The Inner Nuclear Membrane Protein Lamin B Receptor Forms Distinct Microdomains and Links Epigenetically Marked Chromatin to the Nuclear Envelope*

Using heterochromatin-enriched fractions, we have detected specific binding of mononucleosomes to the N-terminal domain of the inner nuclear membrane protein lamin B receptor. Mass spectrometric analysis reveals that LBR-associated particles contain complex patterns of methylated/acetylated histones and are de-void of “euchromatic” epigenetic marks. LBR binds heterochromatin as a higher oligomer and forms distinct nuclear envelope microdomains in vivo . The organization of these membrane assemblies is affected significantly in heterozygous ic (ichthyosis) mutants, resulting in a variety of structural abnormalities and nuclear defects. A significant proportion of heterochromatin is localized in the periphery of the cell nucleus and maintains a close spatial association with the inner nuclear membrane (1–5). This spatial association reflects a multiplicity of interactions between chromatin

Using heterochromatin-enriched fractions, we have detected specific binding of mononucleosomes to the N-terminal domain of the inner nuclear membrane protein lamin B receptor. Mass spectrometric analysis reveals that LBR-associated particles contain complex patterns of methylated/acetylated histones and are devoid of "euchromatic" epigenetic marks. LBR binds heterochromatin as a higher oligomer and forms distinct nuclear envelope microdomains in vivo. The organization of these membrane assemblies is affected significantly in heterozygous ic (ichthyosis) mutants, resulting in a variety of structural abnormalities and nuclear defects.
A significant proportion of heterochromatin is localized in the periphery of the cell nucleus and maintains a close spatial association with the inner nuclear membrane (1)(2)(3)(4)(5). This spatial association reflects a multiplicity of interactions between chromatin components and integral or peripheral proteins of the nuclear envelope (NE) 1 (6,7).
Because chromatin is extensively and differentially modified (8,9), it is tempting to think that certain epigenetic marks or factors associated with histone modifying enzymes provide binding sites for NE proteins. However, it is equally possible that transcriptionally active, noncondensed chromatin is subjected to silencing and "heterochromatinization" upon contact to the NE. Both of these hypotheses receive experimental support: chromatin that is silenced through binding to SIRs can tether itself to the NE (10), whereas targeting of marker genes to the inner nuclear membrane suppresses their expression (11).
One of the factors that have been implicated in chromatin anchorage to the NE is the lamin B receptor (12). LBR is a polytopic inner nuclear membrane protein consisting of a long, N-terminal domain, seven or eight hydrophobic transmembrane regions, and a C-terminal tail (13). The N-terminal part of the molecule protrudes to the nucleoplasm and contains multiple serine-arginine motifs that are phosphorylated by the SRPK1 and the cdc2 kinases (14,15); the hydrophobic region represents, instead, a (functional) form of sterol reductase and is involved in cholesterol metabolism (16).
Immunodepletion of LBR from detergent-solubilized NE vesicles results in proteoliposomes with a diminished ability to bind chromatin. Furthermore, direct binding of electrophoretically purified LBR to metaphase chromosomes can be demonstrated by in vitro assays (17). Corroborating these observations, anti-LBR antibodies block nuclear assembly in sea urchin egg extracts (18), whereas direct (19) and indirect (20) interactions with heterochromatin protein 1 (HP1) have been claimed in the literature.
Two critical parameters in LBR-chromatin interactions are the physical state of LBR and the molecular features of LBRassociated chromatin. To address these issues, we have isolated fragments of peripheral heterochromatin attached to the inner nuclear membrane. These subcellular fractions were utilized to affinity select mononucleosomes that associate with LBR and investigate LBR-LBR interactions. The results obtained reveal specific modification "signatures" in LBR-associated histones and indicate that LBR interacts with peripheral heterochromatin as a higher oligomer.

EXPERIMENTAL PROCEDURES
Cells-Chicken and turkey erythrocytes were obtained from whole blood. HeLa cells were grown in Dulbecco's modified Eagle's medium containing 2 mM L-glutamine, 2 mM penicillin/streptomycin, and 10% fetal calf serum. Mouse fibroblasts were maintained in RPMI medium containing 2 mM L-glutamine, 2 mM penicillin/streptomycin, 10% fetal calf serum, and 50 M 2-mercaptoethanol.
Antibodies-The characterization of anti-chicken LBR antibodies has been described previously (17,21). Anti-mammalian LBR antibodies were produced using an antigen His-tagged N-terminal domain of human LBR. Both antibodies were affinity-purified in Affi-Gel matrices to which the N-terminal domains of chicken and human LBR were coupled. Anti-dimethylated Lys 4 and acetylated Lys 14 histone H3 polyclonal antibodies were purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). The characterization of anti-trimethylated Lys 9 histone H3 antibodies has been described (22). Affinity-purified secondary * This work was supported by grants from the Association Francaise contre les Myopathies (France) and the General Secretariat of Research and Technology (Greece) (to S. D. G.), a core grant from the Biotechnology and Biological Sciences Research Council (United Kingdom) (to P. B. S.), a Royal Society (United Kingdom) exchange award (to P. B. S. and S. D. G.), and National Institutes of Health Grant CA20408 (to L. D. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  antibodies and Alexa Fluor goat anti-rabbit and donkey anti-rat IgGs were obtained from Molecular Probes (Eugene, OR).
Cell Fractionation and Preparation of Nuclear Extracts-NEs were prepared from turkey erythrocyte nuclei, essentially as specified by Refs. 17 and 23 with slight modifications. Briefly, after three rounds of DNase I digestion (80 g/ml in 10 mM NaPO 4 , 2 mM MgCl 2 , 10% sucrose, 1 mM DTT, 1 mM PMSF/protease inhibitors for 15 min at room temperature, with 200 g/ml RNase A included in the last digestion step), the resulting nuclear ghosts (NGs) were washed in buffer I (150 mM NaCl, 20 mM Tris-HCl, pH 7.5, 250 mM sucrose, 2 mM MgCl 2 , 0.1 mM EGTA, 1 mM DTT, 1 mM PMSF), sonicated to induce vesiculation, and stored at Ϫ80°C. NE vesicles were further purified in flotation gradients (70, 50, 30, and 20% sucrose cushions; 100,000 ϫ g for 18 h at 4°C), collecting at the 50 -30% interface. The extracts were prepared by thorough resuspension of NE vesicles or sonicated NGs in 300 -600 mM NaCl, 20 mM Tris-HCl, pH 7.5, 250 mM sucrose, 2 mM MgCl 2 , 0.1 mM EGTA, 1 mM DTT, 1 mM PMSF, protease inhibitors, plus or minus 1% Triton X-100 (buffers S and ST, respectively). After ultracentrifugation (200 -350,000 ϫ g for 30 min at 4°C), the soluble extracts (SE, "just salt" extract using buffer S; STE, salt and Triton extract using buffer ST) were collected and either used immediately, or further fractionated in 5-20% sucrose density gradients (100,000 ϫ g at 4°C for 18 h). A euchromatin-enriched fraction was prepared by digesting whole nuclei with DNase I (80 g/ml, as described above). The digest was adjusted to 300 mM NaCl and 1% Triton X-100 and centrifuged at 15,000 ϫ g for 30 min at 4°C. The soluble phase was used in pull-down assays (see below).
Expression and Purification of Recombinant Proteins-Glutathione S-transferase (GST), GST-fused, and His 6 -tagged proteins were ex-pressed in BL21(DE3) cells and purified from lysates according to standard procedures (24).
Microscopy-For light microscopy, the samples were fixed with 1-4% formaldehyde in phosphate-buffered saline, permeabilized with 0.2% Triton X-100, and blocked with 0.5% fish skin gelatin. DNA staining (4Ј,6-diamidino-2Ј-phenylindole-dihydrochrolide, propidium iodide) and probing with the relevant primary and secondary antibodies was performed according to Ref. 25. The specimens were visualized in a Leica SP confocal microscope. An asterisk points to traces of nLBR "leaking" from the column during elution by hot SDS sample buffer (this is commonly seen when using this matrix). Lanes M correspond to markers with the indicated molecular masses. E, electrophoretic profile of DNA extracted from the LBR-GST precipitate (lane 1) and the GST control (lane 2). The positions of the 1000-and 300-bp markers are indicated. F, detection of Lys 14 -acetylated and Lys 9 -trimethylated histone H3 in LBR-precipitated chromatin, as documented by Western blotting. Equal proportions of the nonbound (lanes 1) and the LBR-bound (lanes 2) material are shown. G, in situ binding of purified GST (ϩGST) and LBR-GST (ϩLBR-GST) to permeabilized HeLa cells. The cells were stained with anti-GST antibodies and visualized in a confocal microscope.
Pull-down Assays-GST fusion proteins (LBR-GST, M31-GST, or GST alone; 10 -30 g) were incubated first for 30 min at room temperature with 30 l of glutathione-agarose beads in buffer S or ST (see above under "Cell Fractionation and Preparation of Nuclear Extracts"). After washing three times with the same medium, the beads were combined with nuclear extracts (SE, STE, or a DNase I digest of whole nuclei) and further incubated for 1 h at room temperature. The beads were washed five times with buffer ST and once with isotonic buffer (buffer I, see "Cell Fractionation and Preparation of Nuclear Extracts") before eluting the bound proteins with hot SDS sample buffer.
Purification of Native LBR and Affinity Chromatography in Affi-Gel Matrices-Full-length, NE-associated LBR was isolated by preparative SDS-PAGE and elution from gel pieces, essentially as described in Ref. 17. The eluted protein was reconstituted in 100 mM Hepes-NaOH, pH 7.5, 2 mM EDTA, 0.1% SDS, 2% Triton X-100, 0.2 mM PMSF, 0.05 mM DTT and dialyzed against 100 mM Hepes-NaOH, pH 7.5, 1 mM EDTA, and 0.2 mM PMSF (for 5 h at room temperature). Coupling to Affi-Gel 10/15 was done as recommended by the manufacturer (Bio-Rad). SEs were applied to the Affi-Gel-LBR column and batch-incubated for 90 min at room temperature. After thorough washing with buffer S (15 column volumes), bound material was eluted either with hot SDS sample buffer or with 1 M NaCl, 5 M urea, 40 mM Tris-HCl, pH 8.0, 1% Triton X-100, 1 mM EGTA, 1 mM DTT, and 1 mM PMSF.
Other Methods-Matrix-assisted laser desorption ionization time-offlight mass spectrometry was performed at the Functional Genomics Unit of Moredun Research Institute (Edinburgh, UK). Protein bands were digested with R-specific protease. ⌬M (the difference between the measured and calculated masses) was at the level of 1/10,000. Peak assignment was done either manually or using appropriate Expasy programs. Ambiguous peaks (e.g. those detected in histones precipitated by recombinant or endogenous LBR but not with the complex of recombinant and endogenous protein) were not taken into account. SDS-PAGE, Western blotting, and column chromatography were practiced according to established procedures.

LBR Associates with Peripheral
Heterochromatin-Subnuclear fractions highly enriched in peripheral heterochromatin were isolated from avian erythrocytes. Intact nuclei were extensively digested with DNase I and RNase A, yielding NGs. Consistent with previous observations (26,27), NGs were devoid of nuclear content and retained only a layer of peripheral heterochromatin (Fig. 1A). 2 This "residual" chromatin was cut at multiple sites by DNase I and represented mononucleosomal particles with variable portions of linker DNA (see broad band at 150 -200 bp in Fig. 1B). As could be expected (8), NEassociated chromatin was essentially depleted of Lys 4 -dimethylated histone H3 but possessed Lys 9 -trimethylated histone H3 (Fig. 1D).
On a mg/mg basis, NGs contained significantly more histones than NE proteins (e.g. lamins and LBR; Fig. 1C). However, upon further fractionation in sucrose flotation gradients, loosely attaching chromatin dissociated from the inner nuclear membrane and the material collecting at the 30 -50% interface (nuclear envelope-peripheral heterochromatin fraction (NEPH)) possessed almost stoichiometric amounts of histones and NE proteins (Fig. 1C).
To examine whether LBR associates with a specific fraction of chromatin, we performed pull-down experiments using exogenous (recombinant) LBR. As shown in Fig. 2 (A, B, and E), a GST fusion protein corresponding to the N-terminal domain of LBR precipitated roughly stoichiometric amounts of heterochromatin-derived core particles obtained from NG and NEPH salt extracts. However, unlike recombinant HP1 (M31), recombinant LBR did not precipitate nucleosomes when NEPH extracts were substituted with euchromatin-enriched DNase I digests of whole nuclei (Fig. 2C). The binding of HP1 to particles released by DNase I digestion is consistent with recent observations that reveal a role of this protein in the organization of both hetero-and euchromatin (28). On the other hand, the selective association of LBR with nucleosomes that cofractionate with the NE indicates a preference for peripheral heterochromatin. Consistent with this idea, when permeabilized HeLa cells were incubated with recombinant LBR, the nuclear periphery was intensely decorated, whereas under the same conditions, the staining of internal nuclear sites was rather faint (Fig. 2G).
As expected from previous studies (17), native (endogenous) LBR purified by preparative electrophoresis and coupled to Affi-Gel also "selected" a fraction of nucleosomes contained in SEs (Fig. 2D, lanes 2 and 3). This enabled comparisons of the chemical modifications in histones H3/H4 associated with either the native or the recombinant protein (see below).
Modification Signatures of NE/LBR-associated Chromatin-To characterize the LBR-associated chromatin fraction, we excised the histone H3/H4 bands from SDS gels (similar to those shown in Fig. 2, A, B, and D) and subjected the samples to mass spectrometry (matrix-assisted laser desorption ionization time-of-flight MS). To be more systematic, we also analyzed histones co-precipitating with a complex of recombinant and endogenous LBR (for details see Fig. 3A). As shown in Table I, nucleosomes affinity-selected by LBR and nucleosomes extracted from NEs yielded almost the same histone H3/H4 modification signatures. Some of the peaks detected in the histones that co-precipitated with recombinant or a complex of endogenous and recombinant LBR were not detected in the samples eluted from Affi-Gel columns containing exclusively endogenous LBR. However, these differences were minor and did not alter the overall picture emerging from the mass spectrometric data (for instance, most of the missing peaks corresponded to either nonmodified or partially modified peptides and not to fully modified ones).
LBR or NE-associated particles exhibited features that clearly differentiated them from bulk chromatin (29 -31). First, Lys 4 of histone H3 was not modified, consistent with the indirect immunofluorescence data presented in Fig. 1D; second, although histone H3 monoacetylated in Lys 18 or Lys 23 could be readily identified, species acetylated in both residues were not detected. This was observed with several different preparations and did not change when the chromatin fractions were isolated in the presence of sodium butyrate.
In LBR-associated nucleosomes, the histone fold domain of H3 and H4 was largely nonmodified (with the exception of Lys 79 in histone H3). Lys 27 and Lys 36 of H3 were modified by trimethylation, whereas H4 was dimethylated in Lys 20 and partially or fully acetylated at Lys 5 , Lys 8 , Lys 12 , and Lys 16 . Histone H3 was heavily modified in the region 9 -17, which contains Lys 9 and Lys 14 . Because the mass difference corresponding to lysine trimethylation is equal to that of lysine acetylation, a peak at 971.5 daltons could be assigned either to a dimethylated/acetylated or to a dimethylated/trimethylated 9 -17 peptide. Likewise, another peak at 985.5 could be attributed either to a twice trimethylated or to a trimethylated/ acetylated peptide. Pull-down assays with recombinant LBR and Western blotting with anti-histone modification antibodies showed that H3 subspecies trimethylated at Lys 9 or acetylated Lys 14 were present in the LBR precipitate, albeit to a different extent (Fig. 2F). Although this does not prove that LBR-associated nucleosomes contain histone H3 that is simultaneously modified at Lys 9 and Lys 14 , it can be safely concluded that acetylation in Lys 14 is not incompatible with heterochromatin association and LBR binding.
LBR Self-associates under in Vitro Conditions-When salt/ detergent extracts (STE) were used in the pull-down experiments instead of just salt SEs, we noticed that recombinant LBR consistently precipitated a protein co-migrating with endogenous, full-length LBR and traces of nuclear lamin B (Fig.  3A). The presence of endogenous LBR in the LBR-GST precipitate could be confirmed by Western blotting (Fig. 3B) and explained either by "cross-linking" through nucleosomes/lamin B and/or direct LBR-LBR binding (oligomerization). To distinguish between these possibilities, we performed the pull-down assay at 0.3 M salt and subsequently washed the precipitates with increasing NaCl concentrations (notice that co-precipitation of native LBR and nucleosomes with recombinant LBR is abolished at 0.6 M NaCl; Fig. 3A). As illustrated in Fig. 3C, native LBR was retained in the LBR-GST precipitate after washing with 0.4, 0.45, or 0.5 M salt, when all of the nucleosomes had detached. Furthermore, direct evidence for in vitro oligomerization was obtained using GST-tagged and His 6tagged LBR. As shown in Fig. 4A, the two forms of recombinant LBR co-precipitated. In addition, LBR-GST/LBR-His 6 complexes were readily detectable in nondenaturing acrylamide gels (Fig. 4B). Finally, LBR-His 6 could be cross-linked by glutaraldehyde, yielding dimers and higher order oligomers. Under the same conditions, a control protein of analogous charge and molecular mass (cytochrome c) was not cross-linked (Fig. 4C).
LBR Resides in Distinct NE Microdomains-That the Nterminal domain of LBR oligomerized in vitro suggested that the native protein might form microdomains at the level of the NE. To explore this idea, we proceeded with morphological experiments. A striking, "lumpy" fluorescence pattern was observed when avian erythrocytes were decorated by chicken LBR-specific, affinity-purified antibodies and examined by confocal microscopy (Fig. 5, C and D). This discontinuous pattern contrasted the smooth, uniform staining of the nuclear periphery with anti-lamin A and anti-lamin B antibodies (Fig. 5, A  and B) and was apparent at both xy and xz optical sections (see below). Discrete LBR "islets" could also be discerned when Triton-permeabilized erythrocytes were examined by pre-embedding immunoelectron microscopy (Fig. 5E).
Heterozygous Mutations Affect LBR Organization and Nuclear Structure-To study in more detail LBR organization in vivo, we examined fibroblasts from wild type and ichthyosis (ic) mouse mutants. Some of these mutations (ic J ) result in a frame shift that changes amino acids 365-385 and introduces a stop codon at residue 386 of the LBR gene. Homozygous animals do not contain immunohistochemically identifiable LBR in the inner nuclear membrane (32), because, most likely, the mutant protein is degraded shortly after synthesis on ER membranes. However, we reasoned that heterozygous ic mutants might be informative; because the truncated ic-LBR has an intact Nterminal domain, oligomerization with wild type LBR may rescue some of this protein and produce an intermediate, analyzable phenotype.
Staining of wild type fibroblasts with affinity-purified, antimammalian LBR-specific antibodies revealed that the intact protein is localized in NE microdomains, very much like its avian counterpart (Fig. 6, wild type, A-D). The LBR islets were of considerable size (500 -800 nm) and in xz sections produced a striking "palisade" pattern along the nuclear periphery (panels E-G). In heterozygotic cells, the rim fluorescence pattern was not immediately obvious; instead, in 50% of the cells LBR appeared to form large nucleoplasmic "blocks" that superficially resembled foci of pericentric heterochromatin or nucleoli (Fig. 6, heteroz. mutant, B and G). Upon closer inspection, it became clear that these LBR aggregates were adjacent to but not coincident with internal heterochromatic foci (Fig. 6, A-D). Furthermore, double staining with anti-LBR/anti-lamin antibodies (Fig. 6, E and F) and confocal series along the xz axis (Fig. 6, K-M) revealed that these blocks, which replace the more or less evenly distributed LBR microdomains, represented large patches of NE-associated material.
As expected, besides the effects on the subnuclear distribution of LBR, some of the mutant cells exhibited gross sorting defects, with a fraction of LBR ending up in adhesion plaques (Fig. 6J, mis-targeting; ϳ60% of the cells). Finally, consistent with the critical role of LBR in nuclear reassembly and architecture (17), some of the mutant cells, often in groups, appeared to have irregularly shaped nuclei (Fig. 6H, lobulation; ϳ30% of the cells). Taken together, these results reinforce the idea that LBR self-associates via its N-terminal domain and forms distinct microdomains in vivo. Apparently, the fine structure of these assemblies is greatly affected in ic cells, in which the protein encoded by the altered allele behaves as a dominantnegative mutant.

DISCUSSION
Features and Epigenetic Marks of LBR-associated Chromatin-To isolate peripheral heterochromatin, we have taken advantage of previous observations (26,27) showing that the central "core" of chromatin is effectively removed from the avian erythrocyte nucleus upon DNase I digestion. The choice of nuclease seems to be critical; in comparative experiments we have found that microccocal nuclease is less efficient than DNase I and, depending on the conditions, yields variable results.
The NGs obtained after DNase I digestion were found to contain a more or less intact NE, underlined by a layer of peripheral heterochromatin. With controlled sonication, these "shells" are converted to small vesicles that have an inside-out orientation (17,27) 2 and can be further fractionated by flotation in sucrose density gradients (NEPH fraction).
We have shown here that mononucleosomes extracted from NEPHs and NGs bind almost stoichiometrically to LBR. Mass spectrometric analysis indicates that LBR-associated particles contain histone H3 that is methylated at Lys 9 and Lys 27 , monoacetylated in Lys 18 and Lys 23 , and nonmodified at all at Lys 4 . The same type of analysis reveals more complicated patterns of post-translational modification, such as trimethylation of Lys 9 combined with acetylation (or, less likely, trimethylation) at Lys 14 (histone H3) and partial acetylation of Lys 5 , Lys 8 , Lys 12 , and Lys 16 (histone H4). Some of these combinations, which clearly occur in native heterochromatin, would not be anticipated from the currently established rules of the "histone code" (8).
Because in vitro acetylation of purified histones H3/H4 by CREB-CBP reduces significantly in vitro binding to LBR (20), nucleosomes containing such acetylated species are not expected to associate directly with the inner nuclear membrane. However, as we argued previously (20), the binding properties of purified histones or histone tail peptides might be different from that of native nucleosomes; sites "occluded" or buried in the fully assembled particle could be partially exposed when histones are dissociated from one another. In addition, long range interactions between different "modification cassettes" (9) may not occur when histone peptides are used instead of the intact H3/H4.
It is theoretically possible that nucleosomes containing partially acetylated histones associate with LBR through "bridging" particles, which consist of nonacetylated histones and are laterally connected to the former via nucleosome-nucleosome interactions. However, this possibility is remote, because the preparations used in this study contained mononucleosomes, whereas binding of core particles to LBR was almost stoichiometric (nonstoichiometric binding should be expected if LBR associates with arrays of polynucleosomes instead of mononucleosomes).
LBR Self-association and NE Microdomains-Accumulating evidence suggests that several integral proteins of the NE (e.g. LAP1, LAP2, and LBR) are organized as multi-subunit complexes and may form chromatin-remodeling platforms. In rat hepatocytes, LAP1 localizes in regularly spaced clusters and is co-immunoprecipitated with B-type lamins and a calcium-dependent kinase (33). Along similar lines, LAP2B can be crosslinked and co-precipitated with at least four other nuclear proteins, including LBR and HA95 (34). LBR is also co-isolated from avian erythrocytes together with the nuclear lamins, an SR kinase (SRPK1), a splicing factor-associated element (p32/p34), and a membrane protein related to peripheral benzodiazepine receptors (35,36). Finally, LBR binding to HP1 (19), or a HP1core histone subcomplex (20), occurs under certain conditions. At this point, it is not clear whether inner nuclear membrane proteins participate in the formation of these complexes as individual monomers or as higher oligomers. Using in vitro assays, we have demonstrated that (at least) LBR self-associates through its N-terminal domain and forms specific oligomers under stringent ionic conditions (i.e. at 0.15-0.5 M salt). Furthermore, employing two different affinity-purified antibodies and a range of morphological techniques, we have shown that native LBR resides in distinct NE microdomains. From the morphological analysis of heterozygous LBR mutants, it looks likely that the normal distribution and fine substructure of LBR microdomains are affected significantly in pathological conditions, such as ichthyosis or the Pelger-Huet anomaly (32). Therefore, heterozygous animals provide an excellent model system for further analysis of LBR-LBR and LBR-chromatin interactions.
Because native heterochromatin contains arrays of tandemly repeated elements (e.g. nucleosome chains with specifically modified histone tails), it is perhaps not surprising that its NE anchor, LBR, may also exist in the form of a multimeric particles. Nonetheless, one idea that needs to be investigated further is whether the hydrophobic domain of LBR, in its capacity as a sterol reductase (16), modifies the local lipid environment and supports the formation of specific LBR "rafts." So far, attempts to isolate such proteolipid assemblies by flotation in sucrose gradients have not been successful. However, this task might be more complicated than it seems, because the composition of LBR-associated lipids and the stability of the putative rafts to different detergents have not yet been systematically explored.