO-1 and rabbit anti-occludin diluted in SuperBlock Blocking Buffer. Protein was visualized by enhanced chemiluminescence with Pico-chemiluminescence substrate (Pierce), using a peroxidase-conjugated secondary antibody (1:5,000). Membranes were stripped with Restore Western Blot Stripping Buffer (Pierce), and stripping was confirmed with visualization using Femto-chemiluminescence substrate (Pierce), a more sensitive chemiluminescence substrate to preclude any residual labeling. Membranes were reprobed with mouse monoclonal primary antibody for either GAPDH (Abcam, Cambridge, MA ) or β-actin (Sigma-Aldrich) as a loading control for densitometric analysis. Membranes were exposed on CL-Xposure film (Pierce) and analyzed using a Kodak RP X-OMAT (model M7B) film analyzer. Imaging and densitometric analysis were performed on a Fluor Chem 8900 Alpha Innotech Imaging System with Alpha Ease FC StandAlone Software (Alpha Innotech, San Leandro, CA).
BME-UV cells were seeded onto Snapwell permeable supports (in groups of six) as described above. The apical medium of three Snapwells in each tray was changed to L-elec medium on day 1, and additional wells were exposed apically to H-elec medium conditions for the duration of the 15-day culture period. Beginning on day 13, one L-elec and one H-elec medium monolayer were cultured in the presence of amiloride (10 μM; a selective blocker of the epithelial Na+ channel, ENaC), and one L-elec and one H-elec medium monolayer were cultured in the presence of 5-(N-ethyl-N-isopropyl)-amiloride (EIPA, 1 μM; an amiloride analog that preferentially inhibits the Na+/H+ exchanger, NHE) in the apical culture media for 24 h prior to evaluation of Rte in the Ussing chamber.
Alternative solutions for Ussing chamber experiments.
Reduced Na+ and/or Cl− solutions were employed for certain experiments by equimolar replacement of 120 mM NaCl with N-methyl-d-glucosamine·HCl (NMDG-Cl), Na-gluconate, or NMDG-gluconate. Alternatively, in some experiments 60 mM NaCl was replaced by 120 mM lactose to test for effects of electrolyte reduction.
BME-UV cell monolayers, cultured in the presence of L-elec apical medium, were exposed to cytokines that have been associated with mammary inflammation (25, 27, 44; reviewed in Ref. 1), including TNF-α (0.5 μg/ml), IL-1β (0.1 μg/ml), or IL-6 (1 μg/ml) in the apical medium for 8 or 12 h prior to assessment of Rte in modified Ussing chambers.
Quantitative results were analyzed via ANOVA using SAS (SAS Institute, Cary, NC). Data are presented as means ± SE. Differences are considered statistically significant when the probability of a type I error is <0.05.
H-elec apical medium decreases transepithelial electrical resistance.
Initial experiments document and extend observations (29) regarding the relationship of apical electrolyte concentration to Rte by systematically assessing the time course over which differences in Rte can be induced by changes in apical electrolyte concentration. Results presented in Fig. 2 demonstrate that L-elec apical medium is consistently associated with the greatest Rte. In the first set of experiments (Fig. 2A), monolayers exposed only to L-elec medium for the entire culture period exhibited more than fourfold greater Rte than monolayers exposed to H-elec medium throughout. More importantly, exposure to H-elec medium for only one day was associated with greater than 25% reduction in Rte (432 ± 64 vs. 309 ± 6 Ω·cm2). In fact, within 1 h of continuous exposure to H-elec medium, barrier function, as measured by Rte, began to decline (data not shown). Four-day exposure to H-elec medium showed further incremental reduction in Rte to less than 30% of the L-elec medium value. Although a trend toward further reduction in Rte was observed with additional exposure to H-elec medium, the incremental changes did not achieve statistical significance. These results demonstrate that short-term apical exposure to H-elec medium compromises the mammary epithelial barrier function, as assessed by electrical means.
L-elec apical medium enhances transepithelial electrical resistance.
A transition of apical medium from H-elec to L-elec medium is associated with greater Rte (Fig. 2B) and results provide a virtual mirror image of Fig. 2A. Monolayers apically exposed to H-elec medium throughout the 14-day culture period exhibited the lowest Rte of all wells in each block (138 ± 17 Ω·cm2). Monolayers exposed to H-elec medium for 13, 10, and 7 days before an apical transition to L-elec medium for 1, 4, and 7 days, respectively, exhibited incrementally greater Rte. Differences in Rte were not associated with longer periods (i.e., up to 14 days) of apical exposure to L-elec medium. These results suggest that a reduction in apical electrolyte concentration, to a value that more closely approximates that of typical (i.e., healthy) milk, rapidly promotes the enhancement of the mammary epithelial barrier and an electrically tight monolayer.
Apical H-elec medium increases paracellular permeability.
Reductions in Rte might reflect an increase in transcellular permeability, paracellular permeability, or some combination of these two permeation pathways. Thus, gradient-driven permeation of 9.5 and 77 kDa dextran was employed to assess changes in the paracellular pathway resulting from differences in the apical solution. As shown in Fig. 3A, monolayers exposed apically to L-elec medium for the entire culture period exhibited the lowest permeation rates (i.e., most substantial barrier) to diffusion compared with other time points for both 9.5 and 77 kDa dextran. Apical exposure to H-elec medium for as little as 1 day, following 13 days of L-elec medium, was associated with significantly greater permeation of both sizes of dextran. Incrementally greater permeation was observed with 7 days exposure to apical H-elec medium. The greatest permeation was observed when monolayers were exposed apically to H-elec medium throughout the entire culture period. These observations demonstrate that short-term exposure to a serum-like medium on the apical face of bovine mammary epithelial cells promotes a loss of the barrier integrity that is associated with the opening of the paracellular pathway to allow the permeation of high molecular weight solutes; an observation that is consistent with the reduction in Rte (Fig. 2A).
Apical L-elec medium decreases paracellular permeability.
To test further the hypothesis that shifts in the apical medium composition modulate paracellular permeability, experiments were conducted in which the apical medium was changed from H-elec to L-elec medium at specified time points. Results presented in Fig. 3B demonstrate that the “leakiness” of the mammary epithelium is reduced by apical exposure to L-elec medium incrementally, depending upon the duration of exposure. As predicted, the greatest permeation was observed in monolayers continually exposed to H-elec medium. However, exposure to L-elec medium for the shortest time tested in these experiments, 7 days, was associated with a decrement of >70% in the magnitude of both 9.5 and 77 kDa dextran permeation. Also as predicted, based upon the results presented in Fig. 2B, longer periods of exposure to L-elec medium were not associated with further reductions in permeation. Taken together, results presented in Figs. 2 and 3 demonstrate that the mammary epithelial barrier, whether quantified by either Rte or by the permeation of large, uncharged solutes, is sensitive to the medium composition in the apical compartment.
Occludin, but not ZO-1, distribution is altered by apical medium.
To test whether reduced integrity of the epithelial barrier was related to changes in junctional proteins, and to rule out the possibility that apical H-elec medium was causing a portion of the cells to “release” from the culture substrate and/or the epithelium, experiments were conducted to visually evaluate the epithelium and to assess the distribution of the tight junction proteins ZO-1 and occludin. Examination of H-elec medium-exposed BME-UV cell monolayers revealed no evidence for the loss of cells. In every case, a survey of at least five objective fields per filter revealed a confluent monolayer with a cobblestone appearance throughout. Monolayers exposed to L-elec medium for 14 days were colabeled with anti-occludin primary antibody that was visualized using an Alexa 488-conjugated secondary antibody, and with anti-ZO-1 primary antibody that was visualized using either Alexa 594- (confocal microscopy) or rhodamine- (light microscopy; data not shown) conjugated secondary antibody. TO-PRO3 staining of nuclear material was included to provide an additional reference point within each cell. Occludin (Fig. 4A) and ZO-1 (Fig. 4B) immunoreactivities were identified in the apical-lateral portions of all cells in the epithelial monolayer. Figure 4 provides the image at a single focal plane that is near the apical aspect of the cells and was selected from a stack of images through the BME-UV cell monolayer. In the BME-UV cell monolayer, each cell is fully circumscribed by immunoreactivity for each of the epitopes. The combined image (Fig. 4D) clearly shows that the distribution of these two tight junction-associated proteins is virtually identical in these conditions.
Experiments were conducted to determine whether distribution or intensity of these tight junction proteins correlated with the differences noted in the Rte and dextran permeability studies. No differences in apparent quantity or localization of ZO-1 immunoreactivity were identified via light or confocal microscopy between monolayers exposed to H-elec or L-elec medium for the entire 14 day culture period (Fig. 5). However, significant differences in occludin immunoreactivity were identified between monolayers exposed to L-elec and H-elec media throughout the culture period (Fig. 6). Figure 6A includes 12 individual plane images (z-stack) from a 14-day-L-elec-treated monolayer. The stacked images were taken at 1 μm intervals sagittally through the monolayer and are presented in sequential order as indicated. Occludin immunoreactivity is clearly present in the apical portion of each cell and is particularly concentrated at the cell margins (panels 2 and 3). Some immunoreactivity is also present in the cytosol at the level of the nucleus (panels 4–7) and little immunoreactivity is present in the basal portion of the cells that is near to the filter substrate (panels 9–12). In contrast, occludin immunoreactivity is substantially less in BME-UV cells that were apically exposed for 14 days to H-elec medium (Fig. 6B). The layout of the confocal images is similar to that presented for the paired filter in Fig. 6A with the upper left image at the apical aspect of the monolayer and the lower right panel nearest to the culture substrate. In this case, progressive optical sections through the monolayer demonstrate that the pattern of occludin immunoreactivity is distinctly different in BME-UV cell monolayers cultured in apical H-elec medium compared with apical L-elec medium-exposed monolayers. These results are consistent with observations presented in Figs. 2 and 3, and suggest that changes in occludin expression or distribution might account for the changes in Rte and dextran permeation. Thus, we next sought to test for redistribution of occludin that could occur within different time frames.
In an experiment tightly paired with the images in Fig. 6, occludin immunoreactivity was assessed in monolayers grown in L-elec medium and switched to H-elec medium for either 1 or 7 days, as well as monolayers grown in H-elec medium and switched to L-elec medium for 4 and 7 days. An apparent redistribution of occludin immunoreactivity was evident with 24 h of apical H-elec medium exposure (Fig. 7A). Compared with Fig. 6A, occludin immunoreactivity is punctate, less intense and less defined, but still predominantly at the apical margins (panels 2 and 3). With 7 days exposure to H-elec medium (Fig. 7B), no occludin immunoreactivity is observed at the expected location of the tight junction. There is little difference between occludin distribution following 7 or 14 days exposure to H-elec medium (compare Fig. 7B to Fig. 6B). Alternatively, occludin immunoreactivity is apparent at the apical cell margins with 4 days exposure to apical L-elec medium after 10-days exposure to H-elec medium (Fig. 7C), although occludin immunoreactivity remains obvious in the cytosol in the nuclear and basal portions of the cells. Results presented in Fig. 7D show that following 7 days of apical L-elec medium exposure, intense occludin immunoreactivity is present at the apical cell margins and lesser immunoreactivity is present in the cytosol. Taken together, data presented in Figs. 2–7 demonstrate that apical exposure of BME-UV cell monolayers to a “blood-like” medium is associated with an apparent loss of barrier integrity, as measured by electrical resistance, solute permeation, and protein distribution. Alternatively, barrier integrity is enhanced with exposure to a “milk-like” medium. The mechanism(s) by which the epithelial cells sense the apical composition and the signaling pathway leading to the response, however, remain to be determined.
Occludin expression is not altered by apical medium.
BME-UV cell monolayers exhibited similar levels of occludin expression regardless of apical culture conditions over the time course of the experiment. Cell lysates from each treatment, containing similar amounts of total protein, were applied in sequential lanes of a gel for Western blot analysis and densitometric analysis, including GAPDH and β-actin controls for normalization of total protein loading. Two clear bands were evident for occludin protein. Although it is possible that these multiple mobilities are due to different phosphorylation states for the occludin protein, experiments were not performed to test this possibility. No significant differences in immunoreactive occludin protein were identified in monolayers exposed to different apical conditions (Fig. 8). These observations indicate that occludin redistribution, rather than expression, is one of the defining factors for loss of barrier function across this epithelium.
Na+ transport blockers fail to affect medium-associated changes in epithelial barrier function.
It was speculated that the effect of H-elec medium on the epithelial barrier might result from the relatively high Na+ concentration that could affect cellular metabolism by changes in membrane potential, cytosolic ion concentration, or cytosolic pH. To investigate these possibilities, cells were cultured in conditions to minimize Na+ entry via conductive or exchange pathways. In the first experiment, amiloride, EIPA, or vehicle control were included in the apical growth medium for 24 h prior to assay. As shown in Fig. 9, the same general trend of lower Rte in the presence of H-elec medium was observed in the presence of amiloride or EIPA. There was, however, a modest, although not statistically significant, trend toward an amiloride-associated increase in Rte in the presence of apical H-elec medium. These results might be construed to indicate that an amiloride-sensitive ion channel (e.g., ENaC) is present in these cells. However, it must be pointed out that the basal ion transport was not different between these treatment groups (data not shown). Additionally, it was reported previously (29, 33) that amiloride-sensitive current is readily observed in corticosteroid-treated monolayers, but not in untreated monolayers, as were used in the current study. Thus, the results suggest that the effect of apical H-elec medium to decrease Rte is not mediated by a cytosolic event that depends on the entry of Na+ into the cell by either ENaC or NHE.
Substitution of impermeant ions fails to affect medium-associated changes in epithelial barrier function.
The hypothesis that electrolyte permeation into the epithelial cells contributed to the change in Rte was tested further by conducting ion substitution experiments. Monolayers were cultured in the presence of apical L-elec medium until the day before assay, when the apical solution was changed to reduce apical lactose concentration and increase apical electrolyte concentrations with permeant monovalent ions (Na+, Cl−), or monovalent ions (N-methyl-d-glucosamine [NMDG+], gluconate) that are typically considered to be impermeant. Results presented in Fig. 10 demonstrate that the Rte is significantly lower in all monolayers exposed to reduced lactose/increased electrolytes in the apical medium. The decrement in Rte at first appears greater in H-elec medium than in the presence of NMDG-Cl−, although this difference is not statistically significant. Furthermore, NMDG-gluconate replacement of both Na+ and Cl−, is associated with a greater decrement in Rte. Taken together, results presented in Figs. 9 and 10 suggest that either an increase in extracellular apical electrolyte concentration or a decrease in lactose concentration is associated with a significant change in Rte, although permeation of electrolytes into the cells is not required for the effect to be observed. Thus alternative experiments must be designed and conducted to identify stimuli and/or signaling pathways that modulate the mammary epithelial barrier.
TNF-α decreases BME-UV monolayer barrier function.
Chemicals derived from inflammatory cells, including the cytokines TNF-α, IL-1β, and IL-6, are increased under pathological conditions in the bovine mammary gland and may directly or indirectly affect epithelial cell function. Experiments were conducted to test for effects of these inflammatory cytokines on mammary epithelial integrity via analysis of Rte. Exposure to TNF-α for 8 h, a time point at which these inflammatory cytokines have been detected to be elevated in vivo, was associated with a 30% decrease in Rte (Fig. 11). Twelve hour exposure to TNF-α was associated with more variable results that were not significantly different from the control (data not shown). No significant effect on Rte was noted after exposure to IL-1β or IL-6 (Fig. 11). Likewise, increased time of exposure, at 24 and 30 h, demonstrated no effect on Rte (data not shown). These observations demonstrate the importance of examining each of these effects both separately and together to determine the overall effect of challenge to the gland within the optimal physiological context.
Mammary epithelial cells respond to differences in the environment in the apical (milk) compartment with rapid changes in the barrier function of the epithelium. To our knowledge, we were the first to demonstrate an association between apical solute composition and mammary epithelial barrier function with an in vitro system (29). These observations have been systematically extended to characterize the time course over which the changes occur and to determine that the barrier can be modulated to either higher or lower resistivity shown in three ways: Rte, dextran permeation, and occludin localization. In addition, the new data show that there can be direct modulation of the mammary epithelial barrier by certain cytokines. These results indicate that mammary epithelial tight junctions can be modulated by a variety of solutes expected to be present in bodily fluids bathing the apical and/or the basolateral aspect of these cells. Inferences can be drawn that relate to nutrition, child nurturing, dairy production/management, and perhaps to breast cancers. Much work has been done to examine the modulation of junctional proteins in small animal models and in mammary carcinoma cells in vitro, however, the mechanisms of tight junction modulation in the healthy gland are less defined (10, 21, 22, 35, 43). Additionally, the in vitro systems that have been used previously are suboptimal representations of the physiological environment because they employed solid supports and/or symmetrical buffers. This project provides a model system representative of mammary epithelium, and critical information that can be extrapolated to other mammary model systems to improve their physiological relevance to the study of both normal and compromised mammary epithelial integrity.
BME-UV cells provide the basis for an in vitro system to characterize mammary epithelial cell function. The original description included the claim that these immortalized cells exhibited many characteristics of mammary epithelia, including a cobblestone appearance when grown on solid supports, cytokeratin production, and synthesis of α-lactalbumin and αs1-casein (41). It has since been reported that BME-UV cells respond to a variety of hormones and pharmacological agents with changes in cell signaling, cellular metabolism, proliferation, and apoptosis (6, 12, 39, 40, 42). It has also been shown that BME-UV cells grow on a permeable support system to form a confluent, polarized monolayer that is capable of regulated ion transport (29). The fact that these cells form a polarized and electrically tight epithelial monolayer provides opportunities to study both barrier function and ion transport across the mammary epithelium in a way that is not possible in the native gland, with freshly isolated cells, or in a 3-dimensional culture system.
Leaders in the field have noted that the mammary epithelium is a structure with complex geometry and, therefore, have suggested that an in vitro model such as this would be amenable to study with powerful techniques such as the Ussing chamber (32). The BME-UV cell system is amenable to tightly controlled systematic experimentation. Optimization of the system by Schmidt et. al. (29) demonstrated the benefit of a milk-like apical medium. The current report utilizes the polarized monolayer configuration and demonstrates that BME-UV cells exhibit functional responses to changes in electrolyte composition and to cytokines that would be predicted based upon in vivo observations. The time course of these changes is important to understanding the progression of epithelial barrier modulation within the mammary gland.
The results indicate that mammary epithelial integrity is modulated by changes in the electrolyte or carbohydrate concentration in the apical medium. Three separate and complementary techniques were employed to document differences between the treatments. First, Rte was employed as a rapid and sensitive indication of the ease with which small charged solutes, especially monovalent ions, move through the epithelium. Since the lactating mammary gland secretes a high volume (30–35 kg/day in dairy cows) of isosmotic solution, one might expect the epithelium to exhibit a relatively low resistance. However, the composition of milk is distinctly different from serum, which might suggest the presence of a “tight” epithelium. Furthermore, it was reported that cultures of freshly isolated mammary epithelial cells exhibited resistances of >1,000 Ω·cm2 (34), although the contribution of the substrates were not reported. Thus, the previous report (29) that BME-UV achieved a maximal Rte of <200 Ω·cm2 was somewhat surprising and suggested that optimal culture conditions had not been fully defined. The hypothesis that a “milk-like” medium on the apical face of the monolayer would enhance resistance was tested and determined to be correct. The goal was to create a medium that was low in total electrolytes, relatively high in K+, relatively high in lactose, and isosmotic with the basolateral medium. There were, however, certain constraints that guided the apical medium construction. First, one cannot remove a solute and maintain osmolarity. Hence, there is not an experiment to test for effects of changing electrolyte concentrations without a concomitant change in carbohydrate to maintain osmolarity. A previous report shows that similar results are obtained when either mannitol or lactose are used as the compensating osmolyte when electrolytes are reduced (29). Additionally, the optimal medium defined by Zavizion et al. (41), included 15% serum, which includes a substantial amount of Na+ and Cl−. It was determined that a systematic evaluation of serum withdrawal was beyond the scope of the current studies. Rather, the current study focused on the time course over which medium-induced differences in Rte could be observed.
In every experimental block, recent duration of L-elec medium exposure is associated with the greatest electrical resistance. Although block to block and experiment to experiment variation in the maximal resistance was observed, continuous exposure to H-elec medium is associated with the lowest Rte in every block. It was previously reported that BME-UV cell monolayers reach a maximal plateau in Rte between 7 and 10 days of culture that is maintained consistently to at least day 14 (29). Thus, the observations that the transition from apical L-elec to H-elec medium for as little as one day at the end of the 14 day culture period results in a >25% reduction in Rte is striking. The results suggest that high electrolyte/low carbohydrate at the apical membrane causes a change in the epithelium that permits gradient-driven flux of small charged solutes, either through transcellular or paracellular routes. Similarly, a one day transition to L-elec medium can result in a doubling of Rte. The pathway(s) that are affected by the change in apical composition can be readily modulated in either direction. Furthermore, the response appears to be complete or nearly complete within four days. There is a fourfold reduction in Rte with four days of apical H-elec medium exposure and there is a fourfold increase in Rte with four days of apical L-elec medium exposure. This time frame for changes in barrier function is important because it parallels the period over which substantial changes are seen in milk composition and in mammary function at the initiation of lactation, at the initiation of involution, and in the course of mastitis.
Dextran permeation across the BME-UV monolayer indicates flux through the paracellular pathway. The electrical parameters described above are blind to transcellular vs. paracellular movement and are insensitive to the movement of any nonionic solute. Numerous mechanisms, both transcellular and paracellular, have been described to account for solute movement across mammary epithelium (3). Thus, a protocol was conducted to assess flux through the paracellular pathway. The experimental design tested for changes in paracellular permeation that paralleled the changes observed in electrical measurements and assessed the approximate size of the solutes that could readily traverse the epithelium. The results show that the BME-UV cells can be a formidable barrier to the diffusional movement of large solutes, with basal permeation rates that are similar to a canine kidney epithelial cell line (MDCK), and to T84 (colonic origin) and Calu-3 (airway origin; Ref. 3 and unpublished observations). The experimental results provide compelling evidence that changes in the apical medium affect the paracellular pathway with a time course that is similar to that observed for the effects on Rte. One day of exposure to H-elec medium is associated with a doubling in the rate of permeation by both the small (9.5 kDa) and large (77 kDa) dextran molecules. Paralleling the changes observed for Rte, dextran permeation was incrementally greater after longer periods of H-elec medium exposure. The results with 7 or more days exposure to L-elec medium are consistent with the Rte measurements in that 7 days exposure to L-elec medium produces a significantly greater barrier to solute flux and that, within these experiments, no additional incremental changes are associated with additional exposure to L-elec medium. Dextran of 77 kDa was selected because of its similarity in size to serum albumin (∼66 kDa). Thus, one can infer from these results that changes in the paracellular pathway that are caused by the modified apical medium are sufficient to allow for the permeation of albumin into the milk. Both the appearance of albumin and the elevation in electrolytes have been used as clinical measures of mastitis (9, 18, 23). The fact that milk conductivity is used as an early indicator of mastitis suggests, but does not demonstrate, that changes in milk electrolyte composition precede changes in the epithelial barrier integrity. Thus the current results provide impetus to conduct systematic measurements in milk electrolyte concentration and serum albumin content at the onset of mastitis.
Immunocytochemistry revealed that occludin, but not ZO-1, was absent from the expected apical lateral location following exposure to H-elec medium. Systematic metabolic assays were not conducted, so no inferences can be drawn regarding the underlying mechanisms that are affected. However, there appears to be a consistent amount of overall occludin immunoreactivity in the epithelial cells following H-elec medium exposure, when examined via Western blot analysis, but less localization of occludin at the level of the tight junctions. The possibility that the H-elec medium treatment caused a general redistribution of proteins or modified overall cell structure is ruled out by the observation that ZO-1 distribution and intensity were unaffected by the H-elec medium treatment, whereas specific redistribution of occludin protein was apparent. It is intriguing that a similar kinetic profile is observed for the decrease in resistance, increase in permeability, and decrease in occludin immunoreactivity. The results clearly show that these modifications in epithelial function occur concurrently during exposure to H-elec medium. Additionally, it is shown that transition to L-elec medium affects all three parameters (Rte, dextran permeation, occludin distribution) in the opposite direction of that seen with H-elec medium and with similar time course.
The magnitude and duration of exposure to altered apical medium composition required to cause a change in epithelial integrity or occludin distribution remain to be determined. The current results show that an abrupt and substantial change in apical medium composition causes a change in barrier Rte, dextran permeation and occludin distribution that can be readily observed within one day. However, one would not necessarily expect to observe such abrupt changes in milk composition in a physiological or clinical setting, although mammary saline infusion has been employed in a research setting (32). Furthermore, ion transport mechanisms that are present in mammary epithelium and account for electrolyte movement into or out of the milk compartment remain to be defined, along with their associated regulatory cascades. Knowledge of these mechanisms is required to establish pharmacological targets that can be used to modulate milk electrolyte composition in vivo. Evidence has been provided for the anion channel that is mutated in cystic fibrosis (CFTR) and for ENaC (2, 29) in mammary epithelial cells, along with regulation by both hormones and neurotransmitters. The magnitude or rate by which milk electrolyte composition can change, however, has not been determined. Clearly, additional experiments are required to test for effects of more subtle changes in apical composition on epithelial function.
A mechanistic link between apical medium composition and tight junction integrity has not been ascertained. The initial supposition was that either Na+ or Cl− in the apical medium might affect membrane electrical potential or, by movement through channels or transporters, cytosol composition to precipitate a change in cell function. This hypothesis was tested both pharmacologically and by substitution with impermeant ions. The results, however, were not particularly instructive except to indicate that these hypotheses were inadequate. Likewise, an osmotic effect is ruled out by meticulous care to insure that all media were isosmotic (290 mosmol/kgH2O). The possibility that a high apical lactose concentration is required to enhance barrier integrity is ruled out by a previous report, which showed that similar results were obtained with mannitol as the compensating osmolyte (29). That a mechanism is not currently revealed does not diminish the impact of the observations that changes in apical composition affect epithelial integrity. Rather, these observations provide additional impetus to identify players in the regulatory cascade.
Times when luminal electrolyte composition would be expected to change, including parturition and involution, are associated with increased incidence of infection, inflammation, and cytokine recruitment (1, 5, 7, 14, 16, 21). Cytokines, including TNF-α, IL-1β, IL-6, and IL-8, have been reported to modulate the mammary epithelial barrier in vivo (44). Intramammary TNF-α infusion is associated with the recruitment of neutrophils, decreased milk proteins, and the appearance of serum proteins in milk (38). TNF-α has also been shown to downregulate occludin expression in other cell systems (36), and to alter Na+ and Cl− movement across epithelial barriers (19). Whether the effects are direct or indirect, however, is still open to question. The results presented in this manuscript indicate that TNF-α has a direct effect on mammary epithelial cells to cause a breakdown in the barrier function, with the same time course that has been observed in vivo. Alternatively, the results failed to support a direct role for IL-1β or IL-6 in modulation of Rte in mammary epithelium. It is important to note that the conditions used in the present assays were not selected to mimic any particular clinical study. Thus, the discordance may reflect differences in the health status of the mammary gland that was used relative to the in vitro cell system. The results demonstrate that the BME-UV cell system is sensitive to selected cytokines and will provide an excellent system to delineate the cellular mechanisms that are involved in the response.
The current results provide a new factor, luminal electrolyte composition, that must be considered as one investigates mammary epithelial function. Changes in epithelial function can occur rapidly and do not require ion permeability into the cells. Rather, mammary epithelial monolayers are extremely sensitive to changes in luminal fluid composition, which is one of the earliest measurable indicators of mastitis. This evidence is complimented by observations that lowering the apical ion concentration leads to an increase in Rte, indicating enhanced barrier function. These results identify new factors that contribute to the progression of mastitis damage in the mammary gland and offer hope for new targets to circumvent (prevent) or treat inflammatory disease within the mammary gland. Reducing direct damage to the epithelium via altering cytokine effects, decreasing electrolytes, or modulating ion transport mechanisms may provide new options for investigation and ultimately for intervention. Work toward understanding the mechanism(s) of action that leads to changes in luminal ion concentration, cytokines, and epithelial remodeling will provide novel targets for prevention and treatment of mastitis at the earliest stages of the disease.
In summary, the current results demonstrate, with an in vitro model, that cytokines and apical electrolytes rapidly affect the epithelium that separates milk from blood. The epithelial barrier function is directly modulated by the local environment and some cytokines. These data provide information for those working with other in vitro mammary models to improve the quality of those systems by more closely representing the in vivo environment. Most importantly, the data indicate that there can be a feedback mechanism whereby the composition of milk that is secreted by the epithelium can affect barrier integrity. This observation provides impetus to define the cellular mechanisms that account for the low electrolyte content of milk and to seek interventions to therapeutically target the activity of these mechanisms.
Research was supported by the United States Department of Agriculture Grant 2003-35206-14157 (to B. D. Schultz), and fellowship support from the Kansas State University Center for Basic Cancer Research (to J. Erickson). This manuscript represents contribution number 06-90-J from the Kansas Agricultural Experiment Station.
The authors thank Ryan Carlin and Dr. James Broughman for technical assistance. We thank Dr. Catherine Uyehara for manuscript review and insightful discussion. We thank the American Physiological Society for the opportunity to participate in the APS Writer's Conference, and the guidance it provided for manuscript preparation and publication. For assistance with confocal microscopy, the authors thank Dr. Miriam D. Burton and the confocal facility supported by NIH P20 RRO17686.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society