To investigate whether the presence of nuclear proteins in DRMs was an artifact or instead reflected physiologic trafficking events, we modified the extraction procedure described by Solomon et al.
by pelleting out nuclei and intact cells from detergent-free homogenates prior to detergent solubilization. We reasoned that this would reduce potential contamination from nuclear material. As a model system, we have used caveolin-negative LNCaP prostate cancer cells stably expressing myristoylated Akt1 (MyrAkt1)[24
]. MyrAkt1 is a robust marker of DRMs in these cells based on (i) its presence in light buoyant density fractions following sucrose gradient centrifugation; (ii) its enrichment in DRMs following differential detergent extraction; and (iii) its accumulation at the plasma membrane based on localization to membranes that stain with CTxB, a marker of the DRM/lipid raft-restricted ganglioside GM1 (Figure )[31
Figure 1 Modification of the detergent extraction procedure eliminates 'nuclear' proteins from detergent-resistant membranes. (A) Cholesterol- and sphingolipid-enriched membranes in LNCaP/MyrAkt1 cells were stained with 0.5 μg/mL Alexa 594-CTxB for 10 (more ...)
To determine the effect of pelleting nuclei and/or intact cells on the overall protein profiles of isolated fractions, we initially compared the conventional extraction method with the modified procedure. As starting material for the extraction, we generated membrane preparations from LNCaP/MyrHA-Akt1 or LNCaP/LacZ cells in log-phase growth by mechanical disruption of cells under detergent-free conditions, as described in Methods. In the conventional method, homogenized samples were centrifuged at 16,000 × g in a Beckman microcentrifuge for 10 min at 4°C. In the modified approach, samples were centrifuged at 500 × g for 5 min at 4°C to pellet intact cells and nuclei; importantly, no brake was applied during the deceleration of the centrifuge. The supernatant was decanted carefully and centrifuged at 16,000 × g in a Beckman microcentrifuge for 10 min at 4°C to pellet membranes. The pellet remaining from the low-speed spin, comprising predominantly nuclei, was lysed in 50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% Igepal CA-630, 0.5% sodium deoxycholate and 0.1% SDS plus protease and phosphatase inhibitors. In each case (conventional or modified), pellets from the high-speed centrifugation steps were then subjected to successive detergent extraction in Triton X-100 and octylglucoside to yield Triton-soluble (TS) and Triton-insoluble/octylglucoside-soluble (TI) fractions, as described in Methods.
Following determination of the protein concentration of fractions by MicroBCA assay, equal amounts of protein from each fraction were resolved by SDS-PAGE and transferred to nitrocellulose membranes. To verify the presence of protein in each fraction, membranes were stained with 0.1% (w/v) Ponceau S in 5% acetic acid (v/v) prior to blotting with the indicated antibodies. The staining pattern revealed both quantitative and qualitative differences in the overall protein profiles of TS and TI fractions generated by the two methods (Figure ). We confirmed the distribution of MyrAkt1 in TS, TI and nuclear (N) fractions by HA immunoblot, and observed robust enrichment of MyrAkt1 in TI fractions in both cases (Figure ). A small amount of MyrAkt1 was detected in the nuclear pellet from the low speed centrifugation, consistent with previous reports [32
]. To determine the impact of the modified extraction procedure on protein distribution, we probed the membranes for selected proteins with defined subcellular localizations, including proliferating cell nuclear antigen (PCNA), a nuclear protein [33
] and the small G-protein, Giα2
, that resides in lipid rafts [34
]. As shown in Figure , PCNA, although present in the TI fraction following the conventional extraction method, was undetectable in TI fractions prepared with the modified approach. As anticipated, the nuclear fraction showed robust signal corresponding to PCNA. In contrast, Giα2
was present in TI fractions generated with either extraction method, consistent with its localization to DRMs. These data were reproducible across three independent biological replicates and similar findings were obtained with fractions isolated from LNCaP cells stably expressing the irrelevant gene (LacZ
). We also probed for two additional proteins that were identified as putative raft-resident binding partners of MyrAkt1 transiently expressed in HEK293 cells (N.K.M. and R.M.A, unpublished observations): scaffold attachment factor B (SAFB), a protein implicated in transcriptional regulation and chromatin organization [36
]; and heterogeneous ribonucleoprotein K (hnRNP K), a pleiotropic nucleic acid binding protein that regulates gene and protein expression [37
]. Although SAFB was present in TI fractions generated by conventional means, it was absent from this material following elimination of nuclei, strongly suggesting that its presence in the TI fraction arose from nuclear contamination. In contrast, hnRNP K was detected in TS and TI fractions with both extraction methods, as well as in nuclei. These observations are consistent with localization of hnRNP K to multiple distinct subcellular sites [37
To extend our analysis beyond characterization of specific proteins and to determine the potential utility of our modified fractionation procedure for proteomic applications, we used LC/MS as a read-out for the proteome compositions of TI fractions generated using either the conventional or modified methods. The TI fractions from the two extraction methods were subjected to in-solution tryptic digestion and mass spectrometry analysis, as described in Methods. According to the criteria outlined in Methods, 388 and 371 proteins were identified from the TI fractions isolated using the conventional and modified methods, respectively [see Additional file 1
]. This proteomic analysis was not meant to be exhaustive. Instead we used the most confident protein identifications to a) ensure that the lipid raft isolation was successful and b) estimate the fraction of nuclear proteins. Common proteins identified in the TI fractions from both extraction methods included a number of proteins previously reported to be associated with DRMs, lipid rafts and/or caveolae (Table ), such as flotillins [16
], heterotrimeric G-protein subunits [25
], components of the vacuolar ATP synthase protein family [16
] and others [29
]. Thus, the modified fractionation procedure did not alter the predicted composition of DRM fractions significantly and yielded fractions consistent with those isolated by density gradient-based methods.
Proteins identified by tandem mass spectrometry.
To assess the impact of the modified extraction procedure on the protein profiles obtained from each fraction, we classified proteins according to subcellular localization. In particular, we focused on whether proteins were nuclear or non-nuclear in nature. Datasets were annotated using Gene Ontology Cellular Component (GO-CC) terms with the cross-reference converter [45
]. As shown in Figure , we observed a striking difference in the number of nuclear proteins differentially identified with the two approaches. Specifically, the modified extraction procedure led to identification of many fewer nuclear proteins, with only 54 out of 371 proteins (15%) classified as nuclear. This was in marked contrast to the conventional method in which 141 out of 388 proteins (36%) were categorized as nuclear. At the same time, the number of proteins annotated as non-nuclear increased from 195 (50%) to 258 (70%). The Venn diagrams in Figure clearly show that the decrease of nuclear proteins in the modified protocol is concomitant to the increase in non-nuclear proteins. These findings are in agreement with our findings by immunoblot analysis that the nuclear proteins PCNA and SAFB are no longer detectable in TI fractions generated using the modified fractionation technique.
Figure 2 Quantitative analysis of proteins in detergent-resistant membrane fractions analyzed by mass spectrometry. (A) Proteins in DRM (TI) fractions generated using the conventional or modified methods were analyzed by tandem mass spectrometry and classified (more ...)
Interestingly, of the 21 proteins annotated as nuclear in localization, that were found exclusively in the modified TI fraction, evidence from the literature and protein databases suggests that 16 of these are also present in additional subcellular locations, and may shuttle between the nucleus and other sites in the cell [46
]. Thus, only 5 proteins (shaded in Table ), out of a total of 156 specifically identified in the modified TI fraction in this experiment, were present exclusively in nuclei. In addition, at least 60% of the 33 nuclear proteins identified with both approaches are also present in other subcellular locations. In contrast, only about 30% of the 108 nuclear proteins specifically identified in the conventional TI fraction are annotated to be present in additional locations.
These findings strongly suggest that the modified DRM extraction method generates fractions that are largely free of nuclear contamination. They also imply that, given the high frequency with which so-called 'nuclear proteins' are found in other organelles, due to shuttling between compartments, nuclear proteins are likely to be detected by most extraction procedures, even when performed under the most stringent conditions. Canonical nuclear proteins, such as nuclear hormone receptors, can shuttle between the cytoplasm and nuclear compartments as part of their hormone-dependent functions. Recently, the androgen receptor was found to process androgenic signals non-genomically by forming a complex with Akt1 in lipid raft microdomains [59
]. The identification of such proteins in DRMs may thus reflect a legitimate physiologic process and not an experimental artifact.