High levels of cellular retinol binding protein-1 expression in leiomyosarcoma: possible implications for diagnostic evaluation

Article

Reference

High levels of cellular retinol binding protein-1 expression in leiomyosarcoma: possible implications for diagnostic evaluation

ORLANDI, Augusto, et al.

Abstract

Retinoid bioavailability is regulated by the activity of specific cytoplasmic receptors. High levels of cellular retinol binding protein-1 (CRBP-1) have been documented during experimental arterial and wound-healing processes, but data concerning neoplastic smooth muscle tissues are scarce and/or controversial. This study reports that the expression of CRBP-1 is markedly higher in uterine and gastrointestinal leiomyosarcomas than in leiomyomas and normal myometrium; CRBP-1 was practically absent in normal gastrointestinal smooth muscle tissue. CRBP-1 positivity was particularly elevated in the epithelioid variant of leiomyosarcoma; it was associated with increased proliferative and apoptotic rates and inversely related to smooth muscle differentiation evaluated by alpha- and gamma-smooth muscle actin and desmin expression. Western blotting and reverse-transcription polymerase chain reaction confirmed the observations concerning CRBP-1 and actin isoform expression and revealed higher NF-kappa-Bp65 and RAR alpha and lower Bax protein levels in leiomyosarcoma than in the other conditions. These findings document that a high [...]

ORLANDI, Augusto, et al . High levels of cellular retinol binding protein-1 expression in

leiomyosarcoma: possible implications for diagnostic evaluation. Virchows Archiv , 2002, vol.

441, no. 1, p. 31-40

PMID : 12111198

DOI : 10.1007/s00428-001-0576-7

Available at:

http://archive-ouverte.unige.ch/unige:155773

Disclaimer: layout of this document may differ from the published version.

1 / 1

Abstract Retinoid bioavailability is regulated by the ac-

tivity of specific cytoplasmic receptors. High levels of cel- lular retinol binding protein-1 (CRBP-1) have been docu- mented during experimental arterial and wound-healing processes, but data concerning neoplastic smooth muscle tissues are scarce and/or controversial. This study reports that the expression of CRBP-1 is markedly higher in uter- ine and gastrointestinal leiomyosarcomas than in leiomyo- mas and normal myometrium; CRBP-1 was practically absent in normal gastrointestinal smooth muscle tissue.

CRBP-1 positivity was particularly elevated in the epithe- lioid variant of leiomyosarcoma; it was associated with in- creased proliferative and apoptotic rates and inversely re- lated to smooth muscle differentiation evaluated by

- and

-smooth muscle actin and desmin expression. Western blotting and reverse-transcription polymerase chain reac- tion confirmed the observations concerning CRBP-1 and actin isoform expression and revealed higher NF-

-Bp65 and RAR

and lower Bax protein levels in leiomyosarco- ma than in the other conditions. These findings document that a high CRBP-1 expression is associated with smooth muscle malignancy and suggest that CRBP-1 expression represents a new useful marker for the classification of unusual variants of smooth muscle tumors.

Keywords Actin isoforms · Myometrium · Retinoic acid ·

Smooth muscle

Introduction

Retinol and retinoid derivatives participate in a wide range of biological processes and play a significant role

in the control of cell growth and differentiation of mesenchymally derived tissues [3, 6]. Trans-retinoic acid (RA), the principal functional retinoid [20, 48], de- rives at least in part from the cytosolic bioconversion of retinol [26]. Bioavailability of retinol is regulated by the presence of specific cytoplasmic receptors; among these, cellular-retinol binding proteins 1 and 2 (CRBP-1 and -2) are members of the fatty acid-binding proteins (FABP)/CRBP family and have distinct physiologic roles [3, 27]. In the endometrium, epithelial and stromal cells express significant levels of mRNA for CRBP-1 through- out the menstrual cycle, whereas levels of CRBP-2 vary and mediate the menstrual cycle effects of ovarian ste- roids [23, 38]. CRBP-1 expression probably reflects the cell requirement for retinol and its active metabolites [5, 27]. The role of retinoids and their receptors has been extensively investigated in smooth muscle (SM) tissue of the arterial district [30]. During experimental arterial healing, intimal smooth muscle cells (SMCs) show tran- sient increased proliferative and apoptotic rates associat- ed with morphologic changes and decreased levels of

-smooth muscle actin (

-SM actin) [4, 28]. At the same time, many intimal SMCs express CRBP-1, where- as quiescent medial SMCs do not [28, 29]. A similar in- creased expression of CRBP-1 has been observed in myofibroblasts during experimental wound healing [45].

Retinoids influence growth and differentiation of intimal but not of normal media SMCs in vitro and inhibit inti- mal thickening after injury in vivo [29]. These data sug- gest that, at least under certain pathophysiological condi- tions, CRBP-1 expression represents a crucial step in the regulation of retinoid-related biological activities.

Experimental data concerning the role of retinoids and their receptors in other normal as well as neoplastic SM tissue are scarce and apparently controversial. Retinoids variably influence the growth of myogenic derived tumor cell lines [14], and no difference is observed comparing the action of retinol on proliferation and morphology of myometrial and leiomyomatous SM cells in vitro [7]. To document the expression of CRBP-1 and its possible rela- tionship with cell morphology, differentiation, prolifera-

Institute of Anatomic Pathology, Tor Vergata University of Rome, Italy S. Clément · P. Ropraz · G. Gabbiani (

✉

Department of Pathology-CMU, University of Geneva, 1 Rue Michel-Servet, 1211 Geneva 4, Switzerland e-mail: [email protected] Tel.: +41-22-7025742, Fax: +41-22-7025746 Virchows Arch (2002) 441:31–40

DOI 10.1007/s00428-001-0576-7 O R I G I N A L A R T I C L E

Augusto Orlandi · Arianna Francesconi Sophie Clément · Patricia Ropraz Luigi Giusto Spagnoli · Giulio Gabbiani

High levels of cellular retinol binding protein-1 expression

in leiomyosarcoma: possible implications for diagnostic evaluation

Received: 29 August 2001 / Accepted: 22 October 2001 / Published online: 31 January 2002

© Springer-Verlag 2002

tion, and apoptosis, we investigated, using morphologi- cal, immunohistochemical, and biomolecular methods, 36 uterine and gastrointestinal leiomyomas and leiomyosar- comas recruited over the past 5 years at the Institute of Anatomic Pathology, Tor Vergata University of Rome, It- aly. Differences in the CRBP-1 level under our conditions are discussed in relation to the possibility of predicting (1) the biological behavior of SM-derived tumors and (2) their potential response to retinoid therapy.

Materials and methods

Tissue samples

Thirty-six uterine and gastrointestinal SM-derived tumors, includ- ing typical leiomyomas (LM, n=10), cellular (n=3), symplastic (n=3), epithelioid (n=4), myxoid LM (n=3), borderline tumors (n=2), and typical (n=8) and epithelioid leiomyosarcomas (LMS, n=3), were obtained from surgical pathology material over the last 5 years at the Institute of Anatomic Pathology, Tor Vergata Uni- versity of Rome, Italy (Table 1). Common diagnostic criteria of malignancy for uterine SM tumors were: ten or more mitotic fig- ures per ten high-power fields (HPF), hypercellularity, tumor cell necrosis, pleomorphism, and hyperchromasia [36]. Gastrointesti- nal tumor criteria included large tumor size (<5 cm), mitotic activ- ity (>5 per 10 HPF) and a negative immunoreaction for CD117 (c-kit), CD34, and S-100 protein to differentiate them from stro- mal tumors [36]. LMS, graded according to Coindre et al. [13], were classified as two of grade 1, five of grade 2 and four of grade 3. Tumors were compared with normal uterine (n=7) and gastroin- testinal (n=7) SM tissues from non-neoplastic surgical pathologic specimens collected within the same period. Some freshly excised samples were prepared for histology, immunohistochemistry, and electron microscopy; others were frozen in isopentane cooled in liquid nitrogen (Table 1) and stored in pools at –80°C for protein and mRNA extraction.

Histology and immunohistochemistry

Tissue samples were fixed in 10% neutral buffered formalin, em- bedded in paraffin and 3- to 4-µm-thick serial sections were stained with hematoxylin and eosin and Masson trichrome. For immunohistochemistry, after deparaffinization, blocking of endog- enous peroxidase activity with 0.2% H₂O₂(20 min), and incuba- tion with normal goat serum (30 min), serial sections were reacted for 1 h with monoclonal anti α-SM actin, anti-desmin, Ki67, and CD34 (Dako) and with polyclonal anti-CRBP-1 [28], CD 117 (c-kit, Santa Cruz), and S-100 protein (Dako) antibodies. Before incubation with CRBP-1 and c-kit antibodies, heat-mediated anti- gen retrieval in a solution of 10 mM sodium citrate buffer (pH 6.0) in a microwave oven (three cycles of 5 min) was performed. Pre- liminary studies have demonstrated that the CRBP-1 antibody spe- cifically reacts with normal human SM and non-SM tissues, simi- lar to what is observed in the rat (data not shown). Diaminobenzi- dine (DAB) was used as the final chromogen. All immunohisto- chemical procedures were performed at room temperature.

Semi-quantitative CRBP-1 and α-SM actin immunoreaction was estimated at ×200 magnification by two different researchers who used a grading system in arbitrary units as follows: negative (–=0), focal (±=0.5), weak (+=1), moderate (++=2), and diffuse positive (+++=3) staining. The inter-observer reproducibility was greater than 95%. For each case, the ratio of the resulting score with the total number of fields analyzed was calculated.

Proliferation and apoptosis

Ki67 positive nuclear staining identified proliferating cells in nor- mal and neoplastic tissues. The positive/negative nuclear ratio was

calculated at ×200 magnification using a Quantimet 920 image an- alyzer (Cambridge Instruments) connected to a Polyvar micro- scope (Reichert Jung) by a HC3077 camera (Hamamatsu). The number of fields required to obtain a significant difference was calculated according to a stereological formula [44]. Two different researchers evaluated a minimum of 12 random fields and 2 serial sections, with inter-variability less that 5%.

To evaluate apoptosis, rehydrated sections were stripped from proteins by incubation with 300 U/ml proteinase K (Sigma) for 15 min at 37°C and endogenous peroxidase blocked with 0.1%

H₂O₂in methanol (20 min) at room temperature. Apoptotic nuclei were revealed by means of terminal deoxynucleotidyl transferase- mediated dUTP-biotin nick end-labeling (TUNEL), according to the method described by Gavrieli et al. [17]. Apoptotic differences were quantified by calculating the positive/negative cell ratio, as reported above.

To verify the myocytic nature and CRBP-1 expression of Ki67 positive and apoptotic cells, a modified double immmunohisto- chemical reaction was performed on serial sections, as previously reported [40], with minor modifications. A modified DAB by NaCl₂precipitation was used as chromogen for the anti-Ki67 and TUNEL, followed by the second immunoreaction for α-SM actin and CRBP-1 using a Vector VIP Substrate Kit for peroxidase (Vector), which produces a purple reaction product.

Ultrastructural study

For transmission electron microscopy, small tissue samples were post-fixed in 1% OsO₄for 2 h and dehydrated through an alcohol series and propylene oxide before embedding in EPON 812. Thin sections were stained with toluidine blue. Ultrathin sections were cut using an 8800 ultramicrotome III (LKB), counterstained with uranyl acetate and lead citrate and studied under a Philips 301 electron microscope.

Sodium dodecyl sulfate–polyacrylamide gel electrophoresis and immunoblotting

For sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), tissue samples from different pools were finely fragmented in the presence of liquid nitrogen and mixed in buffer according to Laemmli [22], immediately sonicated, boiled for 3 min, and centrifuged. Protein content was determined according to the method described by Bradford [8]. All procedures were re- peated three times. Forty micrograms of protein was electrophore- sed on a 5–20% gradient gel and stained with Coomassie Brilliant Blue (R 250, Fluka) in methanol, acetic acid, and water (45:10:45). For Western blotting [41], 5–60 µg protein was electrophoresed on a 5–20% gradient gel. Separated proteins were transferred to nitrocellulose filters (0.45 mm, Schleicher &

Schuell) which were alternatively incubated with anti-α-SM actin or anti-desmin (1:500) followed by a goat anti-mouse immuno- globulin G (IgG, 1:10⁵), anti-CRBP-1 (1:100), α, βand γisotypes of RARs (Santa Cruz, 1:500) and Bax protein (Santa Cruz, 1:100), followed by a goat anti-rabbit IgG (1:10⁵) or Bcl-2, (Santa Cruz, 1:100), and NFκBp65 (Santa Cruz, 1:200) followed by a donkey anti-goat IgG (1:10⁵). Enhanced chemiluminescence was used for detection (Amersham). Quantification of X-Omat R films was per- formed as previously reported [32].

RNA extraction and reverse-transcription polymerase chain reaction

Tissue samples were finely homogenized for 60 s using a T25 Ultra-Turrax (Janke & Kunkel) and total RNAs isolated using Tri-Reagent (Molecular Research Center, Cincinnati, Ohio.), ac- cording to the manufacturer’s instruction. Each RNA sample was quantified spectrophotometrically at 260 nm. Reverse-transcrip- tion polymerase chain reaction (RT-PCR) was performed as previ- ously reported [12, 28] using the following specific primers syn-

thesized (Invitrogen, Switzerland) based on published sequence data:

1. CRBP-1 sense 5′-GGAAGATGTTGGTCAACGAG-3′

-Antisense 5′-GTGGAAGGTGTGGTCTGCAA-3′

2. α-SM actin sense 5′-AGAACGCAAATATTCTGTCTG-3′ -Antisense 5′-TAACCAGTAGCCTATTTCAGA-3′ 3. γ-SM actin sense 5′-AGAGCGGAAGTACTCAGTCTG-3′

-Antisense 5′-CATGACTGGTAACAGAGTAGT-3′

4. β2-microglobulin sense 5′-ATCTTTCTGGTGCTTGTCTC-3′

-Antisense 5′-AGTGTGAGCCAGGATGTAGT-3′

PCR products were analyzed by means of electrophoresis on 2.5%

agarose gels using 1 µg pGEM DNA (Promega) as marker. The amount of amplified products was analyzed by densitometric read- ing in triplicate, and the final amount of PCR was expressed as the ratio of each gene amplified to that of β2-microglobulin [12].

Statistical analysis

Statistical analysis was performed using an SPSS program (Statis- tical Package for the Social Sciences, 4th edn., MJ Norusis/SPSS 33 Table 1 Clinical data, immunohistochemical evaluation and ex-

perimental procedures. – negative, +/– focally positive, + weakly positive, ++ moderately positive, +++ diffusely positive. Methods

used: 1 histology and immunohistochemistry, 2 proliferation and apoptosis, 3 electron microscopy, 4 Western blotting, 5 reverse- transcription polymerase chain reaction

Histology Age Sex Primary Cellular retinol α-Smooth Desmin Methods

(years) site binding protein-1 muscle actin

Nontumor tissue 46 Female Uterus +/– +++ +++ 1, 2

50 Female Uterus +/– +++ +++ 1, 2, 4, 5

48 Female Uterus – ++ +++ 1, 2

31 Female Stomach – +++ ++ 1, 2

41 Male Bowel – +++ +++ 1, 2, 4

55 Female Bowel – +++ +++ 1, 2, 4

43 Female Uterus +/– ++ +++ 1, 2, 4

48 Female Uterus +/– +++ +++ 1, 2, 4, 5

51 Male Stomach – +++ +++ 1, 2

51 Female Uterus – +++ ++ 1, 2

65 Male Bowel +/– ++ +++ 1, 2, 4

48 Female Bowel – ++ +++ 1, 2, 4

46 Female Uterus +/– +++ +++ 1, 2, 4, 5

69 Male Bowel – +++ +++ 1, 2

Leiomyoma 51 Female Uterus^d + +++ ++ 1

51 Female Uterus^a +/– +++ +++ 1, 2, 4

49 Female Uterus^a +/– +++ ++ 1, 3, 4, 5

61 Male Bowel^a – +++ ++ 1, 2

39 Female Stomach^a +/– ++ + 1, 2

45 Female Uterus^d + ++ ++ 1

57 Female Uterus^a +/– ++ ++ 1, 4, 5

42 Female Uterus^a +/– +++ ++ 1, 2

65 Female Bowel^a – ++ ++ 1, 2, 4

49 Female Uterus^b +/– ++ +++ 1, 2

48 Female Uterus^a +/– +++ +++ 1, 2, 3, 4, 5

43 Female Uterus^d +/– ++ ++ 1, 2, 4, 5

38 Female Uterus^f +/– ++ ++ 1

41 Female Bowel^d + ++ + 1, 2

36 Female Uterus^f + ++ ++ 1

54 Female Uterus^b +/– ++ ++ 1

55 Female Uterus^c + ++ ++ 1

48 Female Uterus^f +/– ++ ++ 1

42 Male Bowel^a +/– ++ + 1, 4

44 Female Uterus^b +/– +++ ++ 1

53 Female Uterus^c +/– ++ ++ 1

49 Female Uterus^c + ++ ++ 1

51 Male Bowel^a +/– ++ ++ 1, 3, 4

43 Female Uterus^e +/– ++ ++ 1

25 Female Uterus^e +/– +++ ++ 1, 2

Leiomyosarcoma 55 Female Uterus^d +++ +/– +/– 1, 2, 4

48 Female Uterus^a ++ ++ + 1, 2, 3, 4, 5

51 Female Uterus^d ++ +/– + 1, 2, 4, 5

62 Female Bowel^a +++ +/– +/– 1, 2

49 Female Bowel^a ++ + ++ 1, 2

64 Female Bowel^a ++ + + 1, 2, 4

61 Male Bowel^a ++ ++ ++ 1, 2, 4

36 Female Uterus^a ++ +/– + 1, 2, 4

49 Female Uterus^a ++ +/– + 1, 2

56 Male Bowel^a ++ +/– ++ 1, 2, 3, 4

58 Female Uterus^d ++ +/– + 1, 2, 3, 4, 5

aTypical ^bCellular ^cSymplastic ^dEpithelioid ^eBordeline tumor ^fMyxoid

Inc). For each parameter, the mean, standard error of the mean, and range were calculated. Differences were evaluated using t-tests and non-parametric Mann-Whitney tests; values of P<0.05 were considered statistically significant.

Results

Nontumor tissue

Clinical and immunohistochemical data are listed in Table 1. The mean age of patients was 49.4±2.5 years.

Normal gastrointestinal SM tissue showed an intense

sent CRBP-1 immunostaining. Endometrial stromal and epithelial cells appeared strongly positive for CRBP-1 and negative for α-SM actin immunostaining (Fig. 1). In 80% of the cases, examination of myometrium revealed a slight CRBP-1 positivity with sparse and scattered elongated positive cells (Fig. 1). Semi-quantitative eval- uation of uterine and gastrointestinal nontumor SM tis- sues confirmed different CRBP-1 (P<0.03) and similar

Leiomyoma

The mean age of patients with LM was 47.2±1.7 years.

The histological appearance of leiomyomatous tissue was characterized by the presence of long and spindle- shaped cells, arranged in rather well-oriented streaming or interweaving bundles embedded in a collagenous stroma. The nuclei generally exhibited characteristically blunt or rounded ends, with no atypia or mitotic activity, and eosinophilic cytoplasms (Fig. 1). An immunohisto- chemical examination showed that LM were strongly positive for

-SM actin and desmin, weakly positive or negative for CRBP-1, and negative for c-kit, S-100 and CD34. Similar to myometrium, uterine LM showed sparse, scattered elongated CRBP-1 positive cells (Fig. 1). Semi-quantitative evaluation (Fig. 2) demon- strated a significant increase of CRBP-1 immunoreactiv- ity of gastrointestinal LM compared with nontumor tis-

sue (P<0.02) without differences compared with uterine LM. Instead,

-SM actin immunoreactivity was similar to that of nontumor tissues. Epithelioid LM showed round or polygonal cells with abundant eosinophilic cytoplasm and moderately atypical nuclei. Symplastic LM were characterized by the presence of pleomorphic or giant cells with angular nuclei, very dark chromatin and abundant eosinophilic cytoplasm, and practically ab- sent mitosis (Fig. 3).

Uterine borderline tumors differed from typical LM due to the presence of low-grade cell atypia and mitotic rate (less than five mitoses for 10 HPF). No further im- munohistochemical differences were observed compar- ing borderline tumors and different LM types, including the presence of scattered, single CRBP-1 positive cells as well as

-SM actin expression (Fig. 2).

Leiomyosarcoma

The mean age of patients with LMS was 53.5±2.4 years.

Histologically, typical uterine and gastrointestinal LMS were composed by slender or slightly plump cells ar- ranged in fascicles of varying sizes. Hypercellular areas, atypia, abundant mitotic figures, necrosis and infiltrative edges were evident (Fig. 3). Cells appeared elongated and blunt-ended, sometimes with large and hyperchro- matic nuclei. Ultrastructural investigation confirmed dif- ferences to LM. In general, LMS showed a reduced in- tercellular matrix relative to LM. Epithelioid LMS were characterized by predominantly round or polygonal cells, which tended to be arranged in sheets and lobules simu- lating an epithelial pattern, with clear cytoplasms, pleo- morphic nuclei, and prominent nucleoli. In some cases,

Fig. 1 Cellular retinol binding protein-1 (CRBP-1) and α-smooth muscle (SM) actin immunoreactivity in nontumor SM tissue and leiomyoma (LM). A strong CRBP-1 positive reaction of stromal and epithelial cells is observed in early proliferative (b), secretive (g), cystic atrophic endometrium (i) and adenomyosis (n) but not in underlying or adjacent myometrium; α-SM actin immunoreac- tion is localized in myometrium (c, h). Immunostaining with anti- CRBP-1 antibody shows gastric positive epithelium and negative muscularis mucosae (d); the latter stains intensely for α-SM actin (e). Myometrial cells exhibit a slight diffuse immunostaining for CRBP-1 with rare positive cells (f); serial sections of uterine LM showing morphological aspect (j), a slight CRBP-1 positivity with rare strongly positive cells (k) and intense α-SM actin immunore- action (l); a rare Ki67 positive LM cell (m); CRBP-1 immuno- staining of cirrhotic liver as control (o) shows markedly positive Ito cells and slightly positive or negative hepatocytes. Original magnifications: a–c, g, h, i, n×100; d, e×125; f×160; j–m and o

×200

Fig. 2 Bar graphs represent mean differences (±SEM) of cellular retinol binding protein-1 (CRBP-1) and α-smooth muscle (SM) actin expression in gastrointestinal and uterine normal SM tissues, leiomyoma (LM) and leiomyosarcoma (LMS). Amounts of CRBP-1 and α-SM actin were evaluated by means of immunohis- tochemistry on serial sections with a grading system in arbitrary units as follows: absent 0; focal 0.5; weak 1; moderate 2; and dif- fuse positive staining 3. For each case, the ratio of the resulting score over the total number of fields analyzed was calculated

▲

epithelioid and sparse typical spindle-shaped areas were adjacent. LMS were negative for c-kit, S-100, and CD34 immunoreactions. Typical LMS areas showed an intense and diffuse positive CRBP-1 staining and a reduced

-SM actin and desmin expression. Staining of serial sections confirmed that LMS areas with importantly re- duced levels of

-SM actin were characterized by high CRBP-1 expression (Fig. 3). Semi-quantitative evalua- tion of both districts (Fig. 2) confirmed the increased CRBP-1 and the decreased

-SM actin expression rela- tive to LM (P<0.01). In some cases, almost negative

-SM actin LMS were slightly positive for desmin and

cle (SM) phenotypic marker immunostainings of leiomyoma (LM) variant and leiomyosarcoma (LMS). Hematoxylin-eosin staining of symplastic LM (a) shows large bizarre, sometimes multinucleated, cells but no necrosis and mitosis, rare CRBP-1 (b) and diffuse α-SM actin (c) positive cells. Serial sections of LMS tissue (d–f) with nu- merous, strongly CRBP-1 positive cells and an adjacent negative tu- nica media of a small artery (d), rare α-SM actin positive neoplastic cells with strong vascular staining (e) and numerous desmin-positive neoplastic cells (f); at low magnification, infiltrative edge of intesti- nal LMS (g) strongly CRBP-1 positive (h), α-SM actin negative (i) and desmin-positive (j); epithelioid LMS (k) diffusely and strongly positive for CRBP-1 and (l) practically negative for α-SM actin.

Original magnifications: a–f, k, l×200; g–j×180

strongly positive for CRBP-1 (Fig. 3). Epithelioid LMS showed the highest expression of CRBP-1 (Fig. 3), al- though the semi-quantitative evaluation did not show a significant difference relative to typical LMS, possibly a consequence of the limited number of cases (data not shown).

Proliferation and apoptosis

Proliferating cells were identified by Ki67 immunostain- ing on serial sections. In myometrium, the Ki67 posi- tive/negative cell ratio was higher than in gastrointestinal SM tissue (P<0.01, Fig. 4). The ratio did not differ be- tween uterine and gastrointestinal LM. Uterine and intes- tinal LMS showed a strong increase of positive/negative cell ratio (P<0.01) but without significant differences be- tween the two districts.

As reported in Fig. 4, apoptotic cells detected using TUNEL were more numerous in myometrium than nor- mal gastrointestinal tissue (P<0.01). Gastrointestinal LM also showed a reduced TUNEL positive/negative cell ra- tio relative to the uterine counterpart (P<0.01). Apoptot- ic rate strongly increased in LMS (P<0.01), but more in the uterine than gastrointestinal district (P<0.02). No sig- nificant differences were observed comparing Ki67 or TUNEL positive/negative cell ratio of typical and epithe- lioid LMS (data not shown). Double immunostaining with

-SM actin confirmed that the majority of Ki67 or TUNEL positive cells were of myocitic origin. Many Ki67 positive cells were also positive for CRBP-1, whereas the majority of TUNEL positive cells were CRBP-1 negative.

Western blotting and RT-PCR analysis

As reported in Fig. 5, densitometric analysis of Western blotting confirmed the strong CRBP-1 expression in LMS relative to LM and nontumor tissues (P<0.01). The opposite was true for

-SM actin (P<0.01). In gastroin- testinal nontumor SM tissue, the CRBP-1 level was ex- tremely low and less than that of LM of the same district (P<0.01), in the absence of a difference of

-SM actin expression. Increased levels of CRBP in LMS were parallel to those of RAR

, whereas RAR

and

levels did not differ relative to LM. Analysis of RNA expres- sion using RT-PCR corroborated the results obtained by means of immunochemical and Western blotting investi- gations of protein expression. Densitometric analysis re-

Fig. 4 Bar graphs represent differences (±SEM) of cell prolifera- tion and apoptosis in uterine and gastrointestinal normal smooth muscle (SM) tissue, leiomyoma (LM) and leiomyosarcoma (LMS). Proliferating cells were calculated as the Ki67 and TUNEL positive/negative cell ratio on serial sections, respective- ly. LMS of both districts show higher proliferative and apoptotic values (P<0.01) than LM and nontumor tissues

Fig. 5 (a) Cytoskeletal proteins, cellular retinol binding protein-1 (CRBP-1), proliferation-related and apoptosis-signaling molecules examined by Western blotting. Densitometric analysis (b) reveals that α-smooth muscle (SM) actin expression is lower in leiomyo- sarcoma (LMS, lanes 3, 6) than in leiomyoma (LM, lanes 2, 5) and nontumor tissues (lanes 1, 4) of uterine (lanes 1–3) and gas- trointestinal (lanes 4–6) districts. The opposite is true for CRBP-1 (c); normal gastrointestinal SM tissue shows a very low CRBP-1 expression (lane 4)

vealed that PCR products of CRBP-1 in uterine LMS were threefold higher than those obtained from normal and LM tissues, whereas the opposite was true for

- and

-SM actin isoforms (P<0.01, Fig. 6).

In order to investigate the expression of prolifera- tion and apoptosis-related proteins, we also performed Western blotting analysis of Bcl-2, Bax and NF-

-Bp65 (Fig. 5). Nontumor tissues showed a low expression of Bcl-2; the latter was increased in LM (P<0.02). Bcl-2 expression was somehow lower in uterine but not in gas- trointestinal LMS relative to LM. Densitometric scan- ning showed that NF-

-Bp65 expression was increased in myometrium relative to gastrointestinal nontumor tis- sue (P<0.02) and in LMS relative to LM and nontumor tissues of both districts (P<0.01). Densitometric analysis also demonstrated an increased bax expression in LMS relative to LM tissues (P<0.01).

Discussion

The use of cytoskeletal markers such as

-SM actin, des- min and SM-myosin heavy chains has been proven to be useful in characterizing histopathological features of SM-derived tumors [16, 25, 37]. In general, malignancy is associated with a reduced expression of these markers, and this is confirmed by the present results, although it has been reported that differentiation, cell growth and apoptosis in LMS are not necessarily linked [10, 11, 42].

Here we report that uterine and gastrointestinal LMS ex- press high levels of CRBP-1, whereas this expression is weak in LM and nontumor SM tissue. CRBP-1 levels are also low in symplastic LM and borderline tumors. These findings indicate that CRPB-1 represents an additional phenotypic marker for the screening and classification of SM-derived tumors in which nuclear pleomorphism

and/or increased mitotic activity suggest malignancy, such as epithelioid or symplastic types [35]. The rela- tively low frequency of these LM variants may cause overestimation of nuclear atypia, especially when epithe- lioid areas are encountered in otherwise typical or cellu- lar areas. As far as we know, this is the first systematic study of the relationships between the expression of CRBP-1 and differentiative, proliferative, and apoptotic features of SM tumors. In particular, CRBP-1 expression is inversely related to SM differentiation evaluated by

chemical results show that these findings are shared by LMS of other districts (Orlandi et al., unpublished re- sults). Arterial SMCs cultured from experimental rat neointima exhibit high levels of CRBP-1, an epithelioid and dedifferentiated phenotype, and increased prolifera- tive activity relative to normal media SMCs [32]. In this respect it is worth noting that CRBP-1 expression was higher in epithelioid than in typical LMS.

Presently, the biological role of CRBP-1 in SM- derived tumor cells is not established. Retinoids contri- bute to the growth and differentiation of many tissues [3, 48]. Bioavailability of retinol is regulated by the pres- ence of specific receptors. Among these, CRBP-1 pro- vides the substrate for RA biosynthesis [5, 34] and medi- ates hormonal effects in myometrium [5]. Mesenchymal cells appear to be, in some cases, more responsive to ret- inol that to RA [1] and, in other cases, e.g., arterial SMCs, more responsive to RA in terms of proliferation and differentiation [29]. RA induces a TGF

-indepen- dent increase of CRBP-1 expression in epithelioid inti- mal SMCs as well as in fibroblasts in vitro [29, 46]. RA influences the expression of many genes through interac- tion between RA receptors and response elements (RARE) located in promoter regions [19]. It is conceiv- able that the accumulation of CRBP-1 favors the conver- sion of retinol into RA and its biological effects through RARE induction. Indeed, high levels of CRBP-1 in LMS are associated with increased expression of RAR

but not of

and

signal gene transducers. RAR

mediates RA effects in CRBP-1-expressing SMCs in vitro [29].

The role of RAR

as nuclear transducer of retinoic ef- fects appears tissue specific, since the correlation be- tween RA-induced growth inhibition and changes of the expression of RA nuclear receptors differs in other neo- plastic derived cell lines [15, 18]. Similarly, DNA syn- thesis inhibition is observed in myometrial and stromal endometrial cells but not in adjacent epithelium [6].

Experimental data suggest that progesterone may act as a promoter for normal and tumoral myometrium growth, whereas estrogens have opposite effects [47]. In the rat uterus, a short-term treatment with RA markedly reduces beta-estradiol-induced DNA synthesis in stromal and myometrial cells [6]. It is noteworthy that systemic administration of RA significantly reduces arterial inti- mal thickening after endothelial injury in vivo [24, 29]

and in vitro induces apoptosis of CRBP-1 expressing in- timal cells but not of normal media SMCs [33]. Our re- sults support the possibility that the expression of

β2-microglobulin (β2-M), α- and γ-smooth muscle (SM) actin transcripts using reverse-transcription polymerase chain reaction.

Myometrial (lane 1) and leiomyomatous (lane 2) tissues contain low amounts of CRBP-1 and high levels of α- and γ-SM actin iso- form transcripts, whereas the opposite is observed in uterine leio- myosarcoma (lane 3); lane M pGEM DNA marker

CRBP-1 represents a target for pharmacological strate- gies aimed at influencing LMS growth through the con- trol of RA availability. It is well known that the efficacy of hormonal therapy for the treatment of breast cancer depends on the expression by neoplastic cells of steroid receptors [9]. In vitro retinoids inhibit proliferation and induce apoptosis in several soft tissue sarcoma cell lines [14, 15, 43]. Variable levels of CRBP-1 expression may also explain the different sensitivity of myogenic tumor cell lines to retinoid administration [14]. Our results show that CRBP-1 overexpression is associated with changes of proliferation and apoptosis-signaling mole- cules. Bcl-2 was expressed more in LM and LMS than in nontumor tissues, confirming that tumors of muscle ori- gin preferentially express Bcl-2 [39]. Changes in Bcl-2 were associated with high levels of Bax protein in LMS but not in LM. Bax protein when overexpressed accelerates cytokine-induced apoptosis [31]. Thus, the increased ratio of Bax to Bcl-2 in LMS compared with LM depends probably on the overexpression of Bax. We also observed an increased NF-

-Bp65 expres- sion in LMS. NF-

-B is composed of two subunits, p50 and p65 [2]. Antisense p65 oligonucleotides inhibit tumorigenicity in mice [21], suggesting a main role of NF

Bp65 in the control of growth progression of SM-derived tumors.

In conclusion, we report an increased expression of CRBP-1 in LMS relative to LM and normal SM tissues.

This increase is associated with reduced expression of

- and

-SM actin and increased proliferative and apo- ptotic rates. Screening for CRBP-1 level may help to ob- tain a more precise diagnosis.

Acknowledgments We thank Sabrina Cappelli, Alfredo Colantoni, Renzo Bernabei, and Angela Ortenzi for their excellent technical work. This study was partially supported by a UICC International Cancer Transfer Fellowship (no. 481/2001) and the Swiss National Science Foundation (grant no. 31-61336.00).

References

1. Achkar CC, Derguini F, Blumberg B, Langston A, Levin AA, Speck J, Evans RM, Bolado J, Jr, Nakanishi K, Buck J, et al (1996) 4-Oxoretinol, a new natural ligand and transactivator of the retinoic acid receptors. Proc Natl Acad Sci USA 93:4879–

4884

2. Baeuerle PA, Baltimore D (1996) NF-kappa B: ten years after.

Cell 87:13–20

3. Blomhoff R (1994) Transport and metabolism of vitamin A.

Nutr Rev 52:S13–S23

4. Bochaton-Piallat ML, Gabbiani F, Redard M, Desmouliere A, Gabbiani G (1995) Apoptosis participates in cellularity regula- tion during rat aortic intimal thickening. Am J Pathol 146:

1059–1064

5. Boerman MH, Napoli JL (1996) Cellular retinol-binding pro- tein-supported retinoic acid synthesis. Relative roles of mi- crosomes and cytosol. J Biol Chem 271:5610–5616

6. Boettger Tong HL, Stancel GM (1995) Retinoic acid inhibits estrogen-induced uterine stromal and myometrial cell prolifer- ation. Endocrinology 136:2975–2983

7. Boettger-Tong H, Shipley G, Hsu C-J, Stancel GM (1997) Cultured human uterine smooth muscle cells are retinoid re- sponsive. Proc Soc Exp Biol Med 215:59–65

8. Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72:248–254 9. Brodowicz T, Wiltschke C, Kandioler-Eckersberger D, Grunt

TW, Rudas M, Schneider SM, Hejna M, Budinsky A, Zielinski CC (1999) Inhibition of proliferation and induction of apopto- sis in soft tissue sarcoma cells by interferon-alpha and re- tinoids. Br J Cancer 80:1350–1358

10. Brooks JJ (1997) Disorders of soft tissue. In: Sternberg SSR (ed) Diagnostic surgical pathology. Raven Press, New York, pp 141–200

11. Capriani A, Chiavegato A, Franch R, Azzarello G, Vinante O, Sartore S (1997) Oestrogen-dependent expression of the SM2 smooth muscle-type myosin isoform in rabbit myometrium.

J Muscle Res Cell Motil 18:413–427

12. Clement S, Chaponnier C, Gabbiani G (1999) A subpopulation of cardiomyocytes expressing alpha-skeletal actin is identified by a specific polyclonal antibody. Circ Res 85:e51–e58 13. Coindre JM, Trojani M, Contesso G, David M, Rouesse J, Bui

NB, Bodaert A, De Mascarel I, De Mascarel A, Goussot JF (1986) Reproducibility of a histopathologic grading system for adult soft tissue sarcoma. Cancer 58:306–309

14. Crouch GD, Helman LJ (1991) All-trans-retinoic acid inhibits the growth of human rhabdomyosarcoma cell lines. Cancer Res 51:4882–4887

15. Engers R, van Roy F, Heymer T, Ramp U, Moll R, Dienst M, Friebe U, Pohl A, Gabbert HE, Gerharz CD (1996) Growth in- hibition in clonal subpopulations of a human epithelioid sarco- ma cell line by retinoic acid and tumour necrosis factor alpha.

Br J Cancer 73:491–498

16. Gabbiani G, Kapanci Y, Barazzone P, Franke WW (1981) Immunochemical identification of intermediate-sized fila- ments in human neoplastic cells. A diagnostic aid for the sur- gical pathologist. Am J Pathol 104:206–216

17. Gavrieli Y, Sherman Y, Ben-Sasson SA (1992) Identification of programmed cell death in situ via specific labeling of nucle- ar DNA fragmentation. J Cell Biol 119:493–501

18. Gerharz CD, Ramp U, Reinecke P, Ressler U, Dienst M, Marx N, Friebe U, Engers R, Pollow K, Gabbert HE (1996) Effects of TNF-alpha and retinoic acid on the proliferation of the clear cell and chromophilic types of human renal cell carcinoma in vitro. Anticancer Res 16:1633–1641

19. Giguere V (1994) Retinoic acid receptors and cellular retinoid binding proteins: complex interplay in retinoid signaling. En- docr Rev 15:61–79

20. Goodman DS (1984) Vitamin A and retinoids in health and disease. N Engl J Med 310:1023–1031

21. Higgins KA, Perez JR, Coleman TA, Dorshkind K, McComas WA, Sarmiento UM, Rosen CA, Narayanan R (1993) Anti- sense inhibition of the p65 subunit of NF-kappa B blocks tu- morigenicity and causes tumor regression. Proc Natl Acad Sci USA 90:9901–9905

22. Laemmli UK (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227:680–

685

23. Loughney AD, Kumarendran MK, Thomas EJ, Redfern CP (1995) Variation in the expression of cellular retinoid binding proteins in human endometrium throughout the menstrual cycle. Hum Reprod 10:1297–1304

24. Miano JM, Topouzis S, Majesky MW, Olson EN (1996) Reti- noid receptor expression and all-trans retinoic acid-mediated growth inhibition in vascular smooth muscle cells. Circulation 93:1886–1895

25. Miettinen M, Lehto VP, Badley RA, Virtanen I (1982) Expres- sion of intermediate filaments in soft-tissue sarcomas. Int J Cancer 30:541–546

26. Napoli JL (1986) Retinol metabolism in LLC-PK1 Cells.

Characterization of retinoic acid synthesis by an established mammalian cell line. J Biol Chem 261:13592–13597

27. Napoli JL, Posch KP, Fiorella PD, Boerman MH (1991) Physi- ological occurrence, biosynthesis and metabolism of retinoic acid: evidence for roles of cellular retinol-binding protein 39

(CRBP) and cellular retinoic acid-binding protein (CRABP) in the pathway of retinoic acid homeostasis. Biomed Pharmac- other 45:131–143

28. Neuville P, Geinoz A, Benzonana G, Redard M, Gabbiani F, Ropraz P, Gabbiani G (1997) Cellular retinol-binding protein-1 is expressed by distinct subsets of rat arterial smooth muscle cell in vitro and in vivo. Am J Pathol 150:509–521

29. Neuville P, Yan Z, Gidlof A, Pepper MS, Hansson GK, Gabbiani G, Sirsjo A (1999) Retinoic acid regulates arterial smooth muscle cell proliferation and phenotypic features in vi- vo and in vitro through an RARalpha-dependent signaling pathway. Arterioscler Thromb Vasc Biol 19:1430–1436 30. Neuville P, Bochaton-Piallat ML, Gabbiani G (2000) Retino-

ids and arterial smooth muscle cells. Arterioscler Thromb Vasc Biol 20:1882–1888

31. Oltvai ZN, Milliman CL, Korsmeyer SJ (1993) Bcl-2 heterodi- merizes in vivo with a conserved homolog, Bax, that acceler- ates programmed cell death. Cell 74:609–619

32. Orlandi A, Ehrlich HP, Ropraz P, Spagnoli LG, Gabbiani G (1994) Rat aortic smooth muscle cells isolated from different layers and at different times after endothelial denudation show distinct biological features in vitro. Artherioscler Thromb Vasc Biol 14:982–989

33. Orlandi A, Francesconi A, Cocchia D, Corsini A, Spagnoli LG (2001) Phenotypic heterogeneity influences apoptotic suscep- tibility to retinoic acid and cis-platinum of rat arterial smooth muscle cells in vitro: implications for the evolution of experi- mental intimal thickening. Arterioscler Thromb Vasc Biol 21:1118–1123

34. Ottonello S, Scita G, Mantovani G, Cavazzini D, Rossi GL (1993) Retinol bound to cellular retinol-binding protein is a substrate for cytosolic retinoic acid synthesis. J Biol Chem 268:27133–27142

35. Prayson RA, Goldblum JR, Hart WR (1997) Epithelioid smooth-muscle tumors of the uterus: a clinicopathologic study of 18 patients. Am J Surg Pathol 21:383–391

36. Rosai J (1998) Ackerman’s surgical pathology, 8th edn.

Mosby-Year Book, St Louis

37. Schürch W, Skalli O, Seemayer TA, Gabbiani G (1987) Inter- mediate filament proteins and actin isoforms as markers for soft tissue tumor differentiation and origin. I. Smooth muscle tumors. Am J Pathol 128:91–103

38. Siddiqui NA, Loughney A, Thomas EJ, Dunlop W, Redfern CP (1994) Cellular retinoid binding proteins and nuclear reti- noic acid receptors in endometrial epithelial cells. Hum Re- prod 9:1410–1416

39. Soini Y, Paakko P (1996) Bcl-2 is preferentially expressed in tumours of muscle origin but is not related to p53 expression.

Histopathology 28:141–145

40. Spagnoli LG, Orlandi A, Santeusanio G (1991) Foam cells of the rabbit atherosclerotic plaque arrested in metaphase by colchicine show a macrophage phenotype. Atherosclerosis 88:87–92

41. Towbin H, Staehlin T, Gordon J (1979) Electrophoretic trans- fer of proteins from polyacrylamid gels to nitocellulose sheets:

procedure and some applications. Proc Natl Acad Sci USA 76:4350–4354

42. Valenti MT, Azzarello G, Vinante O, Manconi R, Balducci E, Guidolin D, Chiavegato A, Sartore S (1998) Differentiation, proliferation and apoptosis levels in human leiomyoma and leiomyosarcoma. J Cancer Res Clin Oncol 124:93–105

43. Waseda N, Kato Y, Imura H, Kurata M (1981) Effects of tamoxifen on estrogen and progesterone receptors in human breast cancer. Cancer Res 41:1984–1988

44. Weibel ER (1979) Stereological methods, vol 1. Practical methods for biological morphometry. Academic Press, London, pp 9–60

45. Xu G, Redard M, Gabbiani G, Neuville P (1997) Cellular retinol-binding protein-1 is transiently expressed in granula- tion tissue fibroblasts and differentially expressed in fibro- blasts cultured from different organs. Am J Pathol 151:1741–

1749

46. Xu G, Bochaton-Piallat ML, Andreutti D, Low RB, Gabbiani G, Neuville P (2001) Regulation of alpha-smooth muscle actin and CRBP-1 expression by retinoic acid and TGF-beta in cul- tured fibroblasts. J Cell Physiol 187:315–325

47. Yamate J, Tsujino K, Kumagai D, Sato K, Tsukamoto Y, Kuwamura M, Kotani T, Sakuma S, LaMarre J (1998) Influ- ence of progesterone and oestrogen on growth and morpholo- gy of a transplantable rat uterine smooth muscle tumour (SMT-Y). J Comp Pathol 119:443–457

48. Zile MH (1998) Vitamin A and embryonic development: an overview. J Nutr 128:455S-458S