Galectins as Markers of Aggressiveness of Mouse Mammary Carcinoma: Towards a Lectin Target Therapy of Human Breast Cancer

Report

Galectins as markers of aggressiveness of mouse mammary carcinoma: towards a

lectin target therapy of human breast cancer

E.V. Moiseeva1,2,�, E.M. Rapoport1,�, N.V. Bovin1, A.I. Miroshnikov1, A.V. Chaadaeva1,M.S. Krasilshschikova1, V.K. Bojenko3, Caspaar Bijleveld4, J.E. van Dijk2, and W. Den Otter21Shemyakin and Ovchinnikov Institute of Bioorganic Chemistry RAS, Moscow, Russia; 2Department of Pathobiology,Utrecht, The Netherlands; 3Russian Scientific Centre of Roentgen – Radiology, Moscow, Russia; 4Department of CellBiology, Utrecht, The Netherlands; � These authors contributed equally to this work.

Key words: breast cancer, galectins, glycoconjugates, metastases, mouse model, selectins

Summary

Galectins, b-galactoside binding proteins, expressed selectively in human breast carcinoma are attractive targetsto employ lectin-aimed therapeutics. We examined b-galactoside binding potency of neoplastic cells using fluo-rescein-labelled synthetic glycoconjugates as probes for flow cytometry. As a result, surface b-galactoside bindingproteins/galectins were discovered on mouse mammary carcinoma cells in vitro and in vivo unlike non-malignantcells from the several tissues; and asialo-GM1 ganglioside carbohydrate part - containing probe was the mostspecific one. However, in liver and lung metastatic cells galectins seem to be expressed within cytoplasm and/ornuclei. Galectin expression correlated directly with aggressive tumour potential in the A/Sn transplantable modelsimilar to findings in several human breast carcinoma cell lines. However, galectin expression was reduced duringtumour progression in more aggressive forms of spontaneous BLRB mammary carcinomas like it was shown forhuman breast carcinoma specimens. Analysis of the histopathological data led, however, to the conclusion thatgalectin expression hardly might be a suitable marker of aggressiveness of heterogeneous mammary carcinomas asthe observed level of galectin expression is influenced by the amount of the stroma in a tumour sample and/orprobably, galectin expression inversely correlates with tumour aggressiveness during the initial and advancedsteps of mammary tumour progression. We conclude that surface b-galactoside binding proteins/galectins that areselectively expressed during mouse mammary carcinoma progression, similarly to human breast carcinomas, seemto be proper targets for asialo-GM1-vectored cytotoxics and our mouse model system might be a relevantinstrument to further test novel modes of anti-breast cancer therapy.

Abbreviations: aGM1, Galb1-3GalNAcb1-4Galb1-4Glcb: asialo-GM1 ganglioside carbohydrate part; bGBP:b-galactoside binding protein; BC: breast cancer; BSA: bovine serum albumin; Glyc-PAA-fluo: polyacrylamide gly-coconjugate probes labeled with fluorescein; LacNAc: Galb1-4GlcNAcb, lactosamine disaccharide; PAS: PeriodicAcid -Schiff staining; PBS: phosphate buffered saline; PBA: PBS containing 0.2% BSA; PI: proliferation index;SiaLex: Neu5Aca2-3Galb1-4(Fuca1-3)GlcNAc tetrasaccharide; Tββ: Galb1-3GalNAcβ disacc haride; Versenesolution: PBS containing 0.02% EDTA

Introduction

Breast cancer (BC) remains the most common malig-nancy in women [1–3]. It develops up to age 70 among55–85% of women; in the Netherlands and formerUSSR there are about 32 and 15 deaths, respectively per100,000 women [4]. Currently used therapeutic ap-proaches to improve survival of patients have limitedsuccess in breast cancer clinic. Unfortunately, variousnovel techniques to treat breast cancer including cyto-kines and/or tumour vaccine application also have sev-eral restrictions [2,5]. Besides, there is a current lack of

adequate preclinical rodent models. Taken together,these circumstances compel to search for new therapymodes using appropriate mouse models of breast cancer.

Being interested in a rational design of carbohy-drate-based lectin-aimed therapeutics against BC wehave developed synthetic glycoconjugates, firstly, asdiagnostic tools to identify lectins [6] and, secondly, asconstructs for targeted delivery of the therapeutics [7].In particular, we developed cytotoxic merphalan bear-ing liposomes equipped with lipophilic tetrasaccharideSiaLex (synthetic selectin ligand) as a vector [8]. Wedemonstrated promising therapeutic effect of this drug

Breast Cancer Research and Treatment (2005) 91: 227–241 � Springer 2005DOI 10.1007/s10549-005-0289-8

formulation in the transplantable BLRB mouse mam-mary carcinoma model. However, cytotoxic liposomesaimed mainly to selectins that expressed in mammarytumours (probably on endothelial cells of the newly arisenintratumour vascularization) were not capable to rejectmammary cancer completely in both transplantable [8]and spontaneous (manuscript is in preparation) BLRBmouse models of human BC.

Other lectins, such as galectins (b-galactoside bindingproteins, bGBPs) seem attractive targets for lectin aimedanti-BC therapy [reviewed in 9]. The invasive and met-astatic behaviour of breast cancer might be associatedwith balance between glycosylation and galectinexpression on the tumour cells; the molecular signifi-cance of this correlation is proposed to be the interac-tions of galectins with polylactosamines on matrixproteins such as laminin, aiding cellular invasion [re-viewed in 10]. Moreover, galectins that recognize b-galactoside containing oligosaccharides have beenimplicated in multitude of cell processes including cellgrowth, adhesion, and apoptosis [reviewed in 9–11].Recently, expression of galectin-1, galectin-3, galectin-4,galectin-7, and galectin-8 on non-invasive and invasivebreast carcinomas was confirmed by immunohisto-chemical methods or Cancer Genome Anatomy Projectlibrary analysis [12]. Galectins were identified both inthe cytoplasm and the nucleus of breast cancer cells[13–18].

Several studies have demonstrated that galectins(mainly, galectin-1 and galectin-3) could be down- orupregulated in breast carcinoma cells in models in vi-tro and in vivo [13,14], and in breast carcinomaspecimens as well [15,16]. However, discrepancies havebeen observed in reported data that have hamperedclear answers on two key questions (i) to what extentdo galectin expression profiles differ for breast cancercells and surrounding non-malignant host tissues, and(ii) does galectin overexpression directly or reverselycorrelate with aggressiveness of human breast carci-nomas [reviewed in 16]. Several authors showed adirect correlation between galectin expression onbreast carcinoma cells in vitro and their invasive po-tential in experimental mouse models in vivo [13, 14].On the other hand, the increase of the histologicalgrade of breast carcinoma in a specimen leads to reducedexpression of galectin-3 resulting in decrease of tumourcell adhesion to extracellular matrix and increase of cancercell motility followed by metastatic spread [15,16]. Thisseems to be due to the ability of galectin-3 to arrest the cellcycle in G1 phase through inhibition of cell cycle regula-tors [17]. Galectin-1 may also interfere with cell cycleregulation via direct interaction with oncogenic Ras anti-gen [18]. Besides, galectin-1 as well as galectin-3 are ex-pressed on the surface of endothelial cells which might beinvolved in adhesion of tumour cells to endothelial cellsleading to enhanced angiogenesis [19].

These data presumes a major role of galectins inthe biological behaviour of breast cancer cells duringtumour progression, and therefore, galectins are

attractive as a target in anti-BC therapy. Detailedinterrelationships between galectin level and invasive-ness of BC remained unconvincing, and little is knownabout expression of galectins on metastatic cells fromhuman BC specimens. We expected that naturalexpression of galectins on mammary carcinoma cellsduring tumour progression (similarly to human breastcarcinoma cells) might be a proper model, which mayexplain current discrepancies in previously observeddata and to provide elucidation of underlying mecha-nism(s) involved. There are, however difficulties to find aproper animal model.

A few general characteristics of human BC arehardly reproducible in mouse models, which are incurrent use. First, the remarkable heterogeneity of hu-man BC is such an attribute, as indolent and fastmanifestations are both frequent in BC clinic. On thecontrary, the majority of mouse models represent rela-tively homogenous cohorts of mice bearing tumourswith similar characteristics. Second, the majority ofnaturally arising mouse mammary tumours metastasizethrough blood vessels to lungs, whereas human BCspreads also via lymphatics. Finally, the entire complexof inherent attributes of human BC is hardly reproducedin one mouse strain/model. Hence, we developed asystem of mouse models with different characteristicsand natural incidences of mammary cancer to reproducevarious forms of human BC [20–22]. Recently, we havereported about our mouse models as appropriate toolsto test alternative novel modes of anti-BC therapy[8,21,22]. Transplantable aggressive mouse mammarycarcinoma in A/Sn mice demonstrated high metastaticspread to liver, spleen, kidney, and lymph nodes com-parable with human BC [21]. In addition, we testedtherapy efficacy in mice bearing mammary tumours withfast and slow growth rates separately. However, thesetransplantable mouse models do not reproduce thewhole range of BC, which is usually a more chronicdisease that arises by gradual transformation of cells andevolutionary expansion of cell populations in situ.Autochthonous murine tumours better reproduce manyof the features of human BC but are experimentallymore cumbersome (tumours arise in females of variousage, range in sizes and tumour growth rates). However,tumour cells from spontaneous mammary carcinomatransplanted to syngeneic mice provide the first in vivopassage, which is generally considered to maintain his-topathological and growth parameter values of theautochthonous precursor and avoids the above men-tioned disadvantages of spontaneous tumour models.Therefore, transplantable mouse mammary tumours inA/Sn mice with various metastatic potentials mightcomplement naturally arising BLRB mammary carci-nomas (together with their first transplanted genera-tions) to mimic a wide range of human malignancy asthey imitate different groups of human BC: both fastgrowing poorly differentiated forms that are highlymetastatic to lymph nodes, liver, kidney and/or lungsand slowly growing hardly metastasizing well-differen-

228 EV Moiseeva et al.

tiated forms as well. Hence, the whole variety of ourmouse models was involved in this research. Thedynamics of galectin expression during mammary tu-mour progression might be an important characteristicof a mouse model per se. In addition, this informationwould be useful for the development of new therapeutictechniques including a lectin target therapy for breastcancer.

Therefore, the major aim of this work was to studythe expression of the bGBPs/galectins on neoplasticmammary cells from different new mouse breast cancermodels in vitro and in vivo.

Earlier, using purified galectin-1 and galectin-3 wehave identified three oligosaccharides, namely LacNAc,Tbb, and aGM1 as the most specific ligands to thesegalectins [23 and unpublished data, manuscript is inpreparation]. Using the corresponding glycoconjugateprobes (Glyc-PAA-fluo) that were labelled by fluores-cein for flow cytometry we focused on bGBP/galectinexpression profiles as possible markers of aggressivenessof mammary carcinomas.

Various mouse mammary carcinoma cell lineswhich differed in invasiveness and metastatic ability insyngeneic murine hosts and mammary carcinoma cellsfrom primary mouse spontaneous mammary tumoursand their metastases were probed by means ofthese three glycoconjugates, whereas selectin ligand,namely SiaLex-containing glycoconjugate probe[24] was mainly used as negative control. To furtherclarify the relationship between the bGBP/galectinexpression profiles and the aggressiveness of mammarytumours, histopathological research has been used. Wefound that carcinoma cells from all mouse modelsunder investigation expressed various bGBP/galectinpatterns on the cell surface. These findings holdpromise as a novel target therapeutic strategy forbreast cancer.

Materials and methods

Mouse models

Mice from the breeding nuclei of A/WySnJCitMoise [21]and BLRB-Rb(8.17)1Iem/Moise [20] (thereafter calledA/Sn and BLRB, respectively) were bred in our MouseBreeding Department at the Shemyakin-OvchinnikovInstitute of Bioorganic Chemistry, Moscow, under care-fully standardized conditions. Females in our colonies ofA/Sn and BLRB mice showed<5% and >95% naturallyarising mammary cancer, respectively. Cell lines weredeveloped from these tumours. They were histologicallydescribed and maintained by in vitro and in vivo passages.Males were used in a transplantation series. The recipientmice were 4–6 months of age at the moment of tumourtransplantation.

All experiments were performed in accordance withthe ‘Guide for the Care and Use of Laboratory Animals’(U.S. Department of Health and Human Services,

National Institute of Health Publication No. 93-23,revised 1985).

(A) In vitro mouse models

1. Established cell lines: VMR-O and VMR-L tumourcell lines were kindly provided by Dr. S.L. Kiselev(Moscow, Russia). VMR-O and VMR-L tumour celllines were established from a single moderately meta-static spontaneous mouse mammary carcinoma in anold A/Sn virgin by in vivo and in vitro passaging of theprimary tumour and metastatic foci to obtain low andhighly metastatic tumour substrains, respectively. Thesecell lines with long transplantation history producetumours in syngeneic A/Sn mice with low (VMR-O) andhigh (VMR-L) metastatic potential to liver after beingtransplanted subcutaneously (s.c.) in mice [25]. Cellswere cultured in DULBECCOmedium (Invitrogen, UK)supplemented with 10% FCS and 2 mM glutamine (In-vitrogen, UK) at 37� in an atmosphere of 5% CO2.

Short-term cell cultures: Short-term cell cultures wereobtained from various spontaneous mammary tumoursof BLRB females as a few of naturally arising mammarycarcinomas became available each week in our mousecolony. A cell suspension was prepared from tumourtissue, filtered under sterile conditions and cultured inRPMI 1640 medium (Invitrogen, UK) supplementedwith 10% FCS and 2 mM glutamine (Invitrogen, UK) at37� in an atmosphere of 5% CO2. After a few days deadcells were removed while viable cancer cells adhered tothe plastic flask.

(B) In vivo mouse breast cancer modelsVarious kinds of mammary carcinoma models in twomouse strains were used.

1. A/Sn i.p. transplantable model: Syngeneic A/Snmales were inoculated intraperitonealy (i.p.) with 1·106

VMR-L mammary carcinoma cells. Two weeks laterascites was collected, cells were cultivated in RPMI-1640in vitro for a few days and subjected to cytology andflow cytometry analyses.

2. Initial A/Sn s.c. transplantable model: Syngeneic A/Sn males were inoculated s.c. near the right flank with1·106 VMR-L mammary carcinoma cells. Three weekslater a tumour of about 10 mm in size and visible meta-static foci from liver were taken and representative tissuepieces were subjected to histological research. The restwas cut and cell suspensions were prepared by means of aMedicon machine (Dako, Denmark). The cells were usedfor the flow cytometry analysis.

3. Secondary A/Sn s.c. transplantable model: The firstA/Sn recipient was inoculated s.c. with VMR-L mam-mary carcinoma cells as described above. Three weekslater primary tumours of about 10 mm in size and liver,

Galectin expression in mouse models of breast cancer 229

spleen, and lymph nodes were taken from this mouse,cut and cell suspensions were prepared as mentionedabove. These tumour cell suspensions were s.c. injectedinto secondary A/Sn hosts (1·106 viable cells permouse). Three weeks later tumours derived from bothprimary tumour and metastatic foci were taken fromsecondary hosts, suspensions were prepared and flowcytometry analysis was performed.

4. BLRB spontaneous mammary carcinoma model:BLRB females have about 95% incidence of sponta-neous mammary carcinomas with almost 100% meta-static spread to lungs. About one third of the BLRBfemales got more than one primary tumour as mam-mary tissue is located in fat pads accompanying tennipples of the female, and thoracic and abdominallocalizations are the most frequent. Only females withtwo tumour nodules were used, namely a fast grow-ing, first appearing thoracic tumour of about 11 mmin size and a later appearing second small abdominalnodule of about 3 mm in size. Both large and smalltumours from the same female were collected forhistopathological examination and galectin detectionby flow cytometry. Cell suspensions from both pri-mary tumours and lungs (as metastatic foci werehardly visible) from the same BLRB female wereprepared as described above and subjected to flowcytometric analysis. The rest of suspensions wereinoculated to syngeneic hosts to create the transplan-table model.

5. BLRB s.c. transplantable model (the first pas-sage): Syngeneic BLRB mice were injected s.c. withsuspensions containing about 10·106 of viable cells aswas previously described in [8]. Cells were taken fromspontaneous large advanced and small later arisenmammary tumours and from lung (as metastaticprone tissue) from the same female to obtain the firsttransplantation generation from both primary tumoursand from lung metastases. Five weeks later tumoursof about 10 mm in size derived from primary tumoursand from lung metastases were taken from the re-cipient mice. A representative part of the tumour wasused for histopathological research; the rest was cutand cell suspensions were prepared as mentionedabove and flow cytometric analysis was performed.

Glycoconjugate probes

We used synthetic glycoconjugates, Glyc-PAA-fluo,where PAA is 30 kDa soluble hydrophilic polymer(polyacrylamide), fluo is fluorescein, Glyc is oligosaccha-ride, i.e. LacNAc (lactosamine disaccharide), aGM1

(Galb1-3GalNAcβ1-4Galβ1-4Glcβ tetrasaccharide), Tββ

(Galb1-3GalNAcβ disaccharide), or SiaLex (Neu5Aca2-3Galb1-4(Fuca1-3)GlcNAc tetrasaccharide). All Glyc-PAA-fluo were obtained from Lectinity (Moscow,Russia). The aGM1-PAA-fluo was synthesized asdescribed in [26]. The aGM1-PAA-fluo, Tbb-PAA-fluo

and LacNAc-PAA-fluo were used as specific galectinprobes ligands, whereas SiaLex-PAA-fluo that is selectinligand [23,24] was applied as a negative control.

Flow cytometry analysis

A cell suspension obtained from a tumour was rinsedwith cold PBA and incubated with Glyc-PAA-fluo(100 lM) for 40 min at 4�. After washing twice withcold PBA, fluorescence analysis was performed with aflow cytometer (Dako Galaxy, Denmark) usingFlowMax 2.0 program. In the case of cell cultures,adherent cells (2 · 106/ml) were harvested with coldVersene solution that causes cells to detach. Obtainedfloating cells were washed three times in cold PBAand incubated with probes.

Percentage increase of fluorescence intensity wascalculated as [100Fi/Fo]-100%, where Fi – fluorescenceintensity of cells preincubated with Glyc-PAA-fluo; Fo –fluorescence intensity of negative control cells that werenot preincubated with Glyc-PAA-fluo.

Representative data from one of the several inde-pendent experiments is shown in each figure.

Permeabilization of cells

Tumour cells (2·106/ml) were rinsed with PBS andincubated with 0.1% saponine (Merck, Germany) inPBS for 40 min at 37�. Cells were washed with PBS andflow cytometric analysis was performed.

Cell cycle analysis

DNA staining for cell cycle analysis of tumour cellswas performed as routine procedure according to [27]and DNA analysis was performed by means of flowcytometer (Dako Galaxy, Denmark). The FlowMax2.0 and ModFit 3.0 programs for cell cycle analysiswere used. Proliferation index (PI) was calculated as(S+G2M)/100.

Statistical analysis

Statistical analyses were performed on the data sets usingthe non-parametric Wilcoxon–Mann–Whitney U-test.

Cytological examination

Cytology research on Giemsa-stained smears was per-formed as routine procedure according to [21].

Histopathological examination

Briefly, fragments of primary s.c. tumours with sur-rounding tissue and regional lymph nodes were excisedand fixed in phosphate buffered 4% formalin solutionfor histopathological examination. The same procedurewas used for other metastatic prone organs – such aslung, liver, spleen, and kidney. Paraffin sections (5 lm)


were prepared and stained with haematoxylin-eosin and/or PAS. Three pathologists evaluated all histologicalsections independently. In several panel discussionsconsensus was obtained.

Mammary tumours and their metastases in vitallyimportant organs were mainly classified accordingSquartini and Pingitore [28] with essential modifications[21] following the logic to describe each specific com-ponent of histologically heterogeneous mammary tu-mour as distinct tumour type. Using this approach notype B was presented in our classification scheme toavoid the confusion with previously reported classifica-tions [28] where type B was widely used to illustrate asignificantly heterogeneous histopathological pattern.

Their main characteristics used for histopathologicaldescriptions are:

1. Mammary carcinoma, type A – adenocarcinomawith fine acinar structure is composed of small undif-ferentiated and mainly hyperchromatic epithelial cellsthat arranged in round cavities and elongate tubules;neoplastic cells rarely show mitotic figures. Intercellularcohesion is evident. The connective tissue is scanty. Thevascular bed is thin; necrosis is not found.

2. Mammary carcinoma, type AD – adenocarcinomaseems to be a result of type A mammary carcinomaprogression and is therefore considered as intermediatetype from A to D with more or less evident acinarstructure similar to type A but composed of larger epi-thelial cells with pale polymorphic nuclei similar to typeD; secretory activity is less evident. The level of loosingintercellular cohesion varies within the same section;mitotic figures are frequent. The amount of stromavaries notably. The vascular bed might be well devel-oped in larger tumours. Necrosis is easily seen in com-pressed sheets from hemorrhagic areas.

3. Pale mammary carcinoma, type D – carcinomamight be considered as an extreme case of an AD car-cinoma with total loss of intercellular cohesion andsecretory activity, composed of large neoplastic cells,rich in acidophilic cytoplasm with large round palestaining nuclei. So, glandular elements are absent withintype D of mammary carcinoma and an epidermoidstructure often prevails. The stroma component is thin.The arrangement of cells within the same section variesconsiderably: they form ductules, cords and/or bands.Neoplastic cells regularly show mitotic figures; necrosisis characteristic inside of these structures. The vascularbed is thin, so hemorrhagic areas may not be conspic-uous. The connective tissue is scanty within D areas, buttubules composed of D carcinoma might be embeddedin abundant stroma.

4. Mammary carcinoma, type C – cystic carcinomaarises within other types, mainly, lobular AD type andform cysts filled initially with fluid (secretion) but laterwith blood; therefore this tumour is often haemorrhagic.In small tumours this areas might resemble the so-calledcribriform pattern. In large tumours neoplastic cellsform intracystic papillary projections, solid cords,sheets, nests or tubes with no sign of glandular differ-

entiation. The amount of stroma varies; necrosis andhaemorrhages are frequent.

5. Mammary carcinoma of the bizarre type E – com-posed of intensely stained polymorphic cells; resemblesanaplastic and displastic forms of D type carcinoma andfrequently follows the latter in the projection of thesame tumour. Brightly basophilic extremely polymor-phic neoplastic cells show nuclei of various bizarreshapes. The cell size varies greatly; there is little if anyintercellular cohesion.

6. Fibrous stroma compounds of type F – present welldifferentiated elongated or stellate cells, which resemblefibroblasts and are mainly found in stroma of themammary carcinomas mentioned above. This mightimply that stromal elements are commonly non-malig-nant in naturally arisen mammary tumours. But intransplanted tumours with long transplantation historythese components might vary significantly in mitoticactivity, probably mimicking consequent stages of hu-man breast lesions, namely from fibromatosis tomalignant fibrosarcoma.

7. Various patterns of spindle cells proliferation, or,type G – presents sarcomatous shaped, elongated orstellate cells, so-called myoepithelial cells, which arebrightly basophilic. They frequently outline tumours ofother types, mainly AD and D type. Masses of these cellssometimes protrude into the stroma, compressing duc-tules and alveoli into thin cords.

Results

Flow cytometric analysis of Glyc-PAA-fluo bindingto VMR-0 and VMR-L cells in vitro with oppositemetastatic potencies in syngeneic hosts

VMR-0 and VMR-L cells from established cell lineswere probed by fluorescein labelled glycoconjugates.Data from one of three independent experiments arepresented in Figure 1. LacNAc-PAA-fluo and aGM1-PAA-fluo exhibited remarkable binding to VMR-L cellsthat were morphologically large polymorphic carcinomacells. This VMR-L cell line was characterized as poten-tially highly metastatic to the liver. However, VMR-0cells that morphologically resembled fibroblasts wereclearly less potent in binding to LacNAc-PAA-fluo andaGM1-PAA-fluo probes in vitro. This VMR-0 cell linewas characterized as potentially non-metastatic to theliver [25]. Both cell lines demonstrated poor affinity toTbb-PAA-fluo compared with aGM1-PAA-fluo, whichrepresented the same terminal Taa disaccharide. It mightimply the importance of other structures, such as innerlactose core in aGM1-PAA-fluo for interaction withtumour cells rather than only the terminal disaccharide.

To further confirm the morphology of VMR-0 andVMR-L cell lines with opposite metastatic potencies,these cells were transplanted s.c. to syngeneic A/Snrecipients and histopathological research was under-taken. VMR-0 originated tumours were characterized


by initially slow tumour growth. However, histopathol-ogical analysis of final stages of VMR-0 tumourrevealed abundant pleomorphic and hyperchromaticspindle-shaped cells arranged in fibrous tissue that wasinvasive to the skin. Advanced nodules were fast grow-ing, highly necrotic tumours. Details of both solid andloose tumour components are presented in Figure 2Aand B. An average of 10 mitoses per high-power fieldwas seen. No metastases were found in the liver, spleen,kidney and regional lymph nodes. Only small metastaticfoci of solid carcinomatoid and loose sarcomatoidcomponents in lungs within blood vessels (Figure 2B)were found in one sample. This facilitated the diagnosis ofmetaplastic spindle cell carcinoma of the G type withabundantmalignant fibromatosis of the stroma, F type forVMR-0 originated tumours.

Histopathological data on initially fast s.c. growingVMR-L tumours revealed abundant metastases in allmetastatic prone parenchymal organs, including liver,spleen, kidney, lungs and lymph nodes. Solid constitu-ents of the type D pale and peculiar type E mammarycarcinoma were detected in primary s.c. growing VMR-L tumour (Figure 2C) and its metastases (Figure 2Dand E; liver and lymph node metastases). Livers of allrecipients were full of metastases. Details from a livermetastatic focus are represented in Figure 2D. Fur-thermore, large macroscopic metastatic foci in the liversof animals bearing advanced tumours were found. Ad-vanced tumours and metastatic foci as well were extremelyhemorrhagic and necrotic. Metastatic foci regularly con-tained more aggressive tumour cells as in average 8–10mitoses per higher magnification field were scored in con-trast to 5–6 mitoses in primary tumours.

We can conclude that (i) LacNAc-PAA-fluo andespecially aGM1-PAA-fluo probes were effective inbinding to the VMR-L cells in vitro and (ii) the VMR-0cell line, as potentially less aggressive spindle cellcarcinoma with poor metastatic pattern exhibited lesssurface bGBPs/galectins in vitro than the VMR-L line aspotentially highly invasive poorly differentiatedmammary carcinoma with an abundant metastatic pat-tern.

I.p. A/Sn model. Flow cytometric analysis of Glyc-PAA-fluo binding to cells of the VMR-L derived ascites

Intraperitoneal (i.p.) transplantation of mammary car-cinoma VMR-L cell line led to an individual ratio ofleucocytes and tumour cells in the peritoneal fluid of eachmouse recipient (Figure 3A). Cell compositions of theascites from 11 mice were analyzed and a range of ratioswas obtained; flow cytometry assessments for all caseswere performed. Flow cytometric data of ascites II and Ifrom two mice were selected for presentation as theyprincipally differed in cell composition. Ascites I (Fig-ure 3B, diagonally striated columns) represented almostonly tumour cells, whereas ascites II (Figure 3B, greycolumns) contained equal amounts of tumour cells andleucocytes (lymphocytes, monocytes, and macrophages).Leucocytes from i.p. fluid of intact mouse were used asnegative control (Figure 3B, white columns). Flowcytometry analysis demonstrated that LacNAc-PAA-fluo and aGM1-PAA-fluo bound strongly to cells fromthe ascites I, whereas SiaLex-PAA-fluo and Tbb-PAA-fluo probes showed poor binding (Figure 3B). All thetested Glyc-PAA-fluo probes exhibited significantlylower biding to cells of ascites II as compared to cells ofascites I. On the other hand, LacNAc-PAA-fluo andaGM1-PAA-fluo probes demonstrated limited bindingto cells from peritoneal cavity of intact mouse. Thisstrongly suggests the overexpression of bGBPs/galectinson tumour cells from ascites.

Lower binding ability of cells from ascites II cannotbe explained by means of simple dilution of tumour cellsby leucocytes. To further find an explanation we sug-gested different properties of tumour cells from theseascites due to heterogeneity of tumour cell compositionthat might correlate with the galactoside-binding pattern.Indeed, two morphologically different subsets of mam-mary carcinoma cells in ascites were found by analysingof Giemsa stained smears, i.e. tumour cells with verydark basophilic cytoplasm (TC-dark); Figure 3A and 1and weak basophilic tumour cells (TC-light); Figure 3Aand 2. The morphology of TC-dark cells was that of apoorly differentiated fast growing tumour consisting ofextremely polymorphic round cells with compact hy-perchromatic nuclei, little cytoplasm, and frequentmitoses. This type of apparently more aggressive cellsprevailed in ascites I and contained almost exclusivelytumour cells that exhibited a high level of affinity toboth LacNAc-PAA-fluo and especially aGM1-PAA-fluo, this implies notable surface bGBP/galectin over-expression. In contrast, TC-light carcinoma like cellswere similar to common mammary carcinoma cellsshowing reddish reticulate nuclei with distinct nucleoliand a more differentiated appearance. This type of lessaggressive cells prevailed in ascites II that containedequal amounts of tumour cells and leucocytes andexhibited lower surface galectin level. Finally, we canconclude that (i) normal leucocytes hardly expressedsurface aGBPs/galectins, and (ii) more aggressive poorly

( ), g

0

50

100

150

200

250

300

350

Incr

ease

of f

luor

esce

nce,

%

L TacNAc aGM1 SiaLexββ

Figure 1. The VMR model, in vitro. Flow cytometric assay of the

Glyc-PAA-fluo binding to VMR-O (white columns) and VMR-L

cells (grey columns). Representative data from one of the three

independent experiments are shown.


differentiated neoplastic component of the VMR-L de-rived ascites expressed more surface galectins.

Initial A/Sn s.c. model. Flow cytometric analysis ofGlyc-PAA-fluo binding to tumour cells obtained fromthe liver metastases

VMR-L mammary tumour cells were s.c. transplantedto syngeneic A/Sn mice to create an initial in vivo model.Abundant metastatic pattern was characteristic in thelivers of all recipients (Figure 2D). Large macroscopi-cally visible metastatic foci in the livers of animalsbearing advanced tumours were found.

Cells from the macroscopic liver metastatic focuswere analysed by flow cytometry with all the threegalectin ligands (Figure 4). Neither cells from a normalliver (data not shown) nor cells from the liver metastaticfocus interacted with Glyc-PAA-fluo probes. However,after permeabilization of cells with saponine, strongbinding of LacNAc-PAA-fluo and aGM1-PAA-fluo tothe metastatic cells was observed (Figure 4). This im-plies that surface bGBPs/galectins are hardly expressedin normal liver cells and in metastatic cells as well,however carcinoma cells from the liver metastatic focusdemonstrated high level of galectin expression within thecell (in cytoplasm and/or nucleus).

Figure 2. The VMRmodel, in vivo. Morphology of the VMRmammary tumours arisen from cell lines after their s.c. transplantation to syngeneic

A/Sn females. (A) VMR-0. Details of abundant loose fibrous malignant compound, F type (left) in metaplastic spindle cell carcinoma, G type

(right) in the advanced necrotic tumour, HE staining; (B) VMR-0. Haematogenous metastatic spread to lungs corresponding to both solid

carcinomatoid compound of the G type (middle left) and sarcomatoid loose compound of the F type (top right) from the same A/Sn female with

s.c. tumour as depicted in (A), HE staining; (C) VMR-L. Details of the solid compounds of the type D pale and polymorphic type E dark stained

in s.c. tumour, HE staining; (D) VMR-L. Details in one of the abundant liver metastases, from the same A/Sn female with the s.c. tumour as

depicted in (C), HE staining; (E) VMR-L. Lymph node metastasis showing necrotic area (upper right), from the same A/Sn female as depicted in

(C), HE staining.


Flow cytometric analysis of Glyc-PAA-fluo binding totumour cells obtained from the secondary A/Sn s.c. model

Cells from primary tumour and liver, spleen, and lymphnode metastases from the initial A/Sn model wereinoculated to secondary A/Sn hosts; tumours of about10 mm from these recipients were analyzed histologi-cally and by means of flow cytometry. Histopathologicalresearch revealed highly aggressive forms of types D andE mammary carcinomas in s.c. growing tumours similarto the pattern depicted in Figure 2D; final stages ofadvanced tumours were extremely necrotic and hemor-rhagic. The LacNAc-PAA-fluo and the aGM1-PAA-fluo displayed strong affinity to cells from all s.c.growing tumours (Figure 5), whereas Taa- and SiaLex-containing probes were significantly less potent (datanot shown).

Cell cycle patterns for all tumours (data not shown)and the lectin binding potency of cells derived frominitially primary tumour and metastatic foci were simi-lar, and comparison of the two groups using a Wilco-xon–Mann–Whitney U-test showed no statisticallysignificant differences, although cells derived from

spleen metastases demonstrated notable increase ofaffinity to both tested probes (Figure 5).

Remarkably, the binding levels of LacNAc-PAA-fluo and aGM1-PAA-fluo were in average twice-higherto cells from s.c. growing VMR-L derived tumours thanto VMR-L cells growing in vitro or in ascites (compareFigure 5 with data depicted in Figure 1, grey columnsand 3B, diagonally striated columns). This implieshigher surface bGBP/galectin level on tumour cells froms.c. A/Sn models.

Flow cytometric analysis of Glyc-PAA-fluo binding tocells from short-term BLRB culture

Cells obtained from primary BLRB spontaneous mam-mary tumour were cultured under standard conditions.Lectin expression was studied on cultured tumour cellsafter they formed a confluent layer. Mammary carci-noma cells in cultures were recognized using Giemsastaining (Figure 6A). These cells were probed by Glyc-PAA-fluo (Figure 6B). LacNAc-PAA-fluo and espe-cially aGM1-PAA-fluo bound to cultured tumour cells;whereas glycoconjugate probe bearing its terminal

, g

Figure 3. A/Sn i.p. model. (A) Leucocytes and two subpopulations of tumour VMR-L cells: TC-dark (1) sarcomatous cells and TC-light

(2) carcinomatoid cells, ascitis, smear, Giemsa staining, original magnification · 400; (B) flow cytometric assay of the Glyc-PAA-fluo binding to

cells obtained from i.p. fluid: ascitis II contained equal ratio of tumour cells/leucocytes (grey columns), ascitis I contained only tumour cells

(striped columns), and leucocytes from peritoneal cavity of a normal mouse (white columns). Representative data from 2 of 11 independently

tested ascites are shown.

Figure 4. Liver metastases from initial A/Sn s.c. transplanted model. Flow cytometric analysis of the Glyc-PAA-fluo binding to liver metastatic

cells before (white columns) and after (grey columns) permeabilization of cells with saponine. Representative data from one of two independent

experiments are shown.


disaccharide fragment Tbb-PAA-fluo displayed poorbinding. The SiaLex-PAA-fluo (selectin ligand) alsoshowed limited binding.

To prove mammary carcinoma origin of cells thatwere maintained in vitro (Figure 6A) 104 culturedtumour cells were transplanted i.p. to a BLRB male host.Six month later this male was sacrified, and histo-pathological analysis revealed characteristic mammarycarcinoma growth in the peritoneum and in the liver.Solids were composed mainly of the D carcinoma (datanot shown); however, abundant liver metastases showedtypical glandular structures surrounded by conspicuoussolid pale carcinoma of the type D (Figure 6C and 3).Adenocarcinoma components of the type A (Figure 6Cand 1) and type AD (Figure 6C and 2) contained PAS-positive material.

Finally, we can conclude that BLRB mammary car-cinoma cells cultured in vitro expressed surface bGBPs/galectins rather than selectins.

Spontaneous BLRB model. Flow cytometric analysis ofGlyc-PAA-fluo binding to tumour cells obtained fromlarge and small BLRB primary spontaneous tumours

Two spontaneous mammary tumours from the sameBLRB female with an initial advanced thoracic noduleand a later arisen small groin nodule were used toidentify surface bGBP/galectin expression; representa-tive data from one of two independent experiments aredescribed here (Figure 7C). Larger tumour (L; 11 mm indiameter) was characterized by initial fast tumourgrowth (histopathological pattern of a representativesection is depicted in Figure 7A), whereas smallertumour (S; 3 mm in diameter) initially demonstrated aslow tumour growth rate (Figure 7B). Cells obtainedfrom the large L and small S primary BLRB sponta-

neous mammary tumours were probed by the LacNAC-PAA-fluo, the Tbb-PAA-fluo, and the aGM1-PAA-fluoprobes (Figure 7, bottom). Both the LacNAc-PAA-fluoand the aGM1-PAA-fluo probes were the more potentsimilar to the case of a short-term BLRB culture,whereas the Tbb-PAA-fluo probe was considerably lessactive both in vitro and in vivo. Cells from the small Stumour interacted stronger with both the LacNAc-PAA-fluo and the aGM1-PAA-fluo probes than cellsfrom the larger advanced tumour. This implies thatbGBP/galectin expression was higher on the surface oftumour cells from the small nodule.

To investigate whether galectin expression correlateswith aggressiveness of used carcinomas histopathologyresearch was undertaken. The large L tumour, whichexpressed less galectins was mainly composed of themore aggressive types D and E in addition to small areasof AD of carcinoma (Figure 7A: 1, 2, and 3, respec-tively); whereas the small S tumour that expressed moregalectins was mainly composed of less aggressive secre-tion positive A and AD types of adenocarcinoma (Fig-ure 7B: 3 and 4, respectively). So, this reversecorrelation between galectin expression and carcinomaaggressiveness seems to be dissonant to data describedabove for the A/Sn model where more aggressive tu-mour samples exhibited more surface galectins. Buttaking into account that spontaneous tumours arecomposed of two components, namely carcinoma andfibrous non-malignant connective tissue stroma, and Ltumour exhibited a broad amount of fibrous tissue(which is common for advanced murine mammary tu-mours), whereas S tumour contained scanty connectivetissue we suggest that sample from L tumour had lessaggressive tumour cell composition at the moment ofgalectin assessment and expressed less surface bGBPs/galectins. So, there was no discrepancy with previouslydescribed data.

Figure 5. Secondary A/Sn s.c. transplantable model. Flow cytometric assay of the Glyc-PAA-fluo binding to cells from the s.c. tumour derived

from the VMR-L primary tumour (white columns), s.c. tumour derived from liver metastatic focus (grey columns), from the spleen metastases

(diagonally striped columns), and from lymph nodes metastases (vertically striped columns). Representative data from one of two independent

experiments are shown.


Spontaneous BLRB model. Flow cytometric analysis ofGlyc-PAA-fluo binding to tumour cells obtained from lungmetastases

Lungs with tiny metastatic foci from the BLRB femalewith large and small spontaneous mammary carcinomaswere subjected to histopathological research and flowcytometry. Cystic mammary carcinoma formed themetastatic foci in lungs (Figure 8A). Detailed analysisrevealed carcinomas of the AD and D types surroundedwith the E type within a focus. The cells from lungs withmetastatic foci were analysed by flow cytometry with theglycoconjugate probes containing LacNAc, Tbb, andaGM1 oligosaccharides (Figure 8B). Neither lung met-astatic cells nor cells from normal lung (negative con-trol; not shown) interacted with these probes(Figure 8B, white columns). However, after permeabi-lization of cells with saponine strong binding of all three

probes to cells from the metastatic foci was observed(Figure 8B, grey columns). This implies that similarly tothe case with liver metastatic foci in A/Sn model, surfacebGBPs/galectins are hardly expressed on normal cells(at least cells from the lung and the liver) and on met-astatic cells as well. However, metastatic cells containeda high level of bGBPs/galectins expressed within the cell(in cytoplasm and/or nucleus).

Transplantable BLRB model. Flow cytometric analysis ofGlyc-PAA-fluo binding to tumour cells obtained from thefirst in vivo passages derived from large and small BLRBprimary spontaneous tumours and from lung metastasesas well

Flow cytometric analysis of tumour cells that were ob-tained from the first transplantation generation from thelarge (L1) and the small (S1) primary spontaneousmammary tumours, and lung metastases (M1) from thesame BLRB female was undertaken. In addition, cellcycles were measured for analysed cell populations andhistopathology analysis of tested mammary carcinomaswas performed to find correlation between galectinexpression, histopathological pattern, and tumour pro-liferative capacity. Reproducible data from one of fourindependent experiments are depicted in Figure 9D. TheaGM1-PAA-fluo probe interacted stronger with tumourcells obtained from L1, S1 and M1 tumours (Figure 9D,grey columns) as comparing with LacNAc-PAA-fluoprobe (Figure 9D, white columns). The Tbb- and SiaLex-containing probes showed poor biding (data are notshown). The levels of aGM1-PAA-fluo and LacNAc-PAA-fluo binding were similar for all samples althoughthe binding to S1 was noticeably higher than binding toL1, and the highest interaction was revealed for bothligands to M1 tumour cells. These three samples werecharacterized by low percentage of cells in S and G2Mphases (PI was about 0.1–0.2). This indicated a lowerproliferative capacity of tumours in the BLRB model ascompared to the A/Sn model where PI was scored asabout 0.50 and galectin overexpression was higher(Figure 5). Histopathological analysis also revealedlower grade of mammary tumour aggressiveness in theBLRB model. First passages of all tested tumoursresembled spontaneous precursors in great lines (com-pare L and L1, for example) with one important addi-tion – there was much more fibrous stromal componentshown in all transplanted tumours. Tumour L1 thatderived from primary large spontaneous tumour wasmainly composed of tubular solid carcinoma of both Dand E types with small glandular areas and plenty offibrous stroma (Figure 9A). Tumour S1 that derivedfrom primary small spontaneous tumour was mainlycomposed of solid adenocarcinoma of AD type withsmall carcinoma D areas and scanty fibrous stroma(Figure 9B). However, tumour M1 that derived fromlung metastases was obviously composed of the moremalignant carcinoma component E prevailed; middleamount of stroma was found in M1 tumour (Fig-

Figure 6. The BLRB model, in vitro. (A) Morphology of the mam-

mary carcinoma cells, primary short-term culture obtained from a

BLRB spontaneous mammary adenocarcinoma, Giemsa staining; (B)

flow cytometry assay of the Glyc-PAA-fluo binding to tumour cells

from the short-term BLRB culture, representative data from one of

four independent experiments are shown; (C) the A (1) and AD (2)

types surrounded with the D type (3) of carcinoma originated from the

BLRB short-term primary culture depicted in (A) after i.p. trans-

plantation of cells to a BLRB male, liver metastasis, PAS staining.


ure 9C). If we propose the absence of surface galectinsin non-malignant stroma, then we can explain whyhigher aggressive sample L1 expressed less galectins thanless aggressive sample S1. Taking into account that non-malignant stroma might simply dilute higher aggressivecarcinoma component in the sample, we can concludethat in general the higher aggressive tissue showedhigher bGBP/galectin expression.

Discussion

Based on a crucial role of galectins in human BC pro-gression we studied the b-galactoside binding protein(bGBP)/galectin expression profiles in a variety ofmouse models of human BC that compliment each other

reproducing various human BC entities. A few pecu-liarities of our methodology should be mentioned asmostly specific antibodies are used for galectin detec-tion, which allows identifying the protein but does notprovide information about its functional potency, i.e.ability to bind carbohydrate ligand. On the contrary, wehave investigated surface bGBPs/galectins functionallymeasuring their potency to bind the correspondingcarbohydrate ligands. In contrast to specific antibodies,the oligosaccharide-containing probes are group re-agents, i.e. they provide identifying of galectins as afamily of proteins independently of particular structure(galectin-1, galectin-2, etc.). In context of the currentmajor aim (to identify lectins for target therapy) weconsider this feature to be an advantage because wemeasured integral b-galactoside-binding potency. Thus,strong binding of used probes, especially aGM1-PAA-fluo to mammary tumour cells might imply surfacegalectin expression (functioning) in all mouse models ofhuman BC under investigation.

Firstly, established cell lines VMR-0 and VMR-Lwith long transplantation history and low and highmetastatic potential, respectively after being trans-planted to syngeneic A/Sn mice were screened in vitro.The VMR-L tumour cell line as potentially fast growingaggressive carcinoma with high metastatic spread to li-ver, lymph nodes, kidney, lungs, and spleen displayed

Figure 7. The BLRB spontaneous model, in vivo. (A) Histopatholog-

ical pattern of the advanced large tumour (L): prevailed solid D (1)

including small E (2) types of carcinoma with large cysts invaded fi-

brous stroma surrounded with small areas composed of the AD (3)

carcinoma, HE staining; (B) histopathological pattern of the small

tumour (S): cystic adenocarcinoma of the A (4) and AD (3) types with

delicate stroma, HE staining; (C) flow cytometric assay of the Glyc-

PAA-fluo binding to tumour cells from the primary large (L, white

columns) and small (S, grey columns) spontaneous mammary tumours

from the same BLRB female. Representative data from one of two

independent experiments are shown.

Figure 8. Lung metastases from the BLRB spontaneous model. (A)

Histopathological pattern of lung metastatic focus composed of cystic

mammary carcinoma; (B) flow cytometry analysis of the Glyc-PAA-

fluo binding to cells obtained from the lung with metastatic foci before

(white columns) and after (grey columns) permeabilization of cells with

saponine. Representative data from one of three independent experi-

ments are shown.


affinity to the LacNAc-PAA-fluo and the aGM1-PAA-fluo in vitro. On the contrary, VMR-0 tumour cell line,as potentially slowly growing polymorphic spindle cellcarcinoma with abundant malignant fibrous stroma,hardly expressed surface galectins in vitro. Furthermore,data in the i.p. VMR-L model showed a direct correla-tion of surface galectins and the proportion of tumourcells in an ascites, whereas normal leucocytes from bothtumour bearing and intact mice hardly expressed galec-tins. The VMR-L transplantable s.c. tumours in A/Snmice were characterized by macroscopic metastatic fociin the liver. To our surprise, no surface galectins weredetected on both metastatic liver cells and normalsplenocytes (from intact mice). But galectin ligandsbound strongly to only metastatic cells after permeabi-lization of cells with saponine. Hence, we proposed theintracellular localization of galectins in metastatic cells(in cytoplasm and/or nucleus). Cells from metastatic fociwere more aggressive than tumour cells from the primarytumour and they retained these higher proliferativecapacities during a next transplantation to a secondaryA/Sn host. Tumour cells from transplanted spleenmetastases exhibited faster proliferation in a secondaryA/Sn host and demonstrated higher level of affinity tothe LacNAc and the aGM1 galectin ligands than tumourcells derived from the transplanted primary tumour. Inconclusion, in all types of A/Sn mouse models the moreaggressive cell populations seem to exhibit higherexpression of bGBPs/galectins suggesting a direct cor-

relation between galectin overexpression and mammarytumour aggressiveness. This conclusion is in agreementwith the literature devoted to investigation of galectin-3expression on human BC cell lines in vitro and theirmetastatic potency in experimental immunodeficientmouse models in vivo [13,14]. Highly metastatic humanbreast carcinoma cells (MDA-MB-435) expressed ahigher level of galectin-3 than their non-metastaticcounterpart (MDA-MB-468) in vitro [13]. Galectin-3 en-hanced directly the metastatic potential of human breastcarcinoma BT549 in an experimental liver metastasismodel in vivo [14]. Interestingly, the analogy in galectinexpression profiles and metastatic potency to the liver ofboth human and murine VMR cell lines is apparent.

Histopathological research highlighted severaladvantages of the A/Sn mouse models reproducingvarious forms of human BC. Murine VMR-0 originatedspindle carcinoma resembles morphologically fibroma-tosis-like carcinoma of the human breast. Moreover,this murine tumour shows initially deceptively benignappearance with potentially aggressive behaviour atlatest stages exactly like it was reported for this malig-nancy in women [29]. On the other hand, murine VMR-L originated tumours with characteristic lymph nodemetastatic patterns remind of medullary human breastcarcinoma. Distinct morphological and functional sim-ilarities between murine and human carcinomas validatethe VMR-originated mouse models as proper tools tomimic certain aggressive forms of human BC.

Figure 9. The BLRB transplantable model. Representative histopathological patterns of mammary carcinomas of the first transplantation

generation derived: (A) from Large, L1 and (B) from small S1 spontaneous BLRB mammary carcinomas; (C) from lung metastases, M1. (D)

Flow cytometric analysis of the binding of the aGM1-PAA-fluo (grey columns) and LacNAc-PAA-fluo (white columns) to tumour cells obtained

from the first passage from large L1 primary spontaneous tumour, small S1 primary spontaneous tumour, and lung metastases, M1. In the box

cell cycles are shown. Representative data from one of four independent experiments are shown.


To reproduce the wider range of human malignancy,other mouse models were used, namely spontaneousmammary carcinomas in the BLRB mouse strain thatmight function as a model of familial BC [22]. Histo-pathological analysis showed that spontaneous BLRBmammary carcinomas are generally characterized by alower grade of malignancy and haematogenous meta-static spread to lungs compared with highly invasiveVMR-L originated tumours associated with lymph nodemetastases. The first stages of the BLRB mammarycarcinoma progression (A + AD) resemble in some waylobular carcinoma in situ of the human breast, but thefollowing stages (AD fi D fi E) of murine carcinomaprogression are closer to pure medullary carcinoma ofthe human breast associated commonly with conspicu-ous lymphocyte infiltration. Interestingly, both men-tioned human BC entities are relatively rare forms insporadic BC and were found more frequently in patientswith family history of BC [30]. This distinct morpho-logical similarity between murine and human carcino-mas is an argument to validate the BLRB strain asproper mouse model of familial BC. Furthermore, thepresence of a distinct proportion of the BLRB femaleswith a few primary nodules resembles multifocal BC andpermits to study the histopathology and lectin expres-sion during mammary tumour progression in one fe-male. Small slowly growing tumours (2–3 mm indiameter for a few weeks) as representatives of the initialphase of mammary tumourigenesis (A + AD) expressedmore surface bGBPs/galectins than large tumours thatwere used as representatives of consecutive invasivesteps of mammary carcinoma (types D + E) in thesame female. Both tumours represented histologicallyheterogeneous mammary carcinoma patterns. More-over, in large advanced tumour non-malignant fibrousstroma was conspicuous, which is a common charac-teristic of advanced tumours in the BLRB mouse strain.Hence, to prove galectin overexpression by the carci-noma component, short-term culture was establishedfrom the spontaneous BLRB mammary tumour andcells were tested by galectin ligands using flow cytome-try. Both LacNAc-PAA-fluo and aGM1-PAA-fluoprobes bound strongly to tumour cells from a culture;whereas selectin ligand (SiaLex-containing probe)demonstrated poor binding. This seems to be in dis-agreement with previously published data, where onlybound to tumour cells from the BLRB short-termculture [8]. The obvious differences in the assessment oflectin expression (aGM1-PAA-fluo probe was synthe-sized later) and in vitro technique here and used earlier[8], might explain this discrepancy.

To prove the reliability of our in vitro procedure andthe mammary carcinoma origin of cultured tumour cellswe transplanted these cells i.p. to a BLRB male. Indeed,tumour cells from the BLRB short-term culture formedtypical mammary carcinoma (A + AD + D) in vivo.Surprisingly, these mammary carcinoma cells metasta-sized to liver: this is uncommon behaviour for sponta-neous mammary carcinomas in the BLRB strain and

usual attribute for the VMR-L originated tumours withlong transplantation history in the A/Sn mice. Thisimplies that the invasive potency of the BLRB carci-noma cells has already been increased after one in vitropassage.

Interestingly, in both the BLRB and the A/Sn in vitromodels bGBP/galectins were less expressed on tumourcells than on neoplastic cells from related s.c. growingtumours. This might support a function of galectins incell adhesion to extracellular matrix in vivo. The level ofgalectin ligand binding to tumour cells from the VMR-Loriginated s.c. tumours in the A/Sn mice was in averageabout three times higher than to tumour cells from thegenerally less aggressive BLRB spontaneous mammarytumours (for both aGM1-PAA-fluo and LacNAc-PAA-fluo). The aGM1-containing probe bound in average1.5–2 times higher to tumour cells than LacNAc-con-taining probe in all mouse models used (Table 1).However, the binding of the aGM1-containing probewas 3.5–7 and 4.5–20 times higher than the values forthe Tbb- and SiaLex-containing probes, respectively.

As disadvantages of spontaneous mouse models arewell known, we tested first s.c. transplanted passages ofthe BLRB mammary carcinomas that originated fromboth primary tumours and their lung metastases. Alltumours expressed surface bGBP/galectins. However, itwas difficult to conclude whether galectin expression andtumour aggressiveness directly or reversely correlated inthe BLRB model, although the data were highly repro-ducible both in spontaneous and first transplantationpassage models. Smaller less aggressive mammary tu-mours (and their first transplanted passages) alwaysdemonstrated higher galectin levels than large advancedtumours (and their first transplanted passages). Mam-mary tumours acquired more aggressive carcinomacomponents (D + E) due to tumour progression andsimultaneously gradually attained abundant non-malignant fibrous stromal components. Estimating onlythe carcinoma constituent, one can conclude that sur-face galectin expression reversely correlated withaggressiveness of the carcinoma in the BLRB model.This is in accordance with conclusion in human BCstudies [15,16] that showed reduced expression ofgalectin-3 in more aggressive histopathological types ofhuman BC, namely in invasive ductal carcinoma com-pared with carcinoma in situ. Notably, non-invasiveforms of human BC also contain less stroma similarly toinitial stages (A + AD) of the murine mammary carci-noma progression. In contrast, both highly invasivecarcinomas of the human breast and aggressive conse-quent phases of spontaneous murine mammary carci-noma (D + E) induce abundant invading stroma. Andfirst transplanted passages from both primary BLRBtumours and their lung metastases resembling histolog-ically their spontaneous ancestors showed even moreimpressive stroma. Therefore, taking into account boththe carcinoma and the stromal part of the tumours, onecan suggest that in a sample from advanced aggressivemammary carcinoma lower galectin level was revealed


as tumour cells were simply ‘diluted’ with the non-malignant stromal components. Similar case was ob-served in the A/Sn i.p. model where individual ascitescontained only half of tumour cells showed less galectinsthan the ascites composed of only the tumour cells.Moreover, lower average bGBP/galectin expression inthe transplanted BLRB tumours that apparently con-tained more fibrous stroma than their spontaneousancestors seem to support this assumption. However,also another explanation is possible. In the light of ourfindings and the literature data, we hypothesized thatexpression of galectins on the surface of neoplastic cellsfrom naturally progressing mammary tumours might beenhanced through the initial phases and reduced duringadvanced stages of mammary carcinoma progression.Normal cells are characterized by nuclear galectinlocalization [16]. From the start of tumour growth sur-face galectin expression seems to be gradually aug-mented following the tumour progression due tointeractions of galectins with polylactosaminesexpressed on matrix proteins and cancer mucins. Thesemolecular interactions could mediate both heterotypicadhesion aiding cellular invasion and homotypic adhe-sion of carcinoma cells as well [10]. On the contrary, forthe advanced invasive mammary carcinoma clones withmetastasising phenotypes surface galectin expressionseems to be gradually diminished that permit tumourcells to escape from the primary localization. So, tumourcells from the endpoint of mammary carcinoma pro-gression might show no surface galectins at all. This wasreally the case in one experiment (data not shown),where a fast growing extremely necrotic and haemor-rhagic, poorly differentiated BLRB transplanted mam-mary carcinoma from lung metastases with scantystroma hardly expressed surface galectins. Furthermore,only neoplastic cells without measurable surface galec-tins seem to be capable to metastasise, as surface

bGBPs/galectins were not detected in the A/Sn liver andin the BLRB lung metastatic foci as well. To the best ofour knowledge selective galectin expression on the sur-face of tumour cells from the primary BLRB mammarycarcinomas and the presence of galectins only withinmetastatic cells were described here for the first time in aspontaneous non-transgenic mouse model of humanBC.

Finally, we conclude that galectin overexpressionhardly might be used as a marker of aggressiveness ofboth heterogeneous mouse and human BC as (i) themeasured galectin expression might be influenced by theamount of stromal component in a sample, and (ii)probably, galectins show inverse correlation withtumour aggressiveness during the initial and theadvanced stages of mammary tumour progression.

In conclusion, (i) Galectin expression (functioning)profiles were observed on mammary tumour cells in allused mouse models unlike non-malignant cells from theseveral murine tissues (intraperitoneal leucocytes,splenocytes, and cells from lungs); (ii) generally, thesurface galectin expression was diminished in the moreaggressive forms of the spontaneous BLRB mammarycarcinoma being reduced during the progression simi-larly to galectin-3 expression in human BC [15,16]; (iii)finally, our experimental system of complementarymouse models that showed several morphological andfunctional similarities to the distinct human BC entitiesappears to be a suitable tool to test novel bGBP/galectinaimed cytotoxic therapeutics against human BC.

Acknowledgments

This work was partly supported by the grants of Rus-sian Foundation for Basic Research No. 01-04-49253and No. 04-04-49689.

Table 1. The higher potential of the aGM1-PAA-fluo binding compared with the LacNAc-PAA-fluo binding to tumour cells from various mouse

models

Figure-model Columns Tested cells Ratio, a/b

1-A/Sn, in vitro White VMR-0 cell line 1.5

Grey VMR-L cell line 1.9

3B-VMR-L, i.p. transplantation Diagonally striated Ascites I 2.1

Grey Ascites II 1.8

4-A/Sn, liver metastasis from initial s.c. model Grey After permeabilization 1.4

5-A/Sn, secondary s.c. model White Tumours from primary tumour 2.1

Diagonally striated Tumours from spleen metastases 2.1

6B-BLRB, in vitro Grey Primary culture 3.6

7C-BLRB, primary spontaneous tumours Grey Small tumour S 1.1

White Large tumour L 1.2

8B–BLRB, Lung metastateses Grey After permeabilization 1.3

9D-Transplantable BLRB model Grey versus white Tumour L1 1.5

Tumour S1 2.0

Tumour M1 1.3

aThe level of aGM1-PAA-fluo binding to tumour cells.bThe level of LacNAc-PAA-fluo binding to tumour cells.


References

1. Morrow M, Gradishar W: Breast cancer. Br Med J 324: 410–414,

2002

2. Ogmundsdottir HM: Immune reaction to breast cancer: for better

or for worse?. Arch Immunol Ther Exp 49(suppl. 2), S75–S81,

2001

3. Pluncett TA, Correa I, Miles DW, Taylor-Papadimitriou J: Breast

cancer and the immune system: opportunities and pitfalls. J

Mammary Gland Biol Neoplasia 6: 467–475, 2001

4. Harris JR, Morrow M, Bonadonna G: Cancer of the breast. In:

De Vita VT, Helman S, Rosenberg SA (eds) Cancer. Principles and

Practice of Oncology. JB Lippincot, Philadelphia 1993, pp 1264–

1332

5. Dillman RO, Beutel LD, De Leon C, Nayak ShK: Short-term

tumour cell lines from breast cancer for use as autologous tumour

cell vaccines in the treatment of breast cancer. Cancer Biotherapy

Radiopharm 16: 205–211, 2001

6. Bovin NV: Polyacrylamide-based neoglycoconjugates as tools in

glycobiology. Glycoconj J 15: 431–446, 1998

7. Yamazaki N, Kojima S, Bovin NV, Andre S, Gabius S, Gabius

HJ: Endogenous lectins as targets for drug delivery. Adv Drug

Deliv Rev 43: 225–244, 2000

8. Vodovozova EL, Moiseeva EV, Grechko GK, Gayenko GP, Ni-

fant’ev NE, Bovin NV, Molotkovsky JG: Antitumor activity of

cytotoxic liposomes equipped with selectin ligand SiaLex, in mouse

mammary adenocarcinoma. Eur J Cancer 36: 942–949, 2000

9. Rabinovich GA: Galectins: an evolutionarily conserved family of

animal lectins with multifunctional properties; a trip from the gene

to clinical therapy. Cell Death Different 6: 711–721, 1999

10. Varki A, Cummings R, Esko J, Freeze H, Hart G and Marth J

(Eds) Essentials of Glycobiology, Consortium of Glycobiology

Editors. La Jolla, California, 1999

11. Dangue A, Camby I, Kiss R: Galectins and cancer. Biochim

Biophys Acta 1572: 285–293, 2002

12. Chiariotti L, Salvatore P, Frunzio R, Bruni CB: Galectin genes:

regulation of expression. Glycoconj J 19: 441–449, 2004

13. Khaldoyanidi SK, Glinsky VV, Sikora L, Glinskii AB, Mossine

VV, Quinn TP, Glinsky GV, Sriramarao P: MDA-MB-435 human

breast carcinoma cell homo- and heterotypic adhesion under flow

conditions is mediated in part by Thomsen-Friedenreich antigen-

galectin-3 interactions. J Biol Chem 278: 4127–4134, 2003

14. Song YK, Billiar TR, Lee YJ: Role of galectin-3 in breast cancer

metastasis: involvement of nitric oxide. Am J Pathol 160: 1069–

1075, 2002

15. Idikio H: Galectin-3 expression in human breast carcinoma:

correlation with cancer histological grade. Int J Oncol 12:

1287–1290, 1998

16. Vanden Brule F, Califice S, Castronovo V: Expression of galectins

in cancer: a critical review. Glycoconj J 19: 537–542, 2004

17. Kim H-RCh, Lin Hu-M, Biliran H, Raz A: Cell cycle arrest and

inhibition of anoikis by galectin-3 in human breast epithelial cells.

Cancer Res 59: 4148–4154, 1999

18. Paz A, Haklai R, Elad-Sfadia G, Ballan E, Kloog J: Galectin-1

binds oncogenic H-Ras to mediate Ras membrane anchorage and

cell transformation. Oncogene 20: 7486–7493, 2001

19. Nangia-Makker P, Honjo Y, Sarvis R, Akahani S, Hogan V, Pi-

enta KJ, Raz A: Galectin-3 induces endothelial cell morphogenesis

and angiogenesis. Am J Pathol 156: 899–909, 2000

20. Mouse Genome Informatics http://www.informatics.jax.org/

external/festing/mouse/docs/BLRB.shtml and http://www. infor-

matics.jax.org/external/festing/mouse/docs/CBRB.shtml

21. Moiseeva EV, Merkulova IB, Bijleveld C, Koten J-W,

Miroshnikov AI, Den Otter W: Therapeutic effect of single peri-

tumoural dose of IL-2 on transplanted murine breast cancer.

Cancer Immunol Immunother 52: 487–496, 2003

22. Moiseeva E, Chaadaeva A, Bojenko V, Mehdipour P, Den Otter

W: Mouse naturally arising mammary cancer: promising link to

familial set of breast cancer. In: Proceedings of the First Interna-

tional Congress on Cancer Genetics, Tehran University of Medical

Sciences, December 13–16 2003, Publisher of Deputy Research,

Tehran, 2003, p 102

23. Rapoport E, Khaidukov S, Baidina O, Bojenko V, Moiseeva E,

Pazynina G, Karsten U, Nifant’ev N, LePendu J, Bovin N:

Involvement of the Galbeta1 – 3GalNAcbeta structure in the

recognition of apoptotic bodies by THP-1 cells. Eur J Cell Biol 82:

295–302, 2003

24. Leffler H: Galectins structure and function – a synopsis in

Mammalian Carbohydrate recognition systems. In: Crocker Paul

R (eds) Mammalian Carbohydrate Recognition Systems. Springer-

Verlag, Berlin–Heidelberg 2001, pp 57–83

25. Korobko EV, Saschenko LP, Prockhorchouk EB, Korobko IV,

Gnuchev NV, Kiselev SL: Resistance to tumour necrosis factor

induced apoptosis in vitro correlates with high metastatic capacity

of cells. Immunol Lett 67: 71–76, 1999

26. Siebert HC, Andre S, Lu SY, Frank M, Kaltner H, van Kuik JA,

Korchagina EY, Bovin N, Tajkhorshid E, Kaptein R, Vliegenthart

JFG, von der Leith CW, Jimenez-Barbero J, Kopitz J, Gabius HJ:

Unique conformer selection of the human growth-regulatory lectin

galectin-1 for ganglioside GM1 versus bacterial toxins. Biochem-

istry 42: 14762–14773, 2003

27. Schmidt I, Uittenbogaart CH, Giorgi JV: A gentle fixation and

permeabilization method for combined cell surface and intracel-

lular staining with improved precision in DNA quantification.

Cytometry 12: 279–285, 1991

28. Squartini F, Pingitore R: Tumours of the mammary gland. In:

Turusov VA (ed) Pathology of Tumours in Laboratory Animals,

Lyon, 1994, Vol. 2, pp 84–92

29. Kinkor Z, Svitakova I, Ryska A, Kodet R, Hrabal P: Metaplastic

spindle-cell (fibromatosis-like) carcinoma of the breast-report of 4

cases. Cesk Patol 38: 164–168, 2002

30. Kriege M, Brekelmans CT, Boetes C, Besnard PE, Zonderland

HM, Obdeijn IM, Manoliu RA, Kok T, Peterse H, Tilanus-Lin-

thorst MM, Muller SH, Meijer S, Oosterwijk JC, Beex LV, Toll-

enaar RA, Koning HJde, Rutgers EJ, Klijn JG: Efficacy of MRI

and mammography for breast-cancer screening in women with a

familial or genetic predisposition. N Engl J Med 351: 427–437,

2004

Address for offprints and correspondence: Dr E.V. Moiseeva, Utrecht

University, Pathobiology, Yalelaan 1, 80185 Utrecht, Netherlands;

E-mail: [email protected]


Documents

Galectins as Markers of Aggressiveness of Mouse Mammary Carcinoma: Towards a Lectin Target Therapy of Human Breast Cancer