Decrease in nuclear phospholipids associated with DNA replication

One of the more-studied models for DNA replication, in in vivo conditions, is represented by liver regeneration after partial hepatectomy. The signals which elicit the proliferating response have been partly identified; hormones like insulin and vasopressin and growth factors such as epidermal growth factor (EGF) (Cruise et al., 1985, 1987; Olsen et al., 1988) play a role in this phenomenon, which involves, as a basic mechanism, protein phosphorylation mediated by protein kinases, both dependent on cyclic nucleotides (Laks et al., 1981) and activated by the inositol lipid cycle (Okamoto et al., 1988).

The recently demonstrated presence of protein kinase C (PKC) in isolated rat liver nuclei (Azhar et al., 1987; Capitani et al., 1987; Rogue et al., 1990) and the increased phosphorylation of exogenous histone H1 by PKC occurring after partial hepatectomy (Martelli et al., 1991a), as well as the demonstration that exogenous PKC is able to hyper-phosphorylate some nuclear proteins in regenerating liver (Mazzoni et al., 1992b), suggest that the onset of DNA replication might involve intranuclear PKC. The enzyme is regulated by lipids and lipid-generated cofactors, which can originate from inositol lipids and other phospholipids (Billah and Anthes, 1990). Inositol lipids, and other phospholipid species as well, have been detected by different methods in isolated nuclei (for a review see Maraldi et al., 1992a); therefore, the nucleus owns a sufficient set of factors capable of eliciting the first steps of DNA replication in response to a flow of lipid-dependent signals either generated at the cell surface or transduced to the cell interior via an unknown mechanism.

Regulation of PKC activity in the nucleus, which can be affected primarily by products of the inositol lipid cycle (Divecha et al., 1991; Martelli et al., 1992; Michell, 1992), could depend also on phosphatidylserine (PS), phosphatidylcholine (PC) and other lipids as assumed for cytoplasmic PKC (Billah and Anthes, 1990). Therefore, variations in the nuclear phospholipid content and relative composition may affect the functional activity of this key enzyme.

The main reports concerned with the demonstration of phospholipids in the nucleus stem from fractionation studies, in which the possibility of a partial removal or displacement of phospholipids cannot be completely ruled out, even though the purified nuclear fractions have been convincingly shown to be free of membrane and cytoplasmic contaminations. We have recently demonstrated, by ultrastructural cytochemistry, the presence of phospholipids in defined nuclear domains in whole cells and in tissues (Maraldi et al., 1992b). These nuclear domains, the interchromatin fibers and granules, and the nucleolus, are sites in which transcription products are processed and they belong to the nuclear matrix compartment, with which PKC and some enzymes of the inositol lipid cycle are also associated (Capitani et al., 1987; Payrastre et al., 1991).

It seems thus worthwhile to determine whether, in association with DNA replication, changes in the amount and localization of nuclear phospholipids occur. By in situ ultrastructural cytochemistry, we show here that the nuclear phospholipids decrease in rat liver nuclei in which, as revealed by bromodeoxyuridine (BrdU) incorporation, DNA replication in response to partial hepatectomy takes place.

Male Wistar rats (150-200 g) were operated on as previously reported (Mazzoni et al., 1992b). Rats were injected intraperitoneally with bromodeoxyuridine (BrdU; 25 mg/kg body weight) 1 h before being killed, which was done 22 h after hepatectomy, in correspondence with the maximum uptake of BrdU into DNA, as monitored by flow cytometry (Vitale et al., 1989).

Histological sections (10 μm thick) were incubated with a monoclonal antibody (MoAb) against BrdU (mouse IgG; Becton-Dikinson, Mountain View, CA) and with goat anti-mouse (GAM) IgG linked to 10 nm colloidal gold (cg) particles and then intensified with the Silver Enhancer Kit (Janssen Life Science, Beerse, Belgium).

For electron microscopy immunocytochemistry, liver samples were immediately fixed in 1% glutaraldehyde in 0.1 M phosphate buffer, pH 7.2, for 1 h at 4°C, dehydrated in ethanol and embedded in Epon. Thin sections, mounted on nickel grids, were treated, on one side, by floating them on drops of the following media: 10% H₂O₂ for 10 min etching; 5% normal goat serum for 30 min; anti-BrdU MoAb (1:100) in 0.05 M Tris-HCl, pH 7.6, containing 0.1% bovine serum albumin (BSA) overnight at 4°C; GAM IgG 10 nm cg (1:10) in 0.02 M Tris-HCl, pH 8.2, containing 0.1% BSA for 1 h (Mazzotti et al., 1990). The grids were then incubated on the opposite side for the detection of phospholipids by phospholipase A₂ (PLA₂)-gold complex essentially as previously described (Maraldi et al., 1992b). Briefly, the grids were treated with 0.5% ovalbumin in PBS, pH 8.0, for 10 min and then incubated with PLA₂-17 nm cg for 30 min. The grids were then stained with uranyl acetate and lead citrate.

Controls for BrdU consisted of samples not incubated with the primary antibody or incubated with a non-immune mouse serum instead of the primary antibody (Mazzotti et al., 1990). The controls for PLA₂ consisted of samples to which PC (12 mg/ml) was added as substrate for the PLA₂ gold complex (1:1, v/v) prior to the incubation and of samples treated with non-conjugated gold solution (Maraldi et al., 1992b).

In regenerating liver thin sections, both DNA-replicating cells and quiescent cells were present side by side; the label due to the presence of BrdU (cg particles of 10 nm) identified the cells in which DNA replication occurred, and its distribution pattern allowed us to dissect the S phase. The numbers of cells which were either BrdU positive or BrdU negative were determined on large fields of histological sections, while the different steps of S phase were identified by electron microscopy. The quantitative evaluation of the PLA₂-gold complex labeling (cg particles of 17 nm), which was present in both BrdU positive and BrdU negative nuclei, was performed according to the following criteria: the comparison between resting and replicating nuclei was done on the same sample, in which the experimental conditions were identical; the cg grain count was done manually, since densitometric detection by image analysis also included metal precipitates, which often occur on the thin sections; computer-assisted image analysis was employed to determine the relative areas of the three nuclear domains, i.e. heterochromatin, interchromatin and nucleolus; BrdU positive cells were separately considered according to the labeling pattern for BrdU, identified as early, middle and late S phase (Mazzotti et al., 1990; Rizzoli et al., 1992); the results were expressed as absolute density, i.e. number of particles counted per μm² of the nuclear domain considered.

At 22 h after partial hepatectomy, a peak of incorporation of BrdU into DNA was revealed with an anti-BrdU antibody, by flow cytometry, as previously described (Vitale et al., 1989) (data not reported). At the same time, histological sections, in which BrdU incorporation was revealed by gold-conjugated secondary antibody and silver intensification, showed that about 30% of the cells exhibited characteristic nuclear labeling which appeared diffused, spotted on the inner and perinuclear chromatin, or concentrated as a ring at the nuclear periphery (Fig. 1). At the ultrastructural level, these three main patterns, corresponding to early, middle and late S, as previously demonstrated (Mazzotti et al., 1990), could be identified as well (Fig. 2). Neighbouring resting cells, which did not incorporate BrdU into DNA, were essentially unlabeled, except for the inner space of some mitochondria. The cells with no nuclear labeling, corresponding to resting cells, constituted an internal reference with which to compare the amount of nuclear phospholipids detectable in neighbouring identically treated cycling cells.

The concomitant use of the labeling for DNA and for phospholipids, employing gold particles of different size, allowed the detection of changes in the phospholipid con-tent and localization in nuclei along with the cell cycle.

A comparison among resting hepatocytes and those incorporating BrdU into DNA at different moments of the S phase is reported in Figs 3A,B and 4A,B. In the nucleus of hepatocytes not incorporating BrdU (Fig. 3A) the only label present was that detecting nuclear phospholipids (PLA₂-17 nm gold). The labeling was more intense in the interchromatin than in other nuclear domains (Fig. 3A). In the nucleus of BrdU-incorporating cells (Fig. 3B) 10 nm cg particles appeared mainly in the interchromatin, either scattered or clustered in distinct replicon domains, corresponding to a typical early S pattern. The PLA₂ labeling appeared reduced with respect to resting hepatocytes (Fig. 3B). Similar reduction in the PLA₂ labeling occurred in hepatocytes exibiting a middle S pattern (Fig. 4A) and a late S pattern of BrdU incorporation (Fig. 4B).

The quantitative evaluation of the label density indicated a significant reduction in the phospholipid label in nuclei of BrdU positive compared to BrdU negative cells (Table 1). This drop occurred in all nuclear domains, but was higher in interchromatin than in the nucleolus and heterochromatin, with a reduction of 50% and 30%, respectively, throughout all the steps of the S phase (Table 1).

The amount and composition of the nuclear phospholipids have long been reported to vary in relationship with key metabolic events and mainly with cell proliferation. Increased amounts of phospholipids associated with the chromatin have been found in hepatoma nuclei, compared to normal liver (Coetzee et al., 1975); variations in the relative composition of the chromatin-associated phospholipids occurred in chronic lymphocytic leukemia cells compared to normal B lymphocytes (Manzoli et al., 1977), during the cell cycle in HeLa cells (Alesenko and Burlakova, 1976), and in regenerating liver after partial hepatectomy (Alesenko and Pantaz, 1983). The neutral lipid/phospholipid ratio increased in Ehrlich ascites tumour with respect to normal cells (Awad and Spector, 1976; Balint and Holczinger, 1978); the treatment with insulinlike growth factor I (IGF-I) of mouse Swiss 3T3 cells caused a transient decrease in nuclear inositol lipid phosphorylation (Cocco et al., 1988, 1989); changes in nuclear inositol lipid phosphorylation occurred in Friend erythroleukemia cells (FELC) induced to erythroid differentiation (Cocco et al., 1987; Capitani et al., 1990, 1991). Consistent with these data are the findings that a discrete subcellular localization of the phosphoinositidases β, γ and δ occurs in different cell types, and that the β isoform predominates in the nucleus (Martelli et al., 1992; Mazzoni et al., 1992a).

The results reported here have been obtained by avoiding the cell manipulation required for fractionation studies, and are virtually unaffected by possible drawbaks such as nonhomogeneous or poorly synchronized cells, or artefactual degradation. Our data indicate variations in the nuclear phospholipid content and localization at the level of single cells, with respect to neighbouring cells in the same tissue sample and under the same experimental conditions. A reduction in nuclear phospholipids occurs throughout the entire S phase, suggesting that this lipid modulation is not restricted to the onset of DNA synthesis as a trigger event. The constancy of the values observed in the three stages of the S phase indicates that the phenomenon is required for each replication wave, which occurs in different nuclear compartments involving different levels of chromatin compaction.

The PLA₂-gold method has been compellingly validated for the study of quantitative variations in the total phospholipid content in electron microscope sections. In fact, quantitative variations in labelling have been reported in different membrane regions, known to present a variable phospholipid content (Coulombe and Bendayan, 1989a) or in experimental conditions which affect the phospholipid accumulation (Coulombe and Bendayan, 1989b). In addition, we have previously shown that the PLA₂-gold system used identifies all phospholipid classes used as pure liposomes, and its efficiency is greater for the phospholipids of the nucleus, conceivably forming lipoprotein complexes, than for the membrane phospholipids arranged in bilayer leaflets (Maraldi et al., 1992b). Therefore, the system appears to be suitable for detecting variations in the total phospholipid amount in the nucleus, though a different affinity of the phospholipids for nuclear proteins or nucleoproteins during the S phase cannot be completely ruled out. In regenerating rat liver, an increased amount of the hydrolysis products of phosphoinositides has been found (Kurichi et al., 1992). Preliminary results on rat liver nuclei after partial hepatectomy confirm that an increased cleavage of phosphatidylinositol 4,5-bisphosphate (PIP₂) occurs and that also PC is degraded to phosphorylcholine (Capitani et al., unpublished data). These data suggest that regenerating liver nuclei express specific variations in activity of lipid phosphomonoesterases and phosphodiesterases. The increased activity of these enzymes could produce a rise in the levels of diacylglicerol and inositol trisphosphate (IP₃), which, in turn, could modulate calcium-dependent nuclear PKC activity. The reported activation of nuclear PKC in regenerating liver (Okamoto et al., 1988; Sasaki et al., 1990) has been suggested to affect the phosphorylation levels of some nuclear-specific substrates (Martelli et al., 1991b; Mazzoni et al., 1992b).

The phosphorylation of specific nuclear substrates, such as chromatin proteins by PKC, or the direct effect of IP₃ on intranuclear calcium level, may therefore be considered as a key regulatory element in the initial steps of cell proliferation (Irvine and Divecha, 1992).

In conclusion, the data here reported further support the contention that phospholipids and enzymes related to their breakdown are involved in the generation of signal molecules in the nucleus, which, eventually, induce DNA replication. However, comprehensive understanding of the cascade of events which follows the intranuclear accumulation of the products of the inositide and PC hydrolysis, such as diacylglicerol and inositol phosphates, requires further studies.

We thank Mr. Aurelio Valmori for skilful photographic assistance. This work was supported by grants from Italian C.N.R. PFIG/PFACRO and by Prog. Ric. San. Fin. 1988-1990.