| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,2
* Department of Dermatology, Medical University of Vienna, Vienna, Austria; and
Centre de Recherches et d'Investigations Epidermiques et Sensorielles, Neuilly sur Seine, France
2Correspondence: Department of Dermatology, Medical University of Vienna, Waehringer GuerA-1090 Vienna, Austria. E-mail: erwin.tschachler{at}meduniwien.ac.at
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS.RESULTSDISCUSSIONREFERENCES |
|---|
Key Words: angiogenesis • lactation • alveologenesis • mammary gland differentiation
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS.RESULTSDISCUSSIONREFERENCES |
|---|
. In adolescent virgin mice, the glands form a branched ductal system embedded in the mammary fat pad. Ductal branching increases during puberty and alveoli bud from the tips of the ducts. Although this process is limited in virgin mice, during pregnancy and lactation, massive alveologenesis and differentiation of the epithelial tissue produce dense milk secretory lobulo-alveolar clusters that completely replace the adipose tissue in the fat pad (reviewed in refs. 1
2
3
). These alveoli consist of a single layer of polarized epithelial cells, the site of milk synthesis, and are surrounded by contractile myoepithelial cells that aid in milk ejection (4)
.
The morphogenesis and differentiation of the mammary gland are controlled by a plethora of hormones, cytokines, adhesion molecules, and signaling factors (reviewed in refs. 1
, 3
, 5
, 6
), and require an increased supply of nutrients, oxygen (7)
, and presumably fluid. Accordingly, concomitant to lobulo-alveolar expansion and differentiation, there is an increase in blood supply to the mammary gland during pregnancy, achieved by the formation of numerous capillary sprouts, as well as a remarkable change in the organization of the vasculature. Thus, by midpregnancy the expanding alveoli are surrounded by a network of capillaries that are in intimate contact with the alveolar epithelial cells (7
, 8)
. During the latter part of pregnancy and during lactation, angiogenesis by capillary sprouting ceases. However, endothelial cell surface area is increased by the remodeling of existing capillaries (intussusception) and by the formation of microvilli and folds on the luminal and basal endothelial cell surfaces, respectively (7
, 9)
. In addition, mitochondria (important for the energy supply) and pinocytotic vesicles (implicated in solute transfer) increase in number (9
, 10)
. Finally, capillary permeability increases, peaking on 
day 10 of lactation (11)
. All of these changes are indicative of highly activated capillary endothelium and optimized substance transfer between the circulation and mammary gland tissue.
VEGF (or VEGF-A), also called vascular permeability factor, is the founding and most important member of a family of proangiogenic cytokines that includes VEGF-B, VEGF-C, VEGF-D, and placenta growth factor (PlGF). VEGF is essential for vasculogenesis: mice lacking either VEGF or its receptors, vascular endothelial growth factor receptor-1 (VEGFR-1, also called flt-1) and vascular endothelial growth factor receptor-2 (VEGFR-2, also called flk-1/KDR), die in utero (see refs. 12
13
14
for recent reviews). VEGF induces potent blood vessel endothelial cell proliferation and stimulates vascular permeability. Its activities are mediated by the specific tyrosine kinase receptors, VEGFR-1 and VEGFR-2, with VEGFR-2 regulating the mitogenic and permeabilizing activity of VEGF, whereas VEGFR-1 is thought to act as a negative regulator of VEGFR-2. These receptors are expressed predominantly on blood vessel endothelial cells (see ref. 12
for a recent review).
In the adult organism, angiogenesis is restricted to pathophysiological processes such as wound healing, inflammation, or tumor formation (13)
, and VEGF-dependent blood vessel formation has been unequivocally implicated in all of these processes. In addition, VEGF and its receptors are up-regulated in the ovary during corpus luteum formation (ibid) and in mammary gland tissue during pregnancy and lactation (15
16
17)
, suggesting that it could be important for regulating the changes in the vasculature at these times.
We have generated mice in which VEGF has been inactivated specifically in keratin 5 (K5)-expressing tissues by means of the Cre-LoxP system and have shown that VEGF is important for angiogenic processes in adult mouse skin (18
; C. Barresi, H. Rossiter, M. Ghanaddan, F. Gruber, E. F. Wagner, and E. Tschachler, unpublished results). Besides being active in epidermal keratinocytes, however, K5 promoter constructs also drive gene expression in other epithelial tissues such as myoepithelial cells of the mammary glands (19
, 20)
. We now report that female VEGF-A
k5-cre/k5-cre mice are unable to successfully nurse pups, and that during lactation, lobulo-alveologenesis and angiogenesis are reduced, but not completely abrogated. In contrast, blood vessel function within the lactating mammary glands is severely impaired. Taken together, these results suggest that VEGF is partially dispensable for angiogenesis during lactation, but is crucial for proper blood vessel function.
|
MATERIALS AND METHODS. |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS.RESULTSDISCUSSIONREFERENCES |
|---|
k5-cre/k5-cre (mutant) mouse lines and the PCR primers used to identify them (VEGFc5R.2 and VEGF322.F) have been described before (18
, 21)
. Briefly, a male mouse bearing the Cre recombinase transgene under the control of the K5 promoter (22)
was mated to females homozygous for the floxed exon 3 of VEGF (21)
and the Cre transgene-bearing offspring were bred to homozygosity for the LoxP sites. VEGF would thus be expected to be inactivated in a tissue-specific manner (i.e., only in tissues with an active K5 promoter) in mice that bear the Cre transgene and are homozygous for the floxed VEGF exon 3 (ibid). Littermates not inheriting the Cre transgene were used as controls as much as possible. Genomic DNA from virgin mammary glands and other tissues was prepared using the DNeasy Tissue Kit (Qiagen, Vienna, Austria) according to the manufacturer’s instructions, and 40–50 ng was amplified by PCR to determine tissue specificity of Cre-LoxP-mediated recombination of VEGF exon3. Unless stated otherwise, whole mammary gland tissue was harvested for all other experiments during the second pregnancy or after the second birth, after which pups survived to day 3 (relevant for litters from mutant dams only). To ensure similar suckling stimuli, all but three pups were removed from control mothers 36 h after birth. Samples from mammary glands and other tissues destined for genomic DNA preparation were snap-frozen and those for RNA were placed in RNAlater (Ambion, Austin, TX, USA). Both DNA and RNA samples were stored at –80°C until use.
Analysis of VEGF expression by ELISA and real-time PCR
Protein lysates of mammary glands from 3–12 mice/group were prepared by sonication in 1% nonidet P40 detergent (IGEPAL; Sigma-Aldrich, St. Louis, MO, USA) containing complete protease inhibitor cocktail (Roche, Mannheim, Germany). After removal of cell debris by centrifugation, protein content was measured by the BCA method (Pierce, Rockford, IL, USA) and VEGF content was measured by ELISA (R&D Systems, Minneapolis, MN, USA). For preliminary gene expression analysis in virgin, pregnant, and lactating mice, RNA from whole mammary glands from two or three mice/group was prepared using Trizol Reagent (Invitrogen, Carlsbad, CA, USA) in combination with chloroform/isopropanol extraction or with the High Pure RNA Tissue Kit (Roche). However, the Trizol isolation technique, followed by the purification kit, were applied for the RNA series that was used for statistical analysis of gene expression in lactating glands, since this double purification procedure yielded the highest quality RNA. RNA samples from five to seven mice/group were reverse transcribed with the GeneAmp Kit using MuLV-RT and oligo dT primers (Applied Biosystems, Penzberg, Germany). mRNA expression was then quantified by quantitative real-time PCR (qPCR) using the LightCycler technology and the LightCycler Fast Start DNA Master SYBR Green I Kit (Roche Applied Science, Basel, Switzerland) according to the manufacturer’s protocol.
Primers for qPCR were designed according to the strategy of Kadl et al. (23)
and gene expression levels related to those of the housekeeping gene amino levulinate synthetase 1 (ALAS) were designed using the formula of Pfaffl (24)
. For comparison of genes that are developmentally regulated in the secretory epithelium, expression was related to keratin 18 (K18), since mutant mice harbor significantly fewer alveoli than controls during lactation. The primer sequences are listed in Table 1
.
|
Immunohistochemistry, confocal laser microscopy, and transmission electron microscopy (TEM)
For histology and detection of cell proliferation and cell death, portions of the abdominal glands were fixed overnight in 4.5% buffered formalin and embedded in paraffin. Sections 4 µm thick were deparaffinized, stained with hematoxylin and eosin (H&E) or subjected to high temperature antigen unmasking using ChemMate target retrieval solution, pH 6.0 (DAKO, Glostrup, Denmark), and incubated with rabbit polyclonal anti-Ki67 (Novocastra, Newcastle, UK, at 1 µg/ml) to detect proliferating cells or with rabbit polyclonal antiactive caspase-3 (at 0.5 µg/ml; R&D Systems) to detect apoptotic cells. The biotinylated secondary antibody was from Vector (at 1:100 dilution; Burlingame, CA, USA) for anti-Ki67, and antibody binding was visualized using the horseradish peroxidase-strept-ABC kit from DAKO. For caspase-3 staining, a fluorescent secondary antibody was used (Alexa-fluor546 goat anti-rabbit, Molecular Probes, Eugene, OR, USA, at 1:500 dilution). Duct numbers and Ki67-positive cells were counted manually at x200 and x400 magnification, respectively. All photomicrographs are representative of at least three mice/group and quantitation was performed on three to five fields of view for each group of three mice. The electron-micrographs are representative of four to nine capillaries derived from one mouse/group, except L3, where two mice were examined.
For blood vessel quantitation using confocal laser microscopy, 50 µm-thick frozen sections were fixed in acetone for 10 min at –20°C and, after washing in PBS, incubated with rabbit polyclonal anti-K5 at 1 µg/ml (BAbCo; Richmond, CA, USA), then subsequently with rat monoclonal anti-CD31 (at 10 µg/ml; PharMingen, San Diego, CA, USA), with appropriate washings in between. Fluorescence-labeled secondary antibodies specific for each primary antibody were from Molecular Probes (Alexa-fluor488 goat ant-rabbit and Alexa-fluor546 goat anti-rat, both diluted 1:500) and were mixed together before application to the tissue. All incubations were at room temperature for 1 h with primary antibodies and 30 min with secondary antibodies. For quantitation of blood vessel density, 50 µm-thick, fluorescently stained mammary gland sections were analyzed with a Zeiss Axiovert 200 M laser scanning microscope using a x40 oil objective and Zeiss LSM 510 software (Carl Zeiss, Inc., Jena, Germany). Each stack was 35 µm thick and comprised 35 slices, each spaced by 1 µm. All 35 images from one stack were projected into one picture and morphometric analyses were performed using the Meta Imaging Series software (Molecular Devices, Ismaning, Germany). Two mice/group and three or four fields of view (x400)/mouse were analyzed.
Tissue samples for TEM were fixed in Karnovsky’s fixative according to standard procedures and embedded in epon resin. Ultrathin sections (0.07 µm) were cut, stained with uranyl acetate and lead citrate, and observed with a JEOUL 1010 transmission electron microscope (Jeol, Peabody, MA, USA) set at 60 kV and photographed with the camera incorporated in the microscope. The lengths of the microvillus processes were measured with the same imaging software as above using one mouse/group (except for control postpartum day 3, hereafter referred to as L3 mice: two mice/group) and four to seven photographs/group.
Detection of mammary gland blood vessel permeability
Pups were removed from their mothers and fluorescein isothiocyanate (FITC) -conjugated albumin from rat (Sigma, 500 µg in 200 µl PBS) was injected into the tail vein (i.v.) of L3 mice (controls and mutants, 4 mice each). Thirty minutes later, mammary glands were harvested, fixed in 4.5% buffered formalin overnight, then embedded in optimal cutting temperature compound (O.C.T., Sakura Finetek, Zoeterwoude, Netherlands). An additional three mutant mice were injected with recombinant murine VEGF (R&D Systems, 2.5 µg/0.1 ml i.v.) 10 min before the FITC-albumin injection. Sections 50 µm thick were examined by fluorescence microscopy using an Olympus AX 70 microscope (Olympus Austria, Vienna, Austria); images for these and all other thin sections were captured with a Spot digital camera (Diagnostic Instruments, Sterling Heights, MI, USA). For a comparison of the fluorescence intensities present in FITC-albumin treated mammary glands, all images were captured using identical exposure times.
Milk collection and detection of milk proteins in mammary gland lysates
At postpartum days 10 and 18 (hereafter referred to as L10 and L18, respectively) pups were removed from their mothers and injected i.p. 1.5 h later with 2 IU oxytocin (Vana, Vienna, Austria, in 0.2 ml). After another 30 min, the mothers were milked by massaging the mammary gland area and milk was collected in capillary pipettes. The milk was stored at –80°C until all samples had been collected, then thawed and spun at high speed for 15 min at 4°C to remove the fat. The defatted milk was used to determine VEGF by ELISA, casein by Western blot, and total protein by silver staining according to the method of Blum et al. (25)
. Three mutant mice at L10 and two control mice at L10 and L18 (the same 2 mice) were milked.
For Western blot, lysates were prepared by sonicating pieces of whole mammary glands in 1% Igepal in PBS (containing proteinase inhibitors) as before. Mammary gland lysate (30 µg) or 10 µl of defatted milk diluted 1:100 was loaded in each lane, the proteins were separated on 8–18% Sepharose gels (Amersham, Uppsala, Sweden), then transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA, USA). After blocking with 5% BSA (Sigma-Aldrich) dissolved in PBS, the membranes were incubated sequentially with rabbit anticasein antiserum (1:3000, a kind gift from Takuya Kanazawa, Ibaraki University, Ibaraki, Japan), horseradish peroxidase-conjugated goat anti-rabbit Ig (1:10000, Pierce), then visualized with the ECL chemiluminescence kit (Amersham). Two mice/group were analyzed.
Statistics
The Mann-Whitney U test was used to compare groups in the qPCR experiments and Student’s 2-tailed t test for all others, with P = 0.05 being defined as the level of significance.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS.RESULTSDISCUSSIONREFERENCES |
|---|
k5-cre/k5-cre mice (hereafter referred to as mutant mice) (18)
, we noticed that mutant females bore very few live young. The pups that were born failed to thrive (Fig. 1
A) and less milk was consistently found in their stomachs (Fig. 1B
). Growth retardation was not due to a defect in the pups themselves, since they were able to gain weight and grow normally after weaning (not shown) or if nursed by control mothers (Fig. 1C, D
), whereas pups born to control mothers but nursed by mutant mothers became runted (Fig. 1D
).
|
Cre-LoxP-mediated recombination of VEGF exon 3 leads to inactivation of VEGF (21)
. Since K5 is expressed in mammary glands, we investigated whether K5-driven, Cre-LoxP-mediated recombination of VEGF exon 3 was occurring in the mammary glands of mutant mice. Figure 2
A shows that in the presence of the Cre transgene and LoxP sequences, recombination, as evidenced by the presence of a band at 517 bp (18)
, can indeed be detected in mammary gland tissue extracts as well as in skin and stomach in which the K5 promoter is also active. In addition, weak bands of the correct size were present in the kidney and heart; the band in the kidney derives from renal pelvis tissue, which expresses high levels of K5 (not shown); the origin of the band in heart tissue is unclear. However, since the renal pelvis represents only a small fraction of the total renal tissue and the heart is not known to express K5, the major band obtained with these primers when kidney and heart genomic DNA is amplified is that of the unrecombined DNA at 2176 bp (18)
. No recombination could be detected in the same tissues in the absence of the Cre transgene. Thus, Cre-mediated recombination at the LoxP sequences flanking exon3 of VEGF is not confined to skin but also occurs in other tissues provided they harbor an active K5 promoter.
|
Lactating mutant mammary glands harbor only low amounts of VEGF and VEGFR2
Since K5 is expressed only in the basally situated myoepithelial cells of the ducts (26)
but VEGF apparently is secreted mainly by the luminal cells during lactation (6)
, we investigated the extent to which K5 promoter-driven inactivation of VEGF was affecting the VEGF pool of the glands as a whole. We analyzed whole-gland lysates for VEGF mRNA by qPCR and whole-gland lysates and milk for VEGF protein by ELISA. Whereas 200 µl of milk could readily be collected after oxytocin injection of control mice at L10 and these mice could be milked a second time at L18, only 50 µl could be collected from mutant mice at L10. Second attempts 8 days later, following oxytocin injection as before, yielded virtually no milk. Virgin mammary glands from control and mutant mice expressed similar levels of VEGF protein and mRNA (Fig. 2B, C
). However, in contrast to control glands, in which VEGF protein levels during lactation reached 4- to 5-fold those of virgin mice, there was no up-regulation at all in mutant glands at these times (Fig. 2B
). To investigate mRNA expression of VEGF exon 3, we designed a primer pair for qPCR so that the forward primer is situated proximal to exon 3 in the intron and the reverse primer is situated within exon 3 (GeneBank accession #NM_900505.2). Thus, if exon 3 has been successfully deleted, no product would be obtained with this primer pair. Indeed, we found that although VEGF mRNA expression increased by 2- to 3-fold in lactating control glands, no increase in native VEGF mRNA could be detected in mammary gland tissue from the lactating mutant mice (Fig. 2C
), suggesting that VEGF exon 3 had been deleted with high efficiency in these mice. No VEGF was detected in milk harvested at L10 from three individual mutant mice, although high levels were present in control milk at both L10 and L18 (Fig. 2D
). Thus, Cre-LoxP-mediated recombination of VEGF in mammary epithelial cells (MECs) results in reduced milk production, lower levels of VEGF mRNA and protein during pregnancy, failure to up-regulate both during lactation, and an absence of detectable VEGF in the milk.
Lobulo-alveolar expansion into the fat pad during lactation is incomplete in mutant females
We next investigated whether mammary gland development during pregnancy and lactation had been affected by the lack of VEGF. Histological examination of H&E-stained sections of mammary glands revealed that lobulo-alveolar structures filled the fat pad completely by L3 in glands from control mice and that the luminal epithelial cells contained only small lipid droplets (Fig. 3
A, B). In contrast, mutant glands retained large areas of adipose tissue even up until postpartum day 7 (hereafter referred to as L7), and luminal epithelial cells contained conspicuously large lipid droplets. Quantitation of alveoli confirmed that significantly fewer were present in the mutant glands during both lactation time points (Fig. 3C
). However, no significant differences in the numbers of either Ki67 or active caspase-3 labeled alveolar cells were noted at any stage examined (postcoitum day 12, hereafter referred to as P12; L3 and L7) (not shown), indicating there was neither a defect in the proliferative capacity of the epithelial cells nor were the glands undergoing premature involution.
|
We did not accurately quantitate the amount of milk synthesized by the lactating mutant females, although it was obvious that less milk could be collected from them (see above). Instead, we investigated milk composition by analyzing whole mammary gland lysates and defatted milk for casein content by Western blot and major protein composition by silver staining. These analyses showed there was normal temporal up-regulation of casein in the mutant mammary glands starting during pregnancy and that neither the casein content nor the whole protein content of milk differed from that of control samples (Supplemental Fig. 1). Thus, milk protein synthesis does not seem to be affected in the mutant mice.
Transcription of differentiation-associated genes is reduced in mutant mammary glands
The apparent defect in milk secretion in the mutant glands prompted an investigation into the regulation of mammary gland differentiation-associated genes. We chose to examine connexin 26 (Cx26) and connexin 32 (Cx32), intercellular gap junction proteins implicated in cell-cell communication that are highly up-regulated in pregnant or lactating mammary gland epithelia (27
28
29
30)
, and butyrophilin and xanthine oxidoreductase (XOR), essential for milk fat globule secretion (31
, 32)
. Preliminary analysis of gene expression by quantitative PCR revealed that these four genes, but not K18 or perilipin, a marker for mature adipocytes (33
, 34)
, were up-regulated in lactating control but not mutant glands (supplemental Fig. 2 shows one of two series of mRNA isolates, yielding similar results).
For statistical analysis, a third series of RNA isolates was prepared from five to seven control and mutant L3 mice (Fig. 4
A, B). Figure 4A
shows mRNA expression of angiogenesis-related genes calculated relative to the housekeeping gene ALAS. Genes that are regulated in the secretory epithelium were quantified relative to K18 expression levels (Fig. 4B
), since the mutant mammary glands had significantly fewer lobules. (Note the logarithmic scale on the y axis.) Each dot represents the results for an mRNA isolate from an individual mouse relative to its own level of ALAS or K18. For each gene, levels of expression determined in the same control mouse were arbitrarily set to 1.0, and all other values were normalized to this. Confirming results of the preliminary experiments described above, expression of Cx26, Cx32, XOR (Fig. 4B
) as well as of VEGFR2, CD31, and K5 (Fig. 4A
) was significantly lower in mutant glands (P<0.05), whereas butyrophilin levels tended to be lower but did not reach statistical significance. In contrast, neither PlGF, a proangiogenic member of the VEGF family (35)
(Fig. 4A
), nor VEGFR1 (not shown) levels were lower, and K18 levels varied widely. Thus, as well as incomplete lobulo-alveologenesis and milk lipid stasis, inactivation of VEGF in mouse mammary glands results specifically in inadequate regulation of genes associated with differentiation and angiogenesis.
|
Angiogenesis and blood vessel function is impaired in mutant glands during lactation
Lower levels of VEGF during pregnancy and the lack of up-regulation during lactation could result in inadequate angiogenesis, with concomitant ineffective delivery of the nutrients and hormones that are necessary for lobulo-alveolar expansion and differentiation. Blood vessels and their location with respect to the alveoli were therefore identified by CD31/K5 double immunofluorescence at P12 (P0 being defined as the day the vaginal plug was found) and at L3 and L7 (L0 being defined as the day following parturition, which normally occurs during the night). Confirming earlier reports (7
, 9)
, blood vessels in the control glands formed basket-like structures around the alveoli during pregnancy (Fig. 5
A). In contrast, blood vessels in the mutant glands were less well organized and not intimately associated with the alveoli. During lactation there were fewer blood vessels in the mutant glands; they appeared thinner, were less convoluted and branched, and seemed to meander randomly throughout the tissue instead of lying in close proximity to the alveoli as they did in the control glands. Computer-assisted quantitation of CD31-stained sections revealed there was no difference in blood vessel density between control and mutant mice (not shown) before or during pregnancy (Fig. 5B
). However, during lactation, vessel density, as judged by pixel number, was significantly less in mutant mice, falling to about half that of controls by L7 (Fig. 5B
). qPCR analysis of CD31 gene expression confirmed the quantitative immunofluorescence data indicating that levels were similar in virgin control and mutant mice, but suggested that there was a 2-fold up-regulation in mutant glands compared with the 4-fold increase in control glands. By lactation, expression had returned to virgin levels in mutant glands, but remained high in controls (Fig. 5C
). Because of the low numbers of mice in this experiment, no statistical analysis is possible (but see Fig. 4A
, where the difference at L3 is highly significant).
|
In addition to the lower blood vessel density, vessels present in the lactating mutant mammary glands differed ultrastructurally from control vessels. During pregnancy and lactation, mammary gland blood vessel endothelial cells increase the numbers of cytoplasmic pinocytotic vesicles and develop microvilli and marginal folds on their luminal surfaces (9
, 10)
. These features, which have been shown to function in transendothelial transport of macromolecules (36)
and are thought to aid in fluid uptake (9)
or regulation of blood velocity (10)
, reach a maximum at midlactation. We did not attempt to quantitate pinocytotic vesicles in this study, but the structure of the control capillary at L3 shown in Fig. 6
A is representative of 18/26 of such capillaries from two mice from this time point. The capillaries displayed an extremely thickened, heterogeneous cytoplasm with numerous pinocytotic vesicles and were decorated on the interior surface with convoluted microvillus processes. In contrast, 0/14 of mutant L3 capillaries had this morphology. Although their cytoplasm was now thinner, at L7 the control vessels still displayed long microvilli whereas those in mutant vessels were much shorter, similar to those found in virgin vessels of both mouse lines. Measurement of the lengths of the microvilli confirmed that, although not more numerous at either time point, they were significantly longer in control glands at L7 (Fig. 6B
). These observations raise the possibility that VEGF induces the ultrastructural changes observed on mammary gland blood vessel endothelial cells during lactation; although to the authors’ knowledge they have not been described in these cells, such an effect for VEGF has been reported for other systems (37
, 38)
.
|
Since VEGF is a major permeability factor for blood vessel endothelial cells, its up-regulation during lactation could facilitate solute and fluid transfer from the blood to the MECs. Indeed, blood vessels from lactating mice have been shown to allow the passage of macromoleclules whereas capillaries in virgin glands did not (11)
. We thus asked whether the mutant blood vessels were compromised in their ability to permit transfer of solutes. First we examined the electron micrographs for evidence of endothelial cell fenestrations, which are induced by VEGF (39
, 40)
. Fenestrations were much more frequent in blood vessels from virgin vs. lactating mice (6/11 total virgin vessels compared with 7/56 total lactating vessels, controls and mutants combined), suggesting that these structures do not play an important role in substance transfer during lactation.
We next investigated the permeability of blood vessels to albumin, an important carrier molecule of substances such as phospholipids, fatty acids, and steroid hormones. Figure 7
A shows that FITC-coupled albumin injected into the tail vein of lactating mice 30 min before sacrifice extravasated efficiently in controls and could be detected as a fluorescent accumulation in the alveolar epithelial cells. In contrast, little fluorescence could be detected in alveoli of mutant mice (Fig. 7B
), but this lack of permeability was restored by injection of VEGF 10 min before the fluorescent tracer (Fig. 7C
). The failed up-regulation of VEGF during lactation thus results in impaired lobulo-alveologenesis and epithelial differentiation, insufficient angiogenesis, improper spatial localization of the blood vessels, reduced endothelial cell activation, and extremely reduced vessel permeability to albumin.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS.RESULTSDISCUSSIONREFERENCES |
|---|
In the postnatal mammary gland, K5 expression is confined to the myoepithelial cells (26)
. By contrast, current data suggest it is the luminal epithelial cells that lack K5 and synthesize and secrete VEGF (6
, 15
, 17)
. Our finding that VEGF production is abrogated in the mutant glands as a whole therefore is compatible with the hypothesis that luminal and myoepithelial cells arise from a common (K5-positive) precursor (1
, 41)
. In addition, the complete lack of VEGF up-regulation in whole-gland lysates from mutant mice during pregnancy and lactation supports the findings by Hovey and co-workers (6)
that it is the epithelial component of the mammary gland tissue, and not stromal cells, that is the major source of VEGF at these times.
Our detection of large quantities of VEGF in the milk of control mice agrees with findings from earlier reports on VEGF content of human milk (15
16
17
, 42)
. However, whereas the concentration of VEGF in human milk declines rapidly after birth, we detected higher levels toward the end of lactation (L18) than at midlactation (L10). The function of VEGF in milk is currently unknown. VEGF receptors have been detected by immunohistochemistry in gut epithelium of newborns (17)
, and it has been suggested that VEGF in the milk may function to facilitate nutrient uptake in the gut (42)
. If this is the case, then this could be an additional factor contributing to the runting of pups suckled by mutant mothers, in whose milk we were unable to detect any VEGF at all.
The lack of up-regulation of VEGF in lactating mutant glands was associated with inadequate mammary gland lobulo-alveologenesis and apparent defects in MEC differentiation, since the differentiation-associated genes we had chosen to examine were expressed at lower levels during lactation in the mutant mice. The direct consequence of the reduced connexin up-regulation in our mutant mice is unclear at present. Connexins function in intercellular communication (43)
, and both Cx26 and Cx32 are highly up-regulated in MEC during pregnancy and lactation, respectively. However, inactivation of either of these connexins individually had no affect on the final stages of mammary gland differentiation in the mouse (44)
, and our mutant mice displayed a milk protein composition indistinguishable from that of controls. In contrast, the inability of our mutant dams to adequately nourish pups, together with the accumulation of lipid droplets observed in luminal epithelial cells of such mice, might be a direct result of the poor up-regulation during lactation of XOR and butyrophilin, whose activity is essential for milk lipid droplet secretion (31
, 32)
. Indeed, mice in which these genes have been inactivated also accumulate lipid droplets in lactating luminal epithelial cells, and their pups are runted (ibid). Thus, reduced availability of VEGF during lactation is associated with lower levels of expression of differentiation-associated genes and lipid stasis in MEC.
The mammary gland defects described here closely parallel those reported for mice in which hypoxia-inducible factor 1
(HIF1) was inactivated in mammary gland epithelium (45)
. In both this and our model, alveologenesis and angiogenesis progressed normally during pregnancy in the mutant mice, but lobulo-alveolar development stagnated and the expression of terminal mammary differentiation markers was impaired during lactation, resulting in reduced volumes of secreted milk and runted pups. Seagroves et al. did not find a reduction of VEGF mRNA in mutant mammary glands during pregnancy, however, and suggest that HIF1 acts independently of VEGF. Unfortunately, these authors examined neither VEGF levels nor blood vessel densities during lactation, when VEGF production is highest. The possibility therefore remains that VEGF fails to be up-regulated in their model at this time. Indeed, we show here that inactivation of VEGF, a downstream target of HIF1, is sufficient to generate defects similar to those caused by inactivation of HIF1 itself.
Angiogenesis (as judged by quantitation of CD31 gene and protein expression) in the pregnant and lactating VEGF-deficient mammary glands is reduced compared with control animals. This suggests that VEGF, as well as being secreted into the milk, normally must also be secreted into blood circulation, as proposed by Nishimura and co-workers (15)
. During the latter part of pregnancy and during lactation, the vascular plexus extends by remodeling rather than by capillary sprouting (7)
, and our finding that CD31 expression did not increase from pregnancy to lactation in the control mice is in agreement with this. Instead, an increase in the number of microvilli, pinocytotic vesicles, and mitochondria in the capillary endothelial cells (9
, 10)
strongly suggests that it is the change in blood vessel properties, rather than number, that is necessary for successful mammary gland function during lactation. This notion is supported in our study by several observations. First, the spatial organization of the vasculature in the mammary glands differs between mutant and control mice in that, besides being less abundant, the capillaries often do not lie in close juxtaposition to the basal cells of the alveoli. Second, as assessed by TEM, the endothelial cells of capillaries in the lactating mammary glands of mutant mice show far fewer signs of activation, such as luminal protrusions and pinocytotic vesicles, than those from controls. The lack of these activation-associated features might hamper efficient fluid and solute transfer from blood vessels to the mammary gland (9
, 10)
. Note that the paucity of fenestrations in lactating capillaries of both mutant and control mice suggests that these structures do not play an important role in fluid and solute transfer. Finally, the impressive reduction of extravasation of FITC-labeled albumin in mutant mammary glands, which could be restored by injection of VEGF, clearly demonstrates that the blood-mammary gland barrier is considerably less permeable in these mice. Based on these observations, we propose that VEGF derived from mammary epithelium during lactation plays a crucial role in regulating the permeability of mammary gland blood vessels, thereby contributing to the trafficking of fluid and macromolecules and, ultimately, to the production of sufficient amounts of milk.
In summary, our data indicate that inactivation of VEGF in mammary gland epithelia results in inadequate lobulo-alveologenesis and epithelial differentiation during lactation. We suggest that this is partly due to a reduced blood vessel density, but also (and perhaps more important) to a functional inadequacy of the blood vessels, which leads to ineffective delivery of fluid, hormones, and other substances, ultimately resulting in decreased milk secretion and malnourished pups. VEGF thus must be added to the long list of factors that control the proper function of the fully differentiated mammary gland.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication April 11, 2007. Accepted for publication June 7, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS.RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. Bobrovnikova-Marjon, G. Hatzivassiliou, C. Grigoriadou, M. Romero, D. R. Cavener, C. B. Thompson, and J. A. Diehl PERK-dependent regulation of lipogenesis during mouse mammary gland development and adipocyte differentiation PNAS, October 21, 2008; 105(42): 16314 - 16319. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. M. Elias, J. Arbiser, B. E. Brown, H. Rossiter, M.-Q. Man, F. Cerimele, D. Crumrine, R. Gunathilake, E. H. Choi, Y. Uchida, et al. Epidermal Vascular Endothelial Growth Factor Production Is Required for Permeability Barrier Homeostasis, Dermal Angiogenesis, and the Development of Epidermal Hyperplasia: Implications for the Pathogenesis of Psoriasis Am. J. Pathol., September 1, 2008; 173(3): 689 - 699. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
P < 0.005. D) VEGF protein could not be detected in milk of mutant mice by ELISA (#: each symbol indicates an individual mouse; L10, L18: postpartum days 10, 18).
