Lipid droplet and milk lipid globule membrane associated placental protein 17b (PP17b) is involved in apoptotic and differentiation processes of human epithelial cervical carcinoma cells

Authors


  • Dedication: This work was performed under the inspiring guidance and careful supervision of the late Prof. Gabor Nandor Than, who passed away in March 31, 2002. All the authors and his colleagues dedicate this work to his memory.

N. G. Than, Department of Biochemistry and Medical Chemistry, University of Pecs, 12 Szigeti Street, Pecs H-7624, Hungary. Fax: +36 72 536 277, Tel.: +36 30 9512 026, E-mail: gabor.than@aok.pte.hu

Abstract

The intracellular role of placental protein 17b (PP17b)/TIP47 has been controversial, because it is considered to be a protein required for mannose 6-phosphate receptor transport from endosome to trans-Golgi as well as a neutral lipid droplet-associated protein. The similarity between the amino acid sequences of PP17 variants, adipophilin and perilipins, and between their gene structures indicate that PP17b as well as other alternatively spliced PP17 variants belong to the lipid storage droplet protein family, containing also some differentiation factors. Using a specific antibody, PP17b was detected in lipid droplet fractions and co-localized with neutral lipid droplets stained by Nile red, and fluorescently labelled PP17 antibody in HeLa cells with confocal microscopy. PP17b was also detected in milk, associated to milk lipid globule membranes. Cytostatic agents induced apoptosis and PP17b synthesis in HeLa cells, which was significantly inhibited by protein kinase C (PKC) inhibitor, indicating the involvement of NF-κB and AP-1 transcription factors in this process, while protein kinase A (PKA) inhibitor had only a modest inhibitory effect. Cell differentiation induced by dibutyryl cyclic AMP or phorbol myristate acetate also increased PP17b synthesis, demonstrating its strong involvement in cell differentiation. PP17b synthesis was higher in M than in G0/G1 phases in control, apoptotic and differentiated cells. This data shows that PP17b is a neutral lipid droplet-associated protein, and its expression is regulated by PKC- and PKA-dependent pathways.

Abbreviations
dbcAMP

dibutyryl cyclic AMP

EST

expressed sequence tag

FITC

fluorescein isothiocyanate

MLGM

milk lipid globule membrane

MPR

mannose 6-phosphate receptor

PKA

protein kinase A

PKC

protein kinase C

PMA

phorbol myristate acetate

PP17

Placental Protein 17

PSD

post source decay

TIP47

tail interacting protein of 47 kDa

Our laboratories performed detailed studies on the expression of placenta-specific genes in pregnancy and in different tumours, indicating possible oncodevelopmental functions of these proteins [1]. The 30-kDa soluble placental protein 17 (PP17) was isolated and characterized physico-chemically among the first proteins identified in human placenta [2]. PP17 serum levels were found to be slightly elevated during pregnancy compared to the nonpregnant status [3]. Later it was also shown that the PP17 protein family consists of four PP17 variants (PP17a, PP17b, PP17c and PP17d of 30, 48, 60 and 74 kDa) [4]. The entire nucleotide and amino acid sequences of C-terminus sharing 30 kDa PP17a (AF051314, AF051315) and 48 kDa PP17b (AF055574) were determined and deposited in GenBank. Furthermore, their expression patterns in various human tissues and serum levels in different conditions were also studied and published in this Journal for the first time [4,5]. The closest homologues of PP17 variants were found to be human adipophilin [6] and mouse adipose differentiation-related protein [7] involved in early adipocyte differentiation; and human [8] and rat [9] perilipins, major hormonally regulated adipocyte-specific phosphoproteins. The subsequently GenBank deposited TIP47 (AF057140) proved to be identical to PP17b. It was shown that TIP47–glutathione S-transferase fusion proteins bind to both the cation-dependent and -independent mannose 6-phosphate receptors (MPRs) in vitro, and thus the protein was named TIP47 (tail-interacting protein of 47 kDa). It was proposed that TIP47 directs the retrieval of MPRs from the prelysosomal compartment with delivery back to the trans-Golgi network through interaction with the cytoplasmic tails of MPRs [10].

In parallel, a debate started on the possible function of TIP47, as a recent paper had stated that TIP47 plays a role in intracellular lipid metabolism rather than in secretory protein sorting, taking into account that there is a high-level amino acid sequence similarity between the N-terminal region of TIP47 and other lipid droplet-associated proteins, which localize on the surface of lipid droplets in a lipid synthesis/storage status-responsive manner [11]. A reply paper reinforcing the protein's MPR transport function emphasized that TIP47 is not a lipid droplet component, and accused probable cross-reactivity of the TIP47 antibody with the N-terminus of adipophilin of leading to that finding [12]. Most recently, evidence was presented using green fluorescent protein-tagged TIP47, that it colocalizes with intracellular lipid droplets, showing that there is discrepancy regarding the cellular function of TIP47 [13].

In the past 4 years, the oncological significance and over-expression of PP17b in human uterine squamous cervical carcinoma tissues and HeLa (squamous cervical cancer) cells were established. Serum PP17b levels were found to be elevated in cervical carcinoma patients, and this declined after radical surgery [5,14,15]. Normal cervical epithelia were negative for PP17b, while cytoplasms of the dysplastic cells were positive in low-grade dysplasias, and strongly positive in high-grade dysplasias. In invasive squamous cervical carcinomas, cytoplasms of basal-type tumour cells were negative, while squamous-type dysplastic cells were strongly positive [16]. Now, by extensive databank search, structural similarities between human PP17 (TIP47), adipophilin and perilipin genes have been revealed, and analysis of the 5′ flanking region of the PP17 gene has shown a number of potential transcription factor binding sites indicating its complex transcriptional regulation.

Using a HeLa cell model, evidence was found for alternative splicing of PP17 variants, the importance of protein kinase A (PKA)- and protein kinase C (PKC)-dependent pathways for the regulation of PP17 gene expression was demonstrated, and the effect of the phase of cell cycle, differentiation and apoptosis on expression of this gene was also studied. Furthermore, evidence was presented for the association of PP17b to lipid droplets and milk lipid globule membranes, showing that PP17b binds to heterologous intracellular lipid droplet surfaces indicating its function in lipid deposition and/or mobilization.

Experimental procedures

Materials

PP17 antigen (Op. 169/195) and anti-PP17 rabbit antibody (54ZB) were prepared by H. Bohn. Fluoresceine isothiocyanate (FITC)-labelled goat anti-rabbit IgG was from BD Biosciences, Heidelberg, Germany, anti-bax (4F11) and anti-Bcl2 (124) monoclonal antibodies, and Universal Kit were from Immunotech, Marseille, France, the HeLa S3 cell line was from the ATCC. We purchased 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride, antibiotic-antimycotic solution, benzamidine, BSA, CnBr-activated agarose beads, dibutyryl cyclic AMP, DNase-free RNase, Dulbecco's modified Eagle's medium, fetal bovine serum, horseradish peroxidase-labelled goat anti-rabbit IgG, leupeptin, oleic acid, and phorbol myristate acetate from Sigma. Nile red was from Molecular Probes Inc., PKA and PKC inhibitors were from Calbiochem, trypsin was from Promega, ZipTipC18 pipette tips were from Millipore, ECL chemiluminescence system was from Amersham Pharmacia Biotech, carboplatin/Paraplatin was from Bristol-Myers-Squibb, 5-fluorouracil (Lederelle) was from Wyeth-Whitehall, Wolf Rats Hausen, Germany, irinotecan/Campto was from Rhone-Poulenc Rorer, West Malling, UK, mitomycin C was from Kyowa Hakko Kogyo Co. Ltd, Tokyo, Japan, and paclitaxel/Taxol was from Bristol Arzneimittel GmBH, München, Germany.

Databank search

PP17b cDNA was compared to different expressed tag sequences (ESTs) and genomic databases by blast[17] and ucsc Genome Browser and alignments of PP17b cDNA and related EST genomic sequences were performed with locuslink[18], all provided by NCBI (Bethesda, MD, USA). The Transfac Database was searched [19] for putative transcription binding sites at the 5′ flanking region of PP17 gene using patsearch (GBF-Braunschweig, Germany). Multiple amino acid sequence alignment of PP17b to its homologues was carried out with clustalw at EMB-net (Lausanne, Switzerland) [20].

Cell culture and drug treatments

Confluent monolayers of synchronized HeLa cells were grown on 100-mm dishes in standard Dulbecco's modified Eagle's medium containing 1% antibiotic/antimycotic solution, supplemented with 10% fetal bovine serum under 5% CO2 conditions and 95% humidified air at 37 °C. For immunocytochemistry and confocal immunofluorescence microscopy, cells were cultured on poly 2-lysine coated glass cover slips, dried overnight and stored at −80 °C. To increase triacylglicerol storage, cells were incubated in culture media supplemented with 600 µm oleic acid complexed to fatty acid-free BSA (molar ratio of 6 : 1) for 20 h. For apoptosis induction, cytostatic drugs (carboplatin 0.75 µg·mL−1, 5-fluorouracil 25 µg·mL−1, irinotecan 5 µg·mL−1, mitomycin 10 µg·mL−1, and paclitaxel 10 nm) were diluted in culture medium and applied for 24 h. To induce differentiation, cells were treated with 0.5 mm dibutyryl cyclic AMP (dbcAMP) for 72 h or 80 nm phorbol myristate acetate (PMA) for 48 h. There were cells incubated with 0.1 µm PKC inhibitor or 0.36 µm PKA inhibitor (10 ×Ki in each cases) parallel to treatments with paclitaxel, dbcAMP or PMA.

Subcellular fractionation

Lipid-loaded HeLa cells were harvested and centrifuged at low speed. Pellets were dispersed by vortexing in hypotonic lysis buffer (10 mm Tris pH 7.4, 1 mm EDTA, 1 mm benzamidine, 100 µm 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride and 10 µg·mL−1 leupeptin) for 10 min at 4 °C. After further cell disruption in a Teflon/glass homogenizer, homogenates were centrifuged for 10 min at 1000 g at 4 °C, the supernatants were mixed with 70% sucrose (w/w) in a ratio of 1 : 1.5, and layered under a linear 0–40% sucrose (w/w) gradient. 6-mL tubes were centrifuged for 4 h at 154 000 g in a Beckman SW41Ti rotor at 4 °C. Lipid droplet fractions (1 mL) were collected by slicing off the tops of the tubes with a Beckman tube slicer, and then five additional 1-mL fractions were collected. Equal portions of the fractions were either separated by SDS/PAGE for Western blot or extracted with solvent for lipid analysis.

Milk lipid globule membrane (MLGM) fractionation and monolayer preparation

Total protein of fresh milk obtained from human volunteers was extracted between five and eight times with chloroform/methanol (1 : 1 and 2 : 1, v/v) at ratios not exceeding 5 mg protein·mL−1. For isolation of MLGM-associated proteins, MLGM fractions were separated from milk by sucrose gradient centrifugation, and then proteins were further separated from lipids by chloroform/methanol extraction. Both total and MLGM-associated proteins were then subjected to SDS/PAGE/Western blot. For immunofluorescence imaging, we developed MLMG monolayers by mixing milk with 0.5% agar (w/v) at 1% at 60 °C, then fixing the mixture on glass cover slips.

SDS/PAGE/Western blot

PP17 antigen (1 ng) and 10–10 µg protein from term placenta total protein extract, HeLa total protein extracts and subcellular fractions, milk total protein extracts and MLMG fractions were subjected to SDS/PAGE (12% acrylamide, w/v). Immunoblots were carried out with anti-PP17 antibody and horseradish peroxidase-labelled secondary IgG as described earlier [21]. Protein bands were revealed by ECL chemiluminescence followed by quantitative densitometry using scion image for Windows.

Lipid analysis

Solvent extraction and TLC of neutral lipids were carried out as described [22]; densitometric quantification was similar as for proteins.

Immunoaffinity purification and protein identification by MS

Anti-PP17 Ig was coupled to CnBr-activated agarose beads and incubated with lipid loaded HeLa cell or milk total protein extracts at room temperature for 30 min. The gels were washed three times with 20 mm Tris/HCl pH 7.4 containing 150 mm NaCl to remove unbound proteins. The immunoreactive proteins were removed with an equal volume of 2 × Laemmli sample buffer, then proteins were separated by gradient (6–18%) one-dimensional PAGE and visualized by Coomassie blue staining. Bands of interest were excised from the gel, reduced, alkylated and in-gel digested with trypsin as described previously [23]. Proteins were identified by a combination of MALDI-TOF MS peptide mapping and MALDI-post source decay (PSD) MS sequencing. The digests were purified with ZipTipC18 pipette tips with a saturated aqueous solution of 2.5-dihydroxybenzoic acid matrix (ratio of 1 : 1). A Bruker Reflex IV MALDI-TOF mass spectrometer (Bruker-Daltonics, Bremen, Germany) was used for peptide mass mapping in positive ion reflector mode with delayed extraction. The monoisotopic masses for all peptide ion signals in the acquired spectra were determined and used for database searching against a nonredundant database (NCBI) using ms fit (UCSF, San Francisco, CA, USA) [24]. Primary structures of tryptic peptide ions were confirmed by PSD MS sequencing.

Immunolocalization of PP17b in squamous cervical carcinoma tissue sections and in HeLa cells

Tissue sections were prepared from routine formalin-fixed, paraffin-embedded samples of invasive uterine squamous cervical carcinoma (n = 20). Four-µm sections were cut, mounted on slides, dried at 37 °C overnight, dewaxed and rehydrated. Both tissue sections and the cell culture samples described above were incubated with anti-PP17 antibody, and with monoclonal anti-bax and anti-bcl2 antibodies for the parallel assessment of apoptosis [25]. Immunostaining was carried out according to the streptavidin/biotin/peroxidase technique, with hydrogen peroxide/3-amino-9-ethylcarbazole development using the Universal Kit [26]. Visual evaluation of haematoxylin-counterstained slides was performed by using an Olympus BX50 light microscope with an integral camera (Olympus Optical Co., Hamburg, Germany).

Confocal immunofluorescence microscopy

Fixed cells and MLMG monolayers were consecutively treated with anti-PP17 antibody followed by FITC-labelled secondary IgG in NaCl/Pi containing 0.1% saponin and 0.1% BSA. For neutral lipid staining, 0.01% Nile red dissolved in dimethylsulfoxide was added parallel to the secondary antibody solutions. Cell fluorescence was monitored with a Bio-Rad MRC-1024ES laser scanning confocal attachment mounted on a Nikon Eclipse TE-300 inverted microscope.

Flow cytometry and cell cycle analysis

Synchronized cultured cells were harvested, washed in NaCl/Pi and fixed with 4% paraformaldehyde for 20 min at 4 °C. Immunofluorescent intracellular PP17 staining was performed in permeabilization buffer (0.1% saponin, 0.1% NaN3 and 0.1% BSA in NaCl/Pi) with a two-step labelling technique [27], using anti-PP17 Ig and FITC-labelled secondary IgG for 30 min each at 4 °C. For cellular DNA content analysis, after intracellular staining, samples were incubated with 100 µg·mL−1 RNase followed with 5 µg·mL−1 propidium iodide for 30–30 min at 24 °C. Between 10 000 events were measured in each sample on a FACSCalibur flow cytometer (Becton Dickinson) and analysed statistically using CellQuest software. PP17 quantities were measured in FL-1, while cellular DNA content was measured in the FL-2 channel. To determine PP17 gene expression in cell cycle phases, gates were set on different peaks of the FL-2 histograms.

Statistical evaluation

Values in the figures, tables and text are expressed as mean ± SEM of n observations. Statistical analysis was performed by analysis of variance followed by Turkey's and chi-square tests. Statistical significance was set at P < 0.05.

Results and discussion

PP17 gene: expression, structure and regulation

A GenBank search revealed a high variety of alternatively spliced human ESTs − related to PP17a and PP17b cDNAs by length and sequence − in almost all types of healthy tissue. ESTs were highly expressed in placenta and epithelial origin tumours. These underlined our previous Northern and Western blot results, showing that PP17a is mostly a steroidogenic tissue protein, while PP17b is an ubiquitously synthesized oncodevelopmental protein, both members of an alternatively spliced protein family, homologous to the perilipins. Genomic alignment of PP17b cDNA and the longest EST (BI561840) sequences mapped the PP17 gene (Locus ID: 10226) to 19p13.3 (genomic contig: NT_011255), containing eight exons sized from 82 to 943 bp, spanning ≈ 29.0 kb, with all exon–intron boundaries conforming to consensus sequences [28]. This gene lacks a canonical TATA box, but a putative initiator element (Inr) was found in it, contained by genes with TATA-less promoter [29]. The 5′ end of the longest EST started at the consensus start site (A) of the Inr, confirming it to be the first nucleotide of the first noncoding exon. A downstream promoter element [30], a pyrimidine-rich element [31] and several GC-rich consensus GCF [32] and SP-1 [33] transcription factor binding sites clustered in the vicinity of the Inr might serve in transcription initiation (Fig. 1).

Figure 1.

Nucleotide sequence and possible transcriptional regulation of the human PP17 gene. The figure displays eight exons in bold type upper case letters, seven introns as well as the 5′- and 3′-flanking regions in lower case italics, and the consensus GT/AG splice junction sites underlined. Start (ATG) and stop (TAG) codons in exons 2 and 8 are inverse typed. In the absence of a canonical TATA box, double underlined pyrimidine-rich element (−23), initiator element (Inr; putative initiation site boxed) and downstream promoter element (DPE; +50) may serve an identical function in PP17 gene. GC-rich consensus binding sequences for transcription initiation factors (GCF highlighted, SP-1 boxed) are also indicated.

Analysis of the 1.5-kb 5′ flanking region, attempting to get further insight into the possible regulation of PP17 gene, showed numerous different consensus transcription factor binding sequences clustered preceding the 5′ end of the first exon (Table 1). Factors potentially involved in the transcription of PP17 gene include: (a) general activators or repressors GCF, SP-1, YY1 [34] and USF [35]; (b) coactivators, AP-4 [36] and P-300 [37]; (c) cAMP/PKA, PKC or phorbol ester responsive elements, AP-1 [38], AP-2 [39], CREB [40], GCF and NF-κB [41]; (d) hematopoietic regulators, AML [42], GATA-1 [43], LYF [44], MZF-1 [45] and PAX-5 [46]; (e) adipose differentiation regulator, PPARγ[47]; (f) myogenic factor, MYO-D [48]; (g) keratinocyte specific factors, AP-2, GCF and PAX-2 [49]; (h) factors abundant in placenta, AHR [50], AP-2 and PPARγ; (i) proliferation and/or apoptosis regulators, AP-2, c-MYC [51] and NF-κB; (j) embryo- and organogenic factors, PAX-2 and PAX-5; (k) proto-oncogenes or their targets, AML, AP-1, AP-2, PAX-2, PAX-5, PEA-3 [52] and PPARγ; (l) aryl hydrocarbon regulators, AHR and ARNT [50]. From these, it may be concluded that: (a) ubiquitous PP17b synthesis could be derived from possible gene regulation by factors involved in development of different cells; (b) oncodevelopmental significance of PP17b must be re-emphasized by locating potential binding sites for factors engaged in proliferation, oncogenesis or development; (c) PP17b could be involved in lipid metabolism and droplet formation regulated by PPARγ; (d) apoptotic and (e) differentiation pathways could utilize the as yet unestablished function of PP17b.

Table 1. Possible transcriptional regulation of the human PP17 gene. Computed positions of binding sites for consensus transcription factors in PP17 gene promoter are indicated relative to the putative Inr.
Transcription
factor
Binding
position
Transcription
factor
Binding
position
Transcription
factor
Binding
position
AML-1−704AHR−790MYO-D−172
AP-1−920 −175NF-κB−169
−892ARNT−569PAX-2−642
−870c-MYC−589PAX-5−209
−817CREB−646PEA-3−928
−539GATA−1−990PPARγ−644
−496GCF−147P-300−639
−384 −144SP-1−961
−27 −90 −542
AP-2−876 −74 −93
−665 −54 −77
−225 +4USF−572
−217LYF−976YY1−1022
+35 −674 −845
AP-4−201MZF-1−292 −221

PP17b is a member of the growing lipid storage droplet protein family

By multiple sequence alignment, PP17b proved to have a close structural similarity to human adipophilin and perilipin, members of the newly discovered lipid droplet-associated protein family, sharing a common N-terminal motif [53]. Alignment of their cDNAs to genomic sequences, and superimposition of exon–intron boundaries to the aligned proteins revealed some common characteristics of their genes (Fig. 2). Although genomic sizes and locations (PP17: 29.0 kb, 19p13.3; hADFP: 12.2 kb, 9p21.3; hPLIN: 15.6 kb, 15q26) and intron sizes were divergent, homology was proven by the similar number and length of exons, the corresponding analogous peptide lengths, and the high number of identical and conserved residues. The most conserved regions in all three proteins were encoded by exons 3 and 4, where PP17b had 38–56% identity and 68–82% similarity to its closest homologues. On its C-terminus PP17b had a lower level of sequence similarity to perilipin (29–42%) than to adipophilin (50–70%), and the number of identical residues with the latter was also significantly reduced (26–43%) (Table 2). This is the first comparison of PP17b with two human members of the newly discovered ‘PAT domain gene family’[53], suggesting their common genetic origin. Shared characteristics in the regulation of PP17 gene with other family members were also found: (a) the promoter region of the murine perilipin gene is similar to the human PP17 gene, lacking also TATA box [53]; (b) the mouse adipose differentiation related protein gene contains several transcription factor binding sites (AP2, PAX-2, C-MYC, SP1) [54], as does PP17 gene; (c) the expression of human adipophilin is highly inducible by PPARγ, which plays a fundamental role in lipid catabolism and adipocyte differentiation, as well as in epithelial differentiation [55]. In light of these findings collectively, the concept of PP17b being a member of the lipid storage droplet protein family was to be analysed further.

Figure 2.

Structural relationship of human PP17, adipophilin and perilipin genes and proteins. After multiple sequence alignment of PP17b, adipophilin (hADFP) and perilipin (hPLIN), and alignment of their cDNAs to genomic sequences, aligned proteins were superimposed with corresponding exon boundaries. Identical amino acids are shown in bold type, subsequent exons are indicated by alternate highlighting.

Table 2. Conserved regions in human PP17, adipophilin and perilipin genes and proteins. Exon lengths, corresponding peptide lengths, identities/similarities to PP17b, and following intron sizes were compared for each gene. Although intron sizes are divergent, the structures of PP17 and hADFP genes are highly conserved, and are closely related to the hPLIN gene. The most conserved regions in all three proteins (bold type) are encoded by exons 3 and 4.
ExonExon length (bp)Peptide length (aa)Identity/similarity (%)Following intron (kb)
PP17bADFPPLINPP17bADFPPLINADFPPLINPP17bADFPPLIN
1111541026.21.02.0
282525822101518/2721/421.30.14.0
319819520466656856/8238/680.12.51.6
484848428282854/7554/797.32.41.2
528528526495958843/7020/424.11.00.3
620118317467615836/7013/292.91.21.7
712613519242456426/5014/405.11.70.4
89438792461141338240/7015/381.0
91577129

PP17b is localized on lipid droplets and milk lipid globule membranes

As previous findings by other groups were contradictory on the function of TIP47 − recently detecting lipid droplet association of the previously believed mannose 6-phosphate receptor transporter with a polyclonal antibody [13] − this question was now examined on invasive squamous cervical carcinomas and HeLa cells using our highly specific anti-PP17 antibody. In fixed embedded tissue sections of squamous cervical carcinoma, mainly tumour cells with squamous differentiation were stained in a punctate pattern. At higher magnification, positive granules showed an unstained core, mimicking lipid droplets (Fig. 3A). Similarly, lipid-loaded HeLa cells had a characteristically granular cytoplasmic PP17 localization (Fig. 3B). By confocal imaging, there was a large difference between cells cultured under low or high lipid concentrations. Compared with control cells (Fig. 3C), in lipid-loaded cells spherical structures stained with anti-PP17 antibody in the cytoplasm (Fig. 3D). Large clusters of these globules strongly double-stained with anti-PP17 antibody and Nile red, appearing to be neutral lipid droplets. At higher magnification, even distinct PP17-positive rings surrounding the droplet surfaces could be detected (Fig. 3E). Confocal images supported our computational finding that PP17b belongs genetically and structurally to a new protein family, and also reinforced the postulation that PP17b is a constituent of lipid droplets. Moreover, the same PP17-positive ring could be detected on the surface of double-labelled milk lipid globule membranes, with weaker reticular PP17 staining inside of MLMGs, which was probably the result of the surface protein internalization as small lipid droplets developed into large MLMGs (Fig. 3F).

Figure 3.

Lipid droplets in invasive squamous cervical carcinomas. HeLa cells and human milk are stained with anti-PP17 antibody. (A) In invasive squamous cervical carcinoma, tumour cells have punctuated, ring-like cytoplasmic PP17 staining (immunohistochemistry, haematoxylin counterstain). (B) Lipid-loaded HeLa cells have dominantly granular PP17 staining (immunocytochemistry, haematoxylin counterstain). (C) Compared to controls (D) in lipid-loaded cells spherical structures stained with anti-PP17 Ig are seen (confocal immunofluorescence microscopy). (E) In lipid-loaded cells, clusters of small lipid droplets are double-labelled with anti-PP17 Ig (green) and Nile red (red), colocalization is represented in yellow. The inset magnifies lipid droplets surrounded by distinct PP17 positive ring (confocal immunofluorescence microscopy). (F) A strong PP17 staining around the surface, and weaker signs inside of double-labelled MLMGs is present.

All of these allow some parallels to be indicated: (a) it is thought that perilipins may bring small lipid droplets together, probably by protein–protein interactions [56], while PP17a and PP17b have coiled-coil structures, and were detected to dimerize or oligomerize in natural or even denatured conditions [5,10], which might enable them to play a role in lipid droplet aggregation and formation; (b) alternatively spliced perilipin isoforms have different distribution in steroidogenic cells or adipocytes [22], while a tissue-specific distribution of PP17 variants was also discovered, as PP17b was ubiquitously expressed, while PP17a expression was restricted to steroidogenic tissues only [5]; (c) adipophilin was purified from milk and its cDNA was isolated from a mammary gland clone collection [57], while human mammary gland and mammary adenocarcinoma ESTs similar to PP17b cDNA was found by blast, and subsequently PP17b cDNA was also found to be differentially expressed in breast cancer cell lines [58], indicating that the staining of MLMGs was probably not due to a simple cross-reaction.

To disclose cross-reaction with adipophilin at all and to assess the exact subcellular distribution of PP17 variants detected by our highly specific antibody, fractionation and Western blotting of HeLa cells were subsequently performed. In cells cultured under low lipid concentrations, small amounts of PP17a, PP17b and PP17c were found in the buoyant lipid droplet fraction, while almost all the staining for these proteins could be detected in the cytosol (Fig. 4A). In lipid-loaded cells, amounts of PP17a, PP17b and PP17c were increased in the cytosol fraction, and in parallel an intense elevation of PP17b in the lipid-droplet fraction was detected, as evidence for droplet association of PP17b (Fig. 4B). In total milk, high amounts of PP17b and PP17c were identified, whereas mainly PP17b was associated to MLMG fractions (Fig. 4C).

Figure 4.

Western blot of PP17 variant distribution in HeLa cells and human milk. (A) In control cells, small amounts of PP17a, PP17b and PP17c (PP17a dimer) was present in the buoyant fraction and moderate amounts in cytosol. (B) In lipid-loaded cells, amounts of PP17a, PP17b and PP17c were slightly increased in cytosol, while the quantity of PP17b was significantly elevated in the lipid droplet fraction. Lane 1, floating lipid droplet fraction; lane 2–5, intermediate fractions; lane 6, cytosol fraction. Amounts of neutral lipids in each fraction were quantified with densitometric scanning and shown semiquantitavely. (C) In total milk (lane 1), high amounts of PP17b and PP17c were found, while in MLMG fraction (lane 2), mainly PP17b was detected. Markers indicate molecular masses in kDa.

Following this, the PP17 immunoreactive 30, 48 and 60 kDa proteins were purified from lipid-loaded HeLa cell extracts and human milk, then MALDI-TOF MS peptide mapping and MALDI-PSD MS sequencing were performed. Each protein band yielded a good quality peptide map, and most of the input masses were matched to the candidate protein sequences. The majority of the tryptic peptides matched with the theoretical masses within 62 p.p.m. MALDI-TOF MS data of the 48-kDa protein permitted the identification of PP17b, and mass maps of the 30- and 60-kDa proteins matched PP17a with 46% coverage of the protein sequence. PSD data obtained for precursors also confirmed the identity of these proteins (Table 3). These data show the specificity of our original antibody, excludes cross-reactivity with its human homologues, reinforces dimerization of PP17a to PP17c, and also confirms the lipid–droplet association of PP17b.

Table 3. Assignments of proteolytic fragments from tryptic digests of PP17 immunoaffinity purified 30-, 48- and 60-kDa proteins. Protein identification and sequencing are described in Materials and methods. Most of the input masses matched candidate protein sequences, and the majority of tryptic peptides matched theoretical masses within 62 p.p.m. MALDI-TOF and PSD MS data identified the 48-kDa protein as PP17b, the 30-kDa protein as PP17a and the 60-kDa protein as PP17a dimer.
Measured
mass (MH+)
Calculated
mass (MH+)
Δ p.p.m.ModificationsFragmentMissed
cleavages
Database sequence
1039.61501039.4849125pyroGlu48–550(R) QEQSYFVR (L)
1056.63561056.5114117 48–550(R) QEQSYFVR (L)
1169.67931169.570393 65–740(R) QHAYEHSLGK (L)
1234.75931234.6717711Met-ox1–120(–) MVLSGVDTVLGK (S)
1295.73971295.659662 132–1420(K) EPPKPEQVESR (A)
1438.84541438.755562 63–741(R) LRQHAYEHSLGK (L)
1438.84541438.755562 65–761(R) QHAYEHSLGKLR (A)
1470.79571470.731644 114–1240(K) LHQMWLSWNQK (Q)
1486.79711486.7265471Met-ox114–1240(K) LHQMWLSWNQK (Q)
1707.86611707.9407−44 63–762(R) LRQHAYEHSLGKLR (A)
1835.05961834.929971 31–470(R) IATSLDGFDVASVQQQR (Q)
2059.15322059.046052pyroGlu125–1421(K) QLQGPEKEPPKPEQVESR (A)
2067.08062066.978349 13–300(K) SEEWADNHLPLTDAELAR (I)
2073.22392073.126647 82–1000(R) AQEALLQLSQALSLMETVK (Q)
2076.20012076.072661 125–1421(K) QLQGPEKEPPKPEQVESR (A)
2089.24532089.1215591Met-ox82–1000(R) AQEALLQLSQALSLMETVK (Q)

PP17b is involved in apoptosis and differentiation of epithelial cells

Several putative transcription factor binding sites involved in apoptosis and differentiation were localized in the PP17 gene promoter. Using well-characterized apoptosis and differentiation models, induction of PP17 gene expression through the supposed pathways were detected, parallel to the morphological changes. PP17 quantities were measured in apoptotic conditions, treating cells with carboplatin, 5-fluorouracil, irinotecan, mitomycin or paclitaxel in clinically achievable concentrations, in various dose–time combinations. Apoptosis was assessed by typical cytomorphological alterations in the nucleus and cytoplasm, and by the elevated bax/bcl2 oncoprotein ratio, widely used for squamous epithelial cells and tissues [25]. The effect and time-course of different apoptosis-inducing agents on PP17 gene expression was varied. Paclitaxel had the highest apoptotic effect, which appeared after 12 h and peaked at 24 h, correlating well with increased PP17 protein synthesis, specifically in small round cells exhibiting clearly apoptotic morphology, with picnotic nuclei and narrow cytoplasm (Fig. 5B and Table 3). By flow cytometry, a strict dose and time dependency of its PP17 inducing effect (+49% after 18 h, +154% after 24 h) were observed (Fig. 6A). Parallel treatment with PKC inhibitor caused significant reduction in PP17 protein synthesis after 24 h (+75%), while PKA inhibitor had less influence on this effect of paclitaxel (+126%) (Fig. 6B).

Figure 5.

PP17 immunostaining of apoptotic and differentiated HeLa cells. A, C, E and G show control cells, B, D, and F paclitaxel (10 nm for 24 h) treated cells, and H dbcAMP (0.5 mm for 72 h) treated cells. In A, B, G and H, cells were stained with PP17, in C and D with anti-bax, in E and F with anti-Bcl2 Ig. Compared with controls, after paclitaxel treatment, synthesis of PP17 variants (B) and bax (D) proteins was strongly increased, whereas Bcl2 (F) was unaltered. During differentiation, PP17 variant synthesis was highly elevated. Punctuated PP17 immunostaining was detected either in apoptotic or in differentiated cells.

Figure 6.

Flow cytometric measurements on PP17 induction during apoptosis or differentiation. (A) Parallel treatment of PKC inhibitor with paclitaxel caused significant (*P < 0.05) reduction in PP17 synthesis compared to paclitaxel alone, while PKA inhibitor did not have so strong an effect. (B and C) During cell differentiation, PP17 synthesis was notably elevated, which could be significantly (*P < 0.05) reduced by PKA and PKC inhibitors only in case of PMA. Values indicated above the bars are the averages of three separate flow cytometric measurements. (#P < 0.05, significant as compared with controls.)

Cells were treated with dbcAMP or PMA to obtain data on PP17 gene involvement in cell differentiation pathways, and both notably induced differentiation and PP17 protein synthesis (Fig. 5H and Table 4). Compared with controls 72 h of treatment with dbcAMP caused the highest PP17 increase (+80%), which did not increase further even at higher concentrations, and could be only moderately reduced by PKA (+61%) or PKC (+63%) inhibitors (Fig. 6C). There was some cell differentiation after PMA treatment, although it was less effective in the induction of PP17 protein synthesis (+72%); however, parallel PKC or PKA inhibitor treatment decreased PP17 induction significantly (+20/+28%) (Fig. 6D).

Table 4. Synthesis of PP17 variants in induced and control HeLa cells. Exposure of HeLa cells to cytostatic and cell differentiation-inducing drugs were carried out as described in Materials and methods. Whole protein extracts were Western blotted, then the revealed bands were analysed by densitometry. Amounts of PP17 variants are shown semiquantitatively.
DrugConcentrationTime (h)Synthesized proteins
PP17aPP17bPP17c
Control cells+/–+/–+/–
Apoptosis induced cells
 Carboplatin/Paraplatin®0.75 μg·mL−124+/–++/–
 Irinotecan/Campto®5.00 μg·mL−124+/–++/–
 5-Fluorouracil/5-FU Lederelle®25.00 μg·mL−124+/–++/–
 Mitomycin/Mitomycin C®10.00 μg·mL−124++++
 Paclitaxel/Taxol®10.00 nm24+++++
Differentiation induced cells
 Phorbol myristate acetate80.00 nm48+/–+++/–
 Dibutyril cyclic AMP0.50 mm72+/–+++/–

In the case of paclitaxel, a time-dependent shift in cell cycle was detected. On average, 65–75% of the control cells were in G0/G1 and 25–35% in M phase. Paclitaxel stopped the cells in M phase after 18 h in parallel with increasing (+49%) PP17 protein synthesis, which peaked after 24 h (+154%) (Fig. 7A). It was remarkable that PP17 protein synthesis was ≈ 40% higher in M than in other phases of the cell cycle in either control, apoptotic (Fig. 7B and C) or differentiated cells, which may also show PP17 gene involvement in differentiation.

Figure 7.

Cell cycle and cycle-dependent PP17 fluorescence in control and paclitaxel induced cells. (A) Cells in G0–G1 phases were gated under R1 and cells blocked in M phase under R2. Propidium iodide fluorescence of control cells is shown by filled bars, paclitaxel induced cells by grey bars. (B and C) Cell cycle-dependent PP17 fluorescence of control and paclitaxel-induced cells. Time-dependent shift in cell cycle was caused by paclitaxel, stopping cells in M phase, with increased PP17 protein synthesis reaching its peak (+154%) after 24 h. PP17 fluorescence was 40% higher in cells in M than in G0–G1 phases in both cases. Figures are representatives of three separate experiments.

It is known that paclitaxel markedly increases the binding of NF-κB and AP-1 transcription factors to their binding sites [59]. The PP17 gene promoter has been shown to contain several NF-κB and AP-1 binding sites, therefore it is likely that paclitaxel induces PP17 gene expression by the activation of NF-κB and AP-1 transcription factors. Furthermore, it is known that PKC inhibitors abolish paclitaxel-induced NF-κB activation [59], which is in concordance with our observation that a PKC inhibitor suppressed paclitaxel-induced PP17 synthesis. Paclitaxel-induced gene expression, cell death and differentiation are regulated by complex protein kinase networks including ERK1,2, c-Jun NH2-terminal kinase and the p38-MAP kinase pathways [60], which may explain the complex regulatory effects that have been seen under different conditions.

It was published that a gene involved in squamous cell differentiation can be effectively induced by PMA using AP-1 binding sites, and its expression is inhibited by PKC inhibitors [61]. This is also consistent with our observations that PMA activated PP17 gene expression, which was decreased by a PKC inhibitor. The PKC/Ras/MEKK1/MKK1-dependent/AP-1 kinase cascade involved in the regulation of PMA-induced gene expressions [62] may be another possible means of PP17 gene regulation.

Conclusions

GenBank analysis of EST clones underlines that alternatively spliced PP17a occurs mainly in steroidogenic tissues, while PP17b is synthesized in almost all types of tissue, especially in placenta and epithelial origin tumours. Sequence data show high level sequence similarity at their N-termini between PP17b and neutral lipid droplet-associated proteins including perilipins and adipophilin, the latter of which was also involved in adipose cell differentiation. Taken altogether, a comparison of PP17b and its gene to perilipins and adipophilin, members of the ‘PAT domain gene family’, similar exon structures, sequence homology and many common transcription factor regulatory sequences in the promoter regions were found, suggesting their common genetic origin and functional similarities.

With different techniques based on immunological reactions, considerable evidence was obtained that PP17b/TIP47 was a neutral lipid droplet-associated protein, which also occurs in significant quantities in milk lipid globule membranes. Because of the controversy in the literature on its function, to avoid possible immunological cross-reactivity a very specific independent technique, MALDI-TOF MS analysis was used, and both PP17 variants − PP17b most markedly − were proved to bind to the surface of neutral lipid droplets. Furthermore, our previous data showed that both PP17a and PP17b could aggregate even in the presence of low concentrations of SDS, raising the possibility that these proteins could be involved in the formation of different-sized lipid droplets. By binding to lipid micelles and having self-aggregating properties, PP17 variants could facilitate lipid droplet aggregation, which is clearly detectable in the case of MLGM. This property of PP17b indicates its function as a neutral lipid droplet associated protein and its involvement in lipid droplet formation/mobilization, in accordance with its possible function in cell and tissue differentiation.

Our previous data on the oncodevelopmental overexpression of PP17b in prematurely aging epithelial-character placentas and squamous epithelial cervical dysplasias and carcinomas indicated a sophisticated regulation of PP17 gene expression. With computer analysis of its 5′ upstream sequence, several transcription factor binding sites were identified, including mostly proliferation and/or apoptosis regulators, embryo- and organogenic factors, proto-oncogenes or their targets, which also points to the possible complex PP17 gene regulation.

Induction of apoptosis and differentiation indeed upregulated PP17 expression, while kinase cascade inhibition led to a transcription factor activation block on the induction of PP17 expression, providing evidence for the importance of those transcription factors in PP17 gene regulation. These data also indicate that PP17b could play an important role in tumour cell development and differentiation. Because providing a rich lipid supply to cells induced lipid droplet formation and PP17b overexpression, this indicates that PPARγ could have a role in the regulation of PP17 expression. Furthermore, these data suggest that the main function of PP17a and PP17b is involvement in lipid droplet formation and in rearrangement of lipid membranes, which processes could also be important in cell differentiation and division. The high concentration of PP17b in milk lipid globule membranes indicates its potential role in exporting lipid droplets and membranes.

In the case of several previously known ‘placental proteins’, which turned out to have a general function in different human tissues, more specific structural or functional names were given, such as galectin-13 (PP13) [63], glycodelin (PP14) [64] or branched-chain aminotransferase (PP18) [65]. As (a) PP17b is synthesized ubiquitously, while PP17a is found mainly in steroidogenic tissues; (b) both PP17 variants are generally involved in lipid droplet formation, like alternatively spliced perilipins, which were shown to bind either to steroid or neutral lipid droplets [66]; (c) neither the name ‘placental protein 17b (PP17b)’ nor ‘tail-interacting protein of 47 kDa (TIP47)’ gives the appropriate information on the structure, function, regulation, or the origin of this protein; (d) there is still a lack of an official name for the ‘PP17/TIP47’ gene; and (e) there is a common need to elucidate this controversial situation, it is therefore now proposed that the PP17 variants be renamed to sandrin A (PP17a) and sandrin B (PP17b) (Steroid And Neutral lipid DRoplet-associated proteIN), and their gene to SNDR.

Acknowledgements

This work was supported by Hungarian Grants ETT T-09 163/01; FKFP 0010/1999, 0166/2001; OMFB-BIO 00041/2001; and OTKA T/020622, T/023076, T/029824, M/36996. We thank J. Bocsi for helpful discussions, S. Paku for technical assistance in confocal immunofluorescence microscopy, R. Keszthelyi and Z. Halas for technical assistance in immunostaining, and S. Starkey for critical reading of the manuscript.

Ancillary