Histochemistry and Cell Biology

, Volume 130, Issue 3, pp 567–581

Cell-type-specific expression of murine multifunctional galectin-3 and its association with follicular atresia/luteolysis in contrast to pro-apoptotic galectins-1 and -7

Authors

  • Michaela Lohr
    • Institute of Physiological Chemistry, Faculty of Veterinary MedicineLudwig Maximilians University
    • Institute of Physiological Chemistry, Faculty of Veterinary MedicineLudwig Maximilians University
  • Martin Lensch
    • Institute of Physiological Chemistry, Faculty of Veterinary MedicineLudwig Maximilians University
  • Sabine André
    • Institute of Physiological Chemistry, Faculty of Veterinary MedicineLudwig Maximilians University
  • Fred Sinowatz
    • Institute of Veterinary Anatomy, Faculty of Veterinary MedicineLudwig Maximilians University
  • Hans-Joachim Gabius
    • Institute of Physiological Chemistry, Faculty of Veterinary MedicineLudwig Maximilians University
Original Paper

DOI: 10.1007/s00418-008-0465-0

Cite this article as:
Lohr, M., Kaltner, H., Lensch, M. et al. Histochem Cell Biol (2008) 130: 567. doi:10.1007/s00418-008-0465-0

Abstract

Galectin-3 is a multifunctional protein with modular design. A distinct expression profile was determined in various murine organs when set into relation to homodimeric galectins-1 and -7. Fittingly, the signature of putative transcription-factor-binding sites in the promoter region of the galectin-3 gene affords a toolbox for a complex combinatorial regulation, distinct from the respective sequence stretches in galectins-1 and -7. A striking example for cell-type specificity was the ovary, where these two lectins were confined to the surface epithelium. Immunohistochemically, galectin-3 was found in macrophages of the cortical interstitium between developing follicles and medullary interstitium, matching the distribution of the F4/80 antigen. With respect to atresia and luteolysis strong signals in granulosa cells of atretic preantral but not antral follicles and increasing positivity in corpora lutea upon regression coincided with DNA fragmentation. Labeled galectin-3 revealed lactose-inhibitable binding to granulosa cells. Also, slender processes of vital granulosa cells which extended into the zona pellucida were positive. This study demonstrates cell-type specificity and cycle-associated regulation for galectin-3 with increased presence in atretic preantral follicles and in late stages of luteolysis.

Keywords

AtresiaCorpus luteumGranulosa cellLectinOvary

Introduction

The emerging insights into rather frequent occurrence of multifunctionality in diverse proteins call for analysis of the parameters governing activity regulation and switching between options (Jeffery 2003). To address this issue the determination of the cell type(s) harboring the protein under study and its cellular localization as well as of ligand availability are features amenable to histochemical monitoring. Our study deals with a protein, which is involved in turning sugar-encoded information into cellular responses (Gabius et al. 2004). Guided by the interest to elucidate the biological roles of glycan epitopes of cellular glycoconjugates, especially when strategically positioned at branch ends as readily accessible biochemical signals, increasing attention is being directed to the endogenous lectins (Gabius 1997, 2008; Solís et al. 2001). In this context, a member of the family of adhesion/growth-regulatory galectins, i.e., galectin-3, prominently meets the criteria for multifunctionality (Smetana et al. 2006; Arnoys and Wang 2007).

This lectin has a modular design, setting it apart from the other galectins. Its structure is composed of the C-terminal lectin site, a collagenase/matrix metalloproteinase-sensitive stalk region relevant for oligomerization and a short N-terminal section with the two sites for serine phosphorylation. The specific binding of distinct glycan motifs to the lectin domain and of various peptides to sites covering all three mentioned modules underlies interaction with a particular set of glycoconjugates and proteins. Combined with the possibility to reside in the extracellular matrix, cytoplasm and nucleus this galectin can therefore be engaged in a series of contacts, resulting in a pleiotropic bioactivity profile (Gabius 2006; Villalobo et al. 2006). It ranges from modulation of cell adhesion, growth and migration by binding to cell surface glycans, for example on β1-integrin and laminin, and anti-apoptotic effects by associating with components of the cytoplasmic apoptosis network such as Bcl-2 and Alix/AIP-1 to involvement in pre-mRNA splicing and regulation of gene expression via activation of certain transcription factors (e.g., thyroid transcription factor-1, nuclear factor of activated T-cell, cyclic AMP-responsive element binding protein as well as AP1 and Sp1) (Liu et al. 2002; Villalobo et al. 2006; Nakahara and Raz 2007). Oligomerization of the lectin in the presence of suitable glycan ligands, establishing cross-linked complexes suited for signaling, and its impairment by proteolytic truncation in situ constitutes a further level of activity regulation (André et al. 2003; Ahmad et al. 2004). Having herewith illustrated galectin-3’s versatility in biochemical and cell biological terms, the interest to define the characteristics of cellular expression in vivo is obvious.

We thus systematically monitored murine organs by immunohistochemistry. The resulting profile could then be set into relation to data on homodimeric galectins-1 and -7 obtained under identical conditions previously (Lohr et al. 2007). Because a substantial portion of evidence for a role of galectin-3 as anti-apoptotic effector originates from work with in vitro models, using ectopic expression and/or external elicitors of stress, there is good reason to flank this work by processing tissues with ongoing physiological cell death. With this particular aim in view, the organ profiling presented in the first part of this study guided our investigation to the analysis of follicular atresia and luteal regression in the ovary. These events embody two cases of apoptosis within natural tissue dynamics in vivo (Davis and Rueda 2002; Matsuda-Minehata et al. 2006). Hormonal synchronization and measurements of markers characteristic of progesterone synthesis and of stages during progression of the cell death program were teamed up with galectin-3 localization. Regarding this parameter, we deliberately paid attention to the possibility of proteolytic truncation in situ. In detail, the following questions will be resolved by the obtained results:
  1. 1.

    Will galectin-3’s presence in adult murine organs show a distinct pattern?

     
  2. 2.

    Will this pattern be disparate from those of the homodimeric galectins-1 and -7, which are pro-apoptotic for various cell types (Villalobo et al. 2006), when using non-cross-reactive antibodies under identical conditions?

     
  3. 3.

    Will the proximal promoter region of the galectin-3 gene reveal a simple or complex array of putative transcription-factor-binding sites and exhibit differences to respective regions of the genes coding for galectins-1 and -7?

     
  4. 4.

    Will galectin-3 presence in ovary be regulated in the estrous cycle and tied to certain stages within the apoptotic process identified by visualizing cleaved caspase-3 and DNA fragmentation?

     
  5. 5.

    Will labeled galectin-3 be suitable for visualizing accessible binding sites on the light and electron microscopical levels?

     

Materials and methods

Animals

Wild-type (WT) C57BL/6 mice were purchased from Charles River Laboratories (Sulzfeld, Germany), the galectin-3-deficient (gal-3−/−, KO) C57BL/6 mice were kindly provided by F.-T. Liu (Department of Dermatology, University of California Davis, Sacramento, USA; now available via the Functional Glycomics Consortium). WT and gal-3−/− mice were backcrossed to ensure identical genetic background, and selection of homozygous WT and gal-3−/− mice was based on genomic PCR assays. Ovulation of females was synchronized and started at day 28 post partum prior to the first ovarian cycle by hormonal induction. Mature females were treated at the age of 6 months. To stimulate follicle growth 8 IU of pregnant mare’s serum gonadotropin (PMSG; Intergonan®, Intervet, Unterschleissheim, Germany) were injected intraperitoneally (i.p., day 1) followed by 8 IU of human choriongonadotropin (HCG; Ovogest® 1500, Intervet) 40 h thereafter to induce ovulation and growth of corpora lutea. Mice were killed by cervical dislocation following ether anesthesia, a panel of organs from both genders including ovaries was immediately removed and either snap frozen in liquid nitrogen or fixed in Bouin’s solution for immunohistochemistry. The total number of hormonally treated females entering immunohistochemical analysis was 34 WT and 33 KO, to which 20 mature WT and 20 mature KO mice (2–12 months old) were added. All animal procedures were approved by the local government authorities.

Galectin purification and labeling

Murine galectin-3 was purified after recombinant production by affinity chromatography on lactosylated Sepharose 4B, obtained by ligand conjugation to divinyl-sulfone-activated resin, as crucial step, controlled for purity by one- and two-dimensional gel electrophoresis, gel filtration and mass spectrometry, also exposed to collagenase for proteolytic truncation and labeled under activity-preserving conditions using the N-hydroxysuccinimide ester derivative of biotin (Sigma-Aldrich, Munich, Germany) with conjugation of up to seven units determined by shifts in isoelectric point and by mass spectrometry (André et al. 2004; Kübler et al. 2008). Activity controls of labeled lectin were performed by solid-phase and cell-binding assays including inhibition with haptenic sugar (André et al. 2006).

Antibodies

Polyclonal antibodies against galectins-1, -3 and -7 were raised in rabbits by repeated booster injections, the titer was monitored regularly by ELISA and serum fractionation to obtain the immunoglobulin G (IgG) fraction was carried out on protein-A Sepharose Fast Flow resin (GE Health Care, Munich, Germany) (Kaltner et al. 2002). These preparations were routinely subjected to rigorous specificity controls by ELISA and Western blot using galectins-1, -2, -4, -7, -8 and -9. Slight cross-reactivity to galectin-7 was completely eliminated using bead-immobilized lectin (12 mg/ml) as ligand (Lensch et al. 2006). Anti-Mac-2 monoclonal antibody (isotype: rat IgG2a) isolated from the hybridoma cell line TIB-166 (ATCC, Rockville, USA) was a gift from N.V. Bovin (Shemyakin and Ovchinnikov Institute of Bioorganic Chemistry, Russian Academy of Sciences, Moscow, Russia). Rat monoclonal antibodies against the murine macrophage marker F4/80 (clone CI:A3-1, rat anti-mouse F4/80, isotype IgG2b) was purchased from Serotec (Düsseldorf, Germany). The polyclonal rabbit antibody against cleaved caspase-3 (Asp175) and its blocking peptide were from Cell Signaling (New England Biolabs, Frankfurt, Germany). The antibody against 3β-hydroxysteroid dehydrogenase/isomerase (3β-HSD), prepared as described (Ullmann et al. 2003), was kindly provided by the author (Dr. S. Ullmann, University of Glasgow, Division of Environmental and Evolutionary Biology). The affinity-purified rabbit anti-actin antibody from Sigma-Aldrich was used in Western blots.

Northern/Western blots and RT-PCR analysis

Digoxigenin-labeled sequence portions of galectin-3 and actin cDNAs were obtained after subcloning of 120 bp-long fragments by incorporation of digoxigenin-11-UTP (DIG-11-UTP; Roche, Mannheim, Germany) with the Ampliscribe T7 in vitro transcription kit (Epicentre, Madison, USA). Briefly, a 20-μl reaction volume contained the following components (final concentrations): 50 ng/μl pGEM®-4Z/galectin-3 cDNA fragment or pGEM®-4Z/actin cDNA fragment, the three ribonucleotides ATP, CTP and GTP (at 7.5 mM), 6 mM UTP, 2 mM DIG-11-UTP, 100 mM dithiothreitol and 2 μl RNA polymerase-containing solution (20 U) in commercial Ampliscribe reaction buffer. Controls for length of the riboprobes and labeling efficiency ascertained their applicability for mRNA detection using total RNA from mice ovaries isolated with the RNAeasy kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol. Four microgram of total RNA from each sample was electrophoretically fractionated on a 1% (w/v) agarose gel containing 0.67 M formaldehyde, then transferred to a positively charged nylon membrane (Roche) using capillary blotting and linked to the surface of the membrane by exposure to UV light (0.12 J/cm2; Bio-Link Crosslinker, Vilber Lourmat, Eberhardzell, Germany). Following the pre-hybridization step incubation with the riboprobe (90 ng/ml) for 16 h at 68°C and thorough washing steps led to visualization of specific signal with alkaline-phosphatase-conjugated anti-digoxigenin Fab fragments (0.1 μg/ml in 1% blocking solution) and the CDP (1,2-dioxetane)-Star reagent (Roche) as well as CL-XPosure™ film development (Perbio Science, Bonn, Germany). Total RNA also served as starting material for RT-PCR analysis with galectin-type-specific primer sets, as described previously (Lohr et al. 2007). For galectin detection tissue samples were homogenized in extraction buffer (20 mM phosphate-buffered saline (PBS), pH 7.2, containing 1% Triton X-100, 0.1% sodium deoxycholate, 50 mM lactose, 2 mM dithiothreitol, 2 mM ethylenediamine tetraacetic acid (EDTA) and the protease inhibitors pefabloc® at 1 mM, aprotinin at 2 μg/ml and leupeptin at 5 μg/ml). Western blots followed an optimized protocol (Lohr et al. 2007), first separating a total of 15 or 50 μg extract protein on a discontinuous SDS polyacrylamide gel (4% stacking gel and 14.5% running gel), subsequently electrophoretically transferring proteins to a nitrocellulose membrane (0.2 μm; Protran, Schleicher & Schuell, Germany). Residual binding sites on the membrane were then blocked (5% dry milk and 0.5% Tween 20 in Tris-buffered saline, pH 7.5), and membranes were routinely incubated overnight at 4°C with an anti-galectin-3 IgG-containing solution at a concentration of 0.5 μg/ml (for anti-actin: 5 μg/ml). After three washing steps with Tris-buffered saline, pH 7.5, containing 0.5% Tween 20, the membranes were then placed in a solution containing the horseradish-peroxidase-labeled goat anti-rabbit IgG (0.5 μg/ml; Sigma-Aldrich) for 1 h at room temperature. Finally, membranes were placed in a darkroom exposed to a mixture of 2 ml 0.1 M Tris-HCl, pH 8.6, with 1.25 mM luminol sodium salt, 0.2 ml of a solution of 6.7 mMp-coumaric acid in dimethylsulfoxide and 0.6 μl H2O2 (30% v/v) for 2 min at room temperature. Exposition periods with CL-XPosure™ X-ray film for signal visualization and subsequent film development were set to yield optimal signal intensity and minimal background.

Promoter analysis in silico

The database of the National Center for Biotechnology Information (NCBI, Bethesda, USA; http://www.ncbi.nlm.nih.gov/Genbank/index.html) served as a source for the sequence information and the putative location of the start site for transcription. The sequence section from 2,000 bp upstream to 150 bp downstream of that site was subjected to promoter analysis to spot putative recognition sites for transcription factors using the latest update of the TRANSFAC® data base for transcription factors, two independent search algorithms and scoring and stringency criteria as given in detail previously when presenting initial analysis for galectins-1 and -7 (Lohr et al. 2007).

Visualization of apoptotic DNA fragmentation in situ

Deparaffinized sections were washed in 10 mM PBS and exposed to microwaves at 350 W in 1 mM EDTA-solution, pH 8.0, for 5 min. Following two further rinses with PBS the sections were covered with a mixture of solutions of 5 μl containing the TUNEL enzyme (50 U/ml calf thymus terminal deoxynucleotidyl transferase; Roche) and 45 μl TUNEL Label solution (200 mM potassium cacodylate, 25 mM Tris-HCl, 1 mM CoCl2, 0.25 mg/ml bovine serum albumin, pH 6.6) and incubated for 60 min at 37°C in a humidified chamber in the dark. Signal development with anti-fluorescein Fab fragments conjugated to alkaline phosphatase (Roche) was carried out for 30 min at 37°C. In parallel, sections were processed for positive controls by including an initial incubation step for 10 min with DNase I (Roche) and for negative controls by omitting the reaction step with terminal deoxynucleotidyl transferase. Color development, counterstaining and mounting were performed as described (Lohr et al. 2007).

Immuno- and galectin histochemistry

Ovaries and other organs from WT and KO mice were immediately fixed, the immersion in Bouin’s solution lasting 24 h at 4°C, then dehydrated by using an ethanol series and embedded in paraffin and routinely processed as described (Lohr et al. 2007). Antigen retrieval was required for detection of the F4/80 antigen (0.01% (w/v) trypsin in a 0.1% (w/v) solution of CaCl2, pH 7.8; incubation for 10 min at 37°C) and also of cleaved caspase-3 (microwaving using a solution containing 1 mM EDTA, pH 8.0; first at 375 W until the solution started boiling, then for 10 min at 112 W followed by cooling down for 30 min). Incubation with a solution of 1% H2O2 in distilled water to quench endogenous peroxidase activity was included when peroxidase-labeled markers were used. Washing with 10 mM PBS (pH 7.2–7.4) and treatment with 1% (w/v) solution of bovine serum albumin (BSA; Sigma-Aldrich) in PBS containing 5% (v/v) normal swine serum (DakoCytomation GmbH, Hamburg, Germany) or normal goat serum (Axxora, Grünberg, Germany) followed to saturate non-specific protein-binding sites, hereby contributing to exclude antigen-independent staining. Subsequently, sections were incubated in a humid chamber overnight at 4°C with a solution containing one of the IgG fractions raised against murine galectin-3 (0.5 μg/ml), 3-β-HSD (10 μg/ml), F4/80 (20 μg/ml), cleaved caspase-3 antibody (5 μg/ml), the anti-Mac-2 monoclonal antibody (2 μg/ml) or biotinylated galectin-3 (20 μg/ml), respectively, dissolved in PBS with 1% BSA. Following an incubation period of 16 h the sections were rinsed in PBS and then incubated at room temperature with a solution containing the biotinylated secondary swine anti-rabbit F(ab′)2 fragments (1.25 μg/ml; DakoCytomation GmbH) or the alkaline-phosphatase-conjugated secondary goat anti-rabbit IgG (0.5 μg/ml; Sigma-Aldrich). Regarding the biotinylated secondary antibody, the sections were thoroughly rinsed after 1 h in PBS again and then incubated at room temperature with Vectastain® Elite ABC Kit (standard) PK-6100 reagents (Axxora), in the case of biotinylated galectin-3 with Vectastain® ABC-Kit alkaline phosphatase standard AK-5000 reagents (Axxora), respectively. Localization profiles of the antigens, when applying the marker conjugate of the Vectastain® Elite ABC Kit, were made visible using the Vector® DAB Substrate Kit SK-4100 reagents (Axxora) under visual control with a maximum incubation period of 10 min. As to alkaline-phosphatase-conjugated secondary antibody and the Vectastain® ABC-Kit alkaline phosphatase standard, the sections were first thoroughly rinsed in PBS and then in 100 mM Tris-buffered saline (pH 8.2–8.5). In this case, the distribution of the marker conjugate was visualized using the Vector® Red Alkaline Phosphatase Substrate Kit I SK-5100 reagents (Axxora) applied to the sections for 20–30 min in the dark. The sections were counterstained with Mayer’s haemalaun. Following dehydration sections were finally mounted in Eukitt® (Kindler, Freiburg, Germany). Controls included omission of first-step reagent as well as applications of pre-immune serum, an IgG preparation depleted of anti-galectin-3 reactivity and the peptide blocking anti-cleaved caspase-3 IgG binding (New England Biolabs), respectively. Biotinylated galectin-3 (20 μg/ml; truncated and full-length proteins) was used in light and electron microscopy, the latter following the pre-embedding staining protocol developed for chicken galectins (Stierstorfer et al. 2000). Carbohydrate-dependent binding was blocked by the presence of 75 mM lactose in control sections. Photographs were taken with an Olympus microscope (type BH-2) equipped with an Olympus SC-35 camera (Olympus, Hamburg, Germany) and also a Zeiss microscope AxioImager.M1 (Carl Zeiss MicroImaging GmbH, Göttingen, Germany) equipped with an AxioCam MRc3 and MRc digital camera and the software Axiovision 4.6. Photo documentation of fluorescence emission was recorded after applying the Vector® Red Alkaline Phosphatase Substrate Kit I reagents (Axxora) using the 560/630 nm filter combination.

Results

Specificity controls

Because the quality of the antibody preparation is a critical factor for the study, we started with rigorous controls. First, systematic assays were run to spot any cross-reactivity with other galectins. Indeed, slight positivity was detected in the case of galectin-7. This result prompted removal of the cross-reactive fraction by affinity chromatography. The next set of assays by Western blots ascertained (a) absence of any side reaction to other proteins in extracts from organs of WT mice, (b) lack of signal with material from mice deficient in galectin-3 expression and (c) antibody reactivity to full-length and proteolytically truncated galectin-3 (Fig. 1a). This reactivity with the lectin domain afforded the opportunity to track down in situ truncation in blots. In principle, this type of analysis was capable to bring a substantial truncation of galectin-3 by proteolysis in situ to light, as exemplarily shown in extracts of jejunum (Fig. 1a). That antigen presence was essential for staining was also proven in sections by comparison of specimen from WT and KO mice (Fig. 1b). Thus, the antibody produced in our laboratory was suitable for systematically mapping galectin-3 expression. Due to the availability of antibody preparations of similar quality for other galectins in our laboratory it was possible, too, to compare characteristics of expression of galectin-3 with those of other members of this lectin family determined under identical conditions. A proof-of-principle example for cell-type specificity in the galectin network is given in Fig. 1b, c.
https://static-content.springer.com/image/art%3A10.1007%2Fs00418-008-0465-0/MediaObjects/418_2008_465_Fig1_HTML.gif
Fig. 1

a Western blots for galectin-3 (gal-3) detection. Chemiluminescence monitoring revealed a weak signal in extracts of jejunum and strong expression for example in ovary, spleen and epididymal extracts (15 μg protein per lane). Evidence for proteolytic truncation in situ was seen in the case of jejunum. Negative control: epididymal extract of a KO mouse (15 μg); positive controls: full-length galectin-3 (10 ng) and truncated galectin-3 (500 ng). Immunohistochemical localization of galectins-3 (b) and -1 (c) in murine epididymis of WT and KO mice, nuclei stained with haemalaun. Presence of gal-3 was seen in epididymal epithelium and in immune cells in interstitium of cauda epididymidis (arrows, b), while sections from gal-3−/− mice were consistently free of staining, ascertaining complete absence of antigen-independent staining (inset, b1). In comparison, galectin-1 was localized in smooth muscle layer and interstitium of this region (arrowheads, c). Bar length 20 μm (b, b1, c)

Fingerprinting of galectin-3 expression

The screening for gene expression by RT-PCR at a higher level of sensitivity revealed galectin-3-specific mRNA to be invariably present (Table 1). Immunohistochemical monitoring followed to map protein presence and distribution. Initial tests with different fixatives had shown Bouin’s solution to yield optimal results when localizing proto-type galectins (Lohr et al. 2007). The respective survey of galectin-3 staining documents cell-type specificity both within an organ and in comparison to two other galectins (Table 1). Epithelial cells and macrophages, also organ-specific resident cells of the latter type, were strongly stained. Setting these properties in relation to two homodimeric galectins, intrafamily differences turned up (Table 1). Equally important, this expression profile intimated a complex pattern of gene regulation, directing interest to inspect the proximal promoter region of the gene for murine galectin-3 for putative transcription-factor-binding sites.
Table 1

Summary of semiquantitative detection of galectins-3, -1 and -7 by RT-PCR/immunohistochemistry (IHC)

 

Galectin-3

Galectin-1b

Galectin-7b

RT-PCR

IHC

RT-PCR

IHC

RT-PCR

IHC

Nervous system

 Cerebrum

+++

++c,a

+++

+

++

 Cerebellum

+++

+++

+

+

Digestive tract

 Liver

+++

+++d

+++

+

 Esophagus

+++

+++e,f

n. t.v

+

n. t.v

+++e

 Stomach

+++

++g,a

+++

+

+++

 Duodenum

+++

+h,a

+++

+

 Jejunum

+++

+h,a

+++

+

+

 Colon

+++

+++h,a

+++

+

+

+++t

 Caecum

+++

+h,a

+++

+

+

Respiratory and circulatory tract

 Lung

+++

++i

+++

+

 Heart

+++

+++

+

++

Urogenital tract

 Kidney

+++

++j,k,a

+++

+/+++k

 Urinary bladder

+++

+l,a

+++

+

++

+++l

 Ovary

+++

+++m

+++

+u

+++

+++u

 Testis

+++

+n

+++

+/+++s

+++

 Epididymidis

+++

+++o

+++

+

+

 Uterus

+++

+++p,a

+++

+

+++

Miscellaneous

 Spleen

+++

+++q

+++

+

++

 Skin

n. t.v

+++r,f

n. t.v

+

n. t.v

+++r

Signal intensity was semiquantitatively grouped into the categories: − negative, + weak but significant, ++ medium, +++ strong

aexclusively cytoplasmic positivity; bfrom Lohr et al. 2007; ccuboidal/columnar epithelial cells of the ependyma; dKupffer cells; estratified squamous epithelium; fLangerhans cells; gepithelium covering mucosal ridges and foveolae as well as epithelial lining of apical glands, immune cells in lamina propria and submucosa; hepithelial lining of villi/crypts, immune cells in lamina propria and submucosa; ialveolar macrophages; jepithelium of collecting tubules; ktransitional epithelium of the renal pelvis; ltransitional epithelium of the urinary bladder; mmacrophages in interstitium and corpora lutea; nsingle macrophages in interstitial spaces; oepididymal epithelium, caput epididymidis: single cells, corpus epididymidis: entire layer, cauda epididymidis: parts of the lining; pendometrial epithelium and macrophages in uterus; qmacrophages at high densities in the red pulp, occasionally in the white pulp; repidermal layers, epidermis-derived cells and external root sheath of hair follicles; stubuli seminiferi contorti; tepithelial lining of villi/crypts; uovarian surface epithelium; vnot tested

Computational promoter analysis

Our analysis with two search algorithms was based on the most recent update of the database for transcription factors and settings to include common motifs while maintaining stringent selectivity. Representation of putative target sites for rather ubiquitous factors such as Sp1, RFX-1, C-EBP or Runx family members is typical for a protein with expression in several cell types (for complete compilation, please see “Supplementary material”). The availability of the respective profiles for galectins-1 and -7, obtained by identical data processing (Lohr et al. 2007), facilitated a detailed intrafamily comparison so far not performed. To fully exploit the presented data on galectin-3, we extended this initial analysis to the same length of the sequence stretch also for galectins-1 and -7. Supplementing intergalectin differences in gene expression seen in Table 1, eight qualitatively discriminatory features emerged from the in silico work for galectin-3, when compared to respective information on galectins-1 and -7: the six motifs COUP-TFII (ARP-1), CUTL-1, E47 (Hand1; Tal-1α, β), FoxA4a, POU2F1/2 (Oct-1) and RFX-1 are unique to the galectin-3 promoter, among them a member of the steroid receptor superfamily with COUP-TFII (ARP-1) present in theca cells in normal human ovary (Sato et al. 2003), the two potential target sequences for GATA-box-binding factor-3 and gut-enriched Krueppel-like factor shared by galectin-1 and -7 promoters were not found in the corresponding sequence of the galectin-3 gene. Except for the case of FoxA4a the unique presence of target sequences concerns transcription factors with rather ubiquitous presence and context-dependent activity profiles. Turning back to in vivo parameters, a case study on regulation of galectin-3 expression and its consequences on the cellular level will be realized when examining the ovaries and estrous cycle. In doing so, the issue of relation of galectin-3 expression to two physiological processes of apoptosis in this organ, i.e., atresia and luteolysis, is also addressed.

Galectin-3 and atresia/luteolysis

In accord with the positivity in RT-PCR analysis for galectin-3-specific cDNA (Table 1) corresponding signals in Northern and Western blots were detected (Fig. 2). In order to probe regulation in the course of the estrous cycle, hormonally induced synchronization was performed in immature animals. A pronounced increase of transcription and subsequent translation was observed (Fig. 2). Mock injections of saline to female littermates had no effect, that is, the signal intensity remained constant at control level, as shown for day 5 (Fig. 2). Western blots also resolved the question on in situ truncation during this period. As already illustrated in Fig. 1a, the assay was capable to detect processed galectin-3. In extracts of ovaries, though, no evidence for a second band was obtained (Fig. 1a), galectin-3 was apparently not an in situ substrate for collagenases/matrix metalloproteinases to a notable extent in ovary. This conclusion based on biochemical work was corroborated by parallel immunohistochemical monitoring with the homemade polyclonal antibody and the anti-Mac-2 antibody reactive with the removed N-terminal section, because it disclosed no visible difference (not shown). Of note, this work using two different galectin-3-specific antibodies also served as inherent control to strengthen the validity of any detected cell-type specificity.
https://static-content.springer.com/image/art%3A10.1007%2Fs00418-008-0465-0/MediaObjects/418_2008_465_Fig2_HTML.gif
Fig. 2

Northern and Western blots for galectin-3 and actin (loading control). Time courses of signal intensity after chemiluminescence detection revealed strong expression on the levels of mRNA (a) and protein probed with specific antibodies (b) after day 6, starting the cycle at day 1 with injection of PMSG (*) followed by hCG injection on day 3 (**); ‡ control treatment with injection of saline solution

Galectin-3 has a distinct profile of cellular expression in the ovary. Positivity was consistently seen in cortical interstitium between developing follicles, in medullary interstitium diffusely associated with vessels and sporadically in the theca cell layer (Table 2). As also documented in Table 2, the cellular expression profiles of galectin-3 vs. galectins-1/-7 were conspicuously different. The ovarian surface epithelium, the only site of positivity for the two homodimeric lectins, did not show any reactivity for galectin-3. To complete the galectin fingerprinting of the ovary we ran RT-PCR analyses for further galectins and could only pick up weak signals for galectins-8/-9-specific cDNAs (not shown). In principle, all produced fragments were cloned, amplified and sequenced to ascertain their identity. Having monitored specimens of a total of 43 WT animals, the illustrations in the following paragraphs can be considered to be representative.
Table 2

Histochemical profiling of presence of galectins-3, -1 and -7, F4/80 antigen, 3β-HSD, cleaved caspase-3, DNA fragmentation and galectin-3-reactive sites in ovaries of WT mice

Tissue part

Marker

Galectin-3h

Galectin-1

Galectin-7

F4/80a,h

3β-HSD

Cleaved caspase-3h

DNA fragmentationh

Galectin-3-reactive sites

Preantral follicles

 Healthy

  Granulosa cell layer

+

++/+++g

  Theca cell layer

+a

+

++f

 Atretic

  Granulosa cell layer

++a

+

+

+++

++/+++g

  Theca cell layer

+a

+

++f

+

Antral follicles

 Healthy

  Granulosa cell layer

+++

+

++/+++g

  Theca cell layer

+a

+

++f

 Atretic

  Granulosa cell layer

+

+

+++

+++

  Theca cell layer

+a

+

++f

+

Ovarian surface epithelium

 Healthy epithelium

+a

++

+++e

 At postovulatory site

+

+

Corpora lutea

 Newly formed

+a,b, +/++a,c

+b, +/++c

+++b, ++c

b, ++c

b, ++c

 In regression

+++a,d

+++d

+d

++d

+++d

Staining intensity was semiquantitatively grouped into categories: − no staining, + weak but significant staining, ++ medium staining, +++ strong staining

aMacrophages; b24–48 h post injectionem (p.i.) of hCG; c72 h p.i. of hCG; d >96 h p.i. of hCG; efrom Lohr et al. 2007; fclusters of stained cells; gcorona radiata; hnumber of positive cells was grouped into categories: − no cells stained, + single cells, ++ clusters of cells, +++ >50% of cell population (intensity rather strong and homogeneous)

Galectin-3 was localized in atretic preantral follicles but not antral follicles and also in corpora lutea, the number of positive cells increasing in intensity during progression in (luteal) regression (Fig. 3a–d). Even remnants retained strong positivity. Lack of galectin-3 expression in KO mice invariably led to complete absence of the signal while tissue organization showed no signs of alteration (Fig. 3e). As a measure for cell-type identification we tested expression of particular markers. The staining profiles of the murine macrophage marker F4/80 and galectin-3 in corpora lutea in regression were indistinguishable (Fig. 3d, f). Galectin-3 staining was thus nearly exclusively due to macrophages (Table 2). To confirm that progesterone-producing luteal cells were not sites of galectin-3 expression, we visualized the key enzyme for this synthesis, i.e., 3β-HSD. The expected differences in staining were indeed verified in terms of spatial and temporal parameters (Table 2; Fig. 3g–j). Furthermore, analysis of effectors in apoptosis will enable to integrate galectin-3 expression into the order of events. Because the activation of caspase-3 is a key step within the effector cascade toward regression of corpus luteum, the appearance of the product of this process was put under scrutiny. As expected, first increased presence of cleaved caspase-3 and then of DNA fragmentation were seen during the estrous cycle in hormonally synchronized immature mice (Table 2). Obviously, galectin-3 presence coincided with the latter parameter in corpora lutea (Fig. 4a–d). This association was also observed in atretic preantral follicles but not in the atretic antral ones (Table 2; Fig. 4e–i). Relevant for the assumption of an indispensable role of galectin-3 in atresia and luteolysis the comparative analysis of tissue specimen from 36 immature mice deficient in galectin-3 expression did not come up with a notable parameter change distinguishing the two groups, as was also true for comparing two cohorts of mature females (12 mice per group). In the final step, we tested labeled galectin-3 to prove its suitability as probe and to localize accessible binding sites.
https://static-content.springer.com/image/art%3A10.1007%2Fs00418-008-0465-0/MediaObjects/418_2008_465_Fig3_HTML.jpg
Fig. 3

Immunohistochemical localization of galectin-3, the F4/80 antigen and 3β-HSD in fixed sections of ovaries removed from hormonally synchronized mice (WT and KO) at different time points after cycle induction. a Granulosa cells of atretic preantral follicles (atf, arrows; inset, a1) and interstitial cells (arrows) between two antral follicles (*) were reactive for galectin-3 40 h after PMSG treatment. Only very limited staining for galectin-3 in corpus luteum (cl) was detectable 24 h after hCG injection in this specimen, numerous positive cells in an atretic follicle (atf) serving as positive control (b), positivity in cell number (arrows) increased 24 h thereafter (c), the intensity being strong in cytoplasm (inset, c1). d Clusters of galectin-3-positive cells in corpora lutea (cl) in regression, but not in antral follicles (*) 96 h after hCG injection. e Sections from gal-3−/− mice were consistently free of staining, ascertaining complete absence of antigen-independent staining (please see also Fig. 1b). f Reactivity for F4/80 antigen resembled the staining pattern of galectin-3 presence illustrated in (d). Abundant positivity for 3β-HSD (g) and limited galectin-3 presence, in this case concerning few cells (arrows, h), detectable in corpora lutea (cl) of mice 72 h after hCG injection. (i) When the number of 3β-HSD-positive cells decreased upon regression of corpora lutea (cl), galectin-3 presence was observed (j). Bar length 50 μm (a–j), 10 μm (a1), 5 μm (c1)

https://static-content.springer.com/image/art%3A10.1007%2Fs00418-008-0465-0/MediaObjects/418_2008_465_Fig4_HTML.jpg
Fig. 4

Immunohistochemical detection of cleaved caspase-3, DNA fragmentation and galectin-3 in ovaries of hCG-treated mice. Presence of cleaved caspase-3 (arrowheads) and galectin-3 (insets) was not correlated in corpora lutea 72 h (a; inset, a1) and 96 h (b; inset, b1) after injection. Positivity with the TUNEL reagent (arrowheads) corresponded to galectin-3 detection in F4/80-positive cells (see insetsa1, b1) in regressing corpora lutea 72 h (c) and 96 h (d) after injection. e, f Granulosa cells in atretic antral follicles (*) adjacent to antral spaces in the membrana granulosa, harboring both cleaved caspase-3 reactivity and DNA fragmentation (*, f), were devoid of galectin-3 (inset, f1). g Atretic preantral follicles were negative for cleaved caspase-3 (positive cells belong to adjacent atretic antral follicles) but presented DNA fragmentation (h, arrows) and galectin-3 (i, arrowheads) in the granulosa cell layer. Bar length 50 μm (a–f, a1, b1, f1), 10 μm (g–i)

Labeled galectin-3 as probe

Processing of sections yielded a distinct staining, which was completely inhibitable by 75 mM lactose (Fig. 5a). In addition to the complete lectin with its three domains we also tested its truncated derivative. In accordance with the inhibition of binding of the full-length protein by haptenic sugar, the parallel testing of the labeled lectin domain resulted in similar staining profiles. Thus, the interaction underlying staining exclusively engaged the lectin domain. On the cellular level, granulosa cell layers in vital preantral and antral follicles and in atretic preantral follicles were positive (Table 2; Fig. 5a). The latter region contained galectin-3, whereas atretic antral follicles were negative for both parameters (Figs. 2a, 5b). Binding profiles in KO mice were very similar. On the electron microscopical level, reactivity for the lectin was assigned to slender processes of granulosa cells, which entered the zona pellucida (Fig. 5c).
https://static-content.springer.com/image/art%3A10.1007%2Fs00418-008-0465-0/MediaObjects/418_2008_465_Fig5_HTML.jpg
Fig. 5

Lectin histochemical localization of galectin-3-reactive sites and immunohistochemical detection of galectin-3 presence in follicles at two levels of magnification. a Visualization via fluorescence after incubation with biotin-labeled galectin-3 revealed reactive sites in granulosa cells (gc) of the corona radiata (arrowheads) in contact with the oocyte (oc) of an antral follicle (*) and in remnants of an atretic preantral follicle (atf). Pre-incubation with lactose resulted in blocking of lectin binding (arrow, inset, a1). b Galectin-3 was present in surrounding corpus luteum. c On the electron optical level, reactivity for the labeled lectin was found in an antral follicle, predominantly in slender processes (arrows) of a granulosa cell (gc), which traverse the zona pellucida (zp) to contact the oocyte. ×20,000. Bar length 10 μm (a, a1, b)

Discussion

A key histochemical step in the overall analysis of multifunctionality of proteins is monitoring cellular characteristics of expression. Following rigorous controls including processing of specimens from KO mice the tested and further refined anti-galectin-3 IgG preparation yielded a non-uniform profile, with various cell types being devoid of lectin presence. It is definitely not a general housekeeping product. Of note for intrafamily diversity, the pattern was disparate from those of the homodimeric galectins-1 and -7. They are potent effectors of negative cell growth regulation in keratinocytes, activated T cells and various tumor cell types, galectin-7 referred to as p53-induced gene-1 owing to its conspicuous transcriptional upregulation prior to onset of apoptosis in human DLD-1 colon carcinoma cells (Polyak et al. 1997; Bernerd et al. 1999; Rappl et al. 2002; Kopitz et al. 2003; Sturm et al. 2004; André et al. 2005, 2007; Fischer et al. 2005). These results implied marked intrafamily diversification of the promoter regions, and the ensuing application of two search algorithms confirmed this expectation. Beyond this, it gave respective insights into the presence and spatial arrangement of putative transcription-factor-binding sites. Mostly, the qualitative differences concerned sites for rather ubiquitous factors, whose activity profiles depend on the context in a combinatorial manner. Admittedly, results of computational searches cannot be free of ambiguities dependent on settings and selection criteria. The acquired compilation of putative targets for transcriptional regulation fitting accepted formal sequence criteria and the resulting potential for combinatorial usage will certainly have to be substantiated by experimental screening using a systematic series of deletions or by monitoring cells or mice deficient for a distinct transcription factor. Definitely, the presented collection gives these efforts a clear direction. To allay concerns of limited relevance and to provide an instructive example the recent documentation of actual regulation of galectin-3 expression, e.g., in growth plate cartilage, by a member of the runt family of transcription factors, listed as likely target in the “Supplementary table”, clearly reinforces the inherent potential of the computational analysis (Stock et al. 2003). Dorsal root ganglia and skin are further examples of combined expression of a runt factor with galectin-3 (Stock et al. 2003).

Special attention was then given to galectin-3 expression in the ovary. Signals in Northern and Western blots informed us about a clear correlation between transcription and protein production, with no apparent indication for an involvement of miRNA-dependent interference via miR-322 (Ramasamy et al. 2007). On the histological level, macrophages were prominently positive, their number increasing with luteal regression. A steroid dependence of galectin-3 expression had so far been observed by differential display of estrogen-responsive genes upon maturation of rat neuroendocrine hypothalamus with a 1.76-fold increase (Choi et al. 2001) and differential sensitivity of rat skin fibroblasts vs. fetal calvaria cells to dexamethasone (Aubin et al. 1996). The recently reported presence of immunopositive cells in the ovarian stroma (Kim et al. 2007) could be attributed to F4/80-antigen-positive cells, and the relation between presence of galectin-3-specific transcripts and regression of corpora lutea (Nio and Iwanaga 2007) was extended to the protein level. Moreover, macrophages with galectin-3 positivity were related to follicular atresia. Monitoring galectins-1 and -7 under identical conditions disclosed marked intrafamily differences. No obvious connection between these two galectins, which regulate growth in several other cell types, and atresia or luteolysis could be drawn. The respective immunohistochemical results also afforded further comparison to literature data based on in situ hybridization. Whereas the reported surface localization of galectin-7-specific mRNA was confirmed, immunohistochemistry for galectin-1 detected positivity equally confined to the surface epithelium, in contrast to presence of transcripts in stroma and some corpora lutea (Nio and Iwanaga 2007). As with the relation between detection of transcript and protein, care should also be exercised regarding interspecies extrapolation. Galectin-1, initially spotted in total extracts of murine ovaries by Northern blots (Choe et al. 1997), was consistently undetectable in murine granulosa cells, in contrast to lysates of porcine granulosa cells (Walzel et al. 2004). The murine granulosa cells, in turn, were positive for galectin-3 in preantral, but not antral atretic follicles. This signal coincided with occurrence of DNA fragmentation, and, of particular note, presence of accessible sites for labeled galectin-3. This result underscores the potential of applying the tissue lectin as probe. Its application has so far revealed staining profiles overlapping with lectin localization, even with prognostic relevance in head and neck tumors, or separate from each other (Gabius et al. 1991; Plzák et al. 2001, 2004; Kübler et al. 2008). As in this case, lectin secretion or transport are then required to let these sites become functional. This lectin’s unique combination of reactivity to particular glycan and also peptide motifs precludes considering a galactoside-specific plant lectin as equivalent substitute.

As seen for an aspect of follicular atresia, regression of corpora lutea was also associated with increase of galectin-3 presence, in parallel with emergence of DNA fragmentation and subsequent to the detection of cleaved caspase-3. This protease is an indispensable effector in luteolysis (Carambula et al. 2002). Obviously, galectin-3 is in this case not involved in modulation of regulatory events at the level of mediators of caspase-3 activation such as Bcl-2 or Bax, both proteins proven to be operative in luteolysis (Ratts et al. 1995; Hsu et al. 1996; Perez et al. 1999; Matsuda-Minehata et al. 2006). Conversely, a role for galectin-3 at a later stage, for instance to assist in removal of remnants as suggested for demyelinization (Smith 2001) and/or to protect against excessive apoptosis as inferred in cholesteatomas (Sheikholeslam-Zadeh et al. 2001), albeit not being essential based on analyzing KO mice, appears to be an option. This conclusion supports the emerging concept of cell-type-specific functions for this multifunctional protein.

It is in this respect illuminating to draw attention to the fact that we saw no evidence for proteolytic truncation of galectin-3 in the ovary, independently tested by Western blots and by immunohistochemistry with the anti-Mac-2 antibody. This monoclonal antibody reacts with a determinant in the detached N-terminal section of galectin-3 (Gong et al. 1999). Consequently, enzymatic processing in situ, for instance by matrix metalloproteinase-9, will eliminate galectin-3’s reactivity and likely also affect its activity as modulator of cell growth. Indeed, this mechanism was delineated to switch off extracellular galectin-3 functionality in endochondral ossification targeting chondrocytes (Ortega et al. 2005). When in the terminal hypertrophic zone, these cells undergo apoptosis. This process is another instance of physiological cell death regulation. Its rate can be retarded and uncoupled from vascular invasion and trabecular bone formation by genetically engineered loss of this enzyme (Ortega et al. 2005). Prolonged maintenance of integrity of galectin-3, that is, its protease-sensitive trimodular arrangement, slows down apoptotic decay of these late chondrocytes. A precedent for switching off the growth-regulatory activity of galectin-3 by removal of the non-lectin domains concerned neuroblastoma cells, likely by impairing the ability to cross-link suitably positioned cell surface glycans (Kopitz et al. 2001). Fittingly, chondrocyte apoptosis became precocious in mice deficient in galectin-3 expression (Colnot et al. 2001). By the way, galectin-3 expression at this site is under the control of the transcription factor Runx2/Cbfa1, a key regulator of osteoblast differentiation and chondrocyte maturation noted above (Stock et al. 2003), potential target sites having been tracked down in the promoter region by our in silico search. To solidify the notion for cell-type specificity of galectin-3 functionality, intimated by these in vivo cases, a brief reference should also be given to the situation in tumors. Here, the correlation to clinical parameters varies significantly even within a distinct tumor class, e.g., head and neck cancer, and the relationship between growth features and presence of galectin-3 not necessarily matches its activity in in vitro models (Honjo et al. 2000; Piantelli et al. 2002; Kayser et al. 2005; Saussez et al. 2006; Moisa et al. 2007).

In summary, the presented results answer the questions given in the “Introduction” by documenting (1.) a cell-type-specific expression profile of galectin-3 in murine organs, (2.) marked differences in relation to two expression profiles of homodimeric proteins of this family of adhesion/growth-regulatory lectins, (3.) the complex and distinctive array of putative transcription-factor-binding sites in the proximal promoter region of the galectin-3 gene, (4.) an association of galectin-3 expression in granulosa cells of preantral but not antral atretic follicles and in regressing corpora lutea with occurrence of DNA fragmentation and (5.) applicability of the labeled lectin for histochemical application.

Acknowledgments

The expert technical assistance of A. Helfrich, L. Mantel and C. Neumüller as well as the generous financial support of the research initiative LMUexcellent and an EC Marie Curie Research Training Network (contract 2005-019561) are greatly appreciated.

Supplementary material

418_2008_465_MOESM1_ESM.doc (116 kb)
MOESM1 (DOC 116 kb)

Copyright information

© Springer-Verlag 2008