Novel localization of the DNA-PK complex in lipid rafts: a putative role in the signal transduction pathway of the ionizing radiation response.

Increased sensitivity to ionizing radiation (IR) has been shown to be due to defects in DNA double-strand break repair machinery. The major pathway in mammalian cells dedicated to the repair of DNA double-strand breaks is by the nonhomologous end-joining machinery. Six components function in this pathway, of which three (Ku70, Ku86, and DNA-PKcs) constitute a protein complex known as DNA-dependent protein kinase (DNA-PK). However, it is now recognized that the cellular radiation response is complex, and radiosensitivity may be also regulated at different levels in the radiation signal transduction pathway. In addition to DNA damage, exposure to IR triggers intracellular signaling cascades that overlap with pathways initiated by ligand engagement to a receptor. In this study, we provide evidence for the novel localization of the DNA-PK complex in lipid rafts. We also show this property is not a generalized characteristic of all DNA repair proteins. Furthermore, we have detected Ku86 in yeast lipid rafts. Our results suggest that the components of this complex might be recruited separately to the plasma membrane by tethering with raft-resident proteins. In addition, we found an irradiation-induced differential protein phosphorylation pattern dependent upon DNA-PKcs in lipid rafts. Thus, we speculate that another role for the DNA-PKcs subunit and perhaps for the holoenzyme is in the signal transduction of IR response.

One of the most damaging lesions that can occur in a cell is a DNA double-strand break (DSB), 1 because it disrupts both strands of the DNA molecule. The importance of rescuing the cell from this lesion is evident by the existence of evolutionarily conserved DNA repair systems that act upon DSBs that are generated under physiological conditions, such as during transposition, meiosis, and recombination, or by external insults like ionizing irradiation (IR) exposure. The major pathway in mammalian cells dedicated to the repair of DSBs is by the nonho-mologous end-joining machinery (NHEJ). Six components function in this pathway, of which three (Ku70, Ku86, and DNA-PKcs) constitute a protein complex, termed DNA-dependent protein kinase (DNA-PK) (for reviews, see Refs. [1][2][3][4]. Characterization of the DNA-PK catalytic subunit (DNA-PKcs) has revealed that it is a Ser/Thr protein kinase, a member of the phosphatidylinositol 3-kinase superfamily, and that it must be bound to DNA in order to be activated. Other protein components of the NHEJ pathway are XRCC4 (5), DNA-Lig IV (6,7), and Artemis (8). There is additional evidence for further functions either for the NHEJ or for roles of the component proteins separate from their function in NHEJ (4, 9 -22).
In addition to DNA damage, exposure to IR triggers intracellular signaling cascades that overlap with pathways initiated by ligand engagement to a receptor (23)(24)(25)(26)(27)(28)(29)(30)(31)(32). A variety of kinases including protein kinase C (PKC) and Raf-1 kinase are activated by IR treatment (33)(34)(35)(36)(37), which in turn supports radiation-induced cascades emanating in the membrane and mediated by the activation of cytoplasmic molecules. In the same context, a signaling pathway conserved from yeast to humans associated with the response to diverse stresses involves the generation of ceramide from sphingomyelin by, at least in part, the action of sphingomyelinase (38).
It is now becoming clear that lipid microenvironments on the cell surface, known as detergent-insoluble glycolipid-enriched complexes (DIGs) or lipid rafts, also take part in signal transduction processes. This lipid environment favors specific protein-protein interactions between ligands, receptors, and kinases, resulting in the activation of signaling cascades (for reviews, see Refs. 39 -43). Proteins incorporated into DIGs fall into different categories: glycophosphatidylinositol-anchored proteins, transmembrane proteins, doubly acylated tyrosine kinases of the Src family, and the ␣-subunit of heteromeric GTP-binding proteins and cholesterol linked proteins like Hedgehog (for a review, see Ref. 43).
Curiously, another member of the phosphatidylinositol 3-kinase family, which has been also associated with the response to IR, the ATM protein, has been reported to localize outside the nucleus in cytoplasmic and membrane-associated vesicles (44). Results from the same study showed that lymphoblast cells from patients carrying a mutation in the ATM gene (locus associated with a human hereditary disease known as ataxia telangiectasia) were defective in the IR-induced activation of a raft-resident protein-tyrosine kinase p53/56 Lyn.
In this study, we provide evidence for the localization of the DNA-PK complex in lipid raft of mammalian cells but not as a generalized characteristic of NHEJ factors. Furthermore, the localization of Ku86 in lipid rafts is conserved in yeast. Because the protein phosphorylation pattern after irradiation is different in lipid rafts isolated from DNA-PKcs-defective cells compared with controls, we are tempted to speculate that besides the role that the DNA-PK has in the DSBR pathway in mammalians, this complex might be involved in signal transduction of IR response.
Horseradish peroxidase-conjugated secondary antibodies against rabbit and mouse IgG were purchased from Amersham Biosciences, and antibodies against goat IgG were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA).
Monoclonal antibody generated in rat and directed against the Cterminal part of yeast Ku86 was a gift from Dr. Heidi Feldmann (Institut fuer Biochemie der Universitate Muenchen, Germany). All other chemicals were purchased from Sigma.
The mixture of inhibitors of proteases was as follows: 0.5 g/ml leupeptin and aprotinin; 40 g/ml bestatin, and 0.5 mM Pefabloc SC. The mixture of inhibitors of phosphatases was as follows: 10 mM disodium pyrophosphate, 25 mM disodium ␤-glydcerolphosphate, 1 mM sodium vanadate, 50 mM sodium fluoride.
Cell Culture-The xrs-6 cell line was derived from the CHO-K1 cell line on the basis of its sensitivity to IR due to a defect in Ku86 (52). MO57J and MO57K were isolated from different areas of a glioblastoma tumor and further characterized. MO57J is radiosensitive due to a defect in DNA-PKcs as described elsewhere (50).
Preparation of Microsomal Fraction-Cells (ϳ5 ϫ 10 7 ) were detached with trypsin, washed with ice-cold PBS, and homogenized by Dounce disruption at 4°C in Tris-HCl, pH 7.5, EDTA 5 mM with protease inhibitors. Cell integrity was checked under microscope using an exclusion dye trypan blue. Before centrifugation at 700 ϫ g, the solution was brought up to 440 mM sucrose. The postnuclear supernatant was centrifuged further, first at 12,000 ϫ g for 15 min at 4°C to pellet mitochondria, and this supernatant was centrifuged again at 120,000 ϫ g for 1 h at 4°C. The pellet (microsomal fraction) was separated from the supernatant (cytosolic fraction) and resuspended in 500 l of the above buffer containing sucrose and stored at Ϫ80°C.
Conditions such as high salt or extreme pH (pH 11.5), which disrupt protein-protein interactions, cause dissociation of membrane-associated proteins but have minimal effects on integral membrane proteins. Treatment (100 l of the microsome fraction) was performed on ice for 1 h, membranes were centrifuged at 120,000 ϫ g, and pellet was separated from supernatant. Pellet was resuspended in 140 l of the above buffer containing sucrose. An equal amount of protein (3 g) was analyzed by SDS-PAGE.
Supernatant was dialyzed against cold PBS for 2 h, and protein was precipitated by adding 2 volumes of ethanol and incubating overnight at Ϫ20°C. After centrifugation for 10 min in a microcentrifuge, the supernatant was removed, the pellet was resuspended in 80 l of loading buffer, and an equal volume (40 l) was analyzed by SDS-PAGE.
Isolation of Lipid Rafts-Because of their high lipid content (sphingolipid and cholesterol), lipid rafts or DIGs are insoluble in the detergent Triton X-100 at 4°C (but soluble at 37°C), and they float to a low density during gradient centrifugation.
We prepared lipid rafts using the method described by Bagnat et al. (79) with some modifications. Briefly, 1-2 ϫ 10 7 adherent cells were detached by trypsination and washed in PBS, and the cell pellet was homogenized in TNE supplemented with Triton X-100 to a final concentration of 1% (v/v) in the presence of a mixture of protease and phosphatase inhibitors and incubated for 30 min on ice. Extracts were brought to 40% Optiprep (Nycomed Pharma, Oslo, Norway) and over-laid with 2 volumes of 30% Optiprep in TNEX and a top layer of 1 volume of TNEX. Samples were spun for 5 h at 4°C either in a RP55.5 rotor at 55,000 rpm (5 ϫ 10 6 cells) or in a SW55.1 rotor at 45,000 rpm (1-3 ϫ 10 7 cells). Fractions of 200 -400 l were collected from the top of the gradient. DIGs are present in fractions 1-4, and the soluble component is represented by fractions 9 -12.
Western Blots-Whole cell extracts or lipid rafts fraction (20 l) were boiled in SDS-PAGE loading buffer and separated by SDS-PAGE (5 or 10% polyacrylamide). When fraction volume needed to be increased (60 l), precipitation with 2 volumes of ethanol and incubation at Ϫ20°C overnight was performed. Proteins were transferred to polyvinylidene difluoride membrane by using a wet apparatus and blocked for 1 h to overnight at 4°C with 5% (w/v) skim milk (Carnation) in TBST buffer solution.
[ 32 P]Orthophosphate Metabolic Labeling-Cells were detached with trypsin, washed in Dulbecco's modified Eagle's medium high glucose PO 4 medium (Invitrogen) without serum, and irradiated with 500 rads (5 grays). After irradiation, cells were recovered for 2 h. at 37°C in rotator, and an aliquot 5-6 ϫ 10 6 cells was resuspended in 1 ml of the same medium containing 1 mCi of [ 32 P]orthophosphate (PerkinElmer). Cells were washed in PBS, and the pellet was stored at Ϫ80°C.
Preparation of lipid rafts was performed as described above, and 100 l of the fractions of interest were precipitated with ethanol and resuspended in 60 l of 0.1 M Tris (pH 8.9), 1.1% (w/v) SDS before performing two-dimensional PAGE.
Two-dimensional PAGE-25 l of lipid raft fractions (fraction 3 or 4) were mixed with 300 l of rehydration buffer containing thiourea (Plu-sOne Reagents) according to protocols provided by Amersham Biosciences.
Isoelectric focusing was performed using Immobiline dry strips (18-cm strips; pH range from 4 to 7; Amersham Biosciences). The dry strips were rehydrated with the solubilized protein sample by in gel reswelling (12 h) and were electrophoresed at 20°C for 8 h following conditions provided by the manufacturer. The second dimension was run in a 12.5% polyacrylamide SDS gel using a Bio-Rad XiII cell.
Silver staining was performed using a kit from Amersham Biosciences. Gels were dried and exposed to Biomax MS film (Eastman Kodak Co.). Autoradiography and gels were scanned using a flat bed scanner (Amersham Biosciences).

DNA-PK Complex Is Localized in Lipid Rafts-In an attempt
to unravel the role that the DNA-PK may have in the signal transduction pathway in response to IR treatment, we investigated its localization in lipid rafts or DIGs. While technical variations are employed to isolate DIGs, one set of conditions, based on the solubilization of membranes in 1% Triton X-100 at 4°C followed by flotation in sucrose or Optiprep gradients, has become the benchmark protocol within the field.
As shown in Fig. 1 (A-C), we localized components of the DNA-PK complex (Ku70, Ku86, and DNA-PKcs) in detergentinsoluble fractions, DIGs. To demonstrate that this was not a general phenomenon of members of the NHEJ machinery, a control was included for another member of this DNA repair pathway, XRCC4. As shown in Fig. 1D, the XRCC4 signal was only present in the soluble fractions. This result strongly suggests that the localization of the DNA-PK complex serves additional roles besides its main function in DSBR. Other controls included membrane hybridization with Ab against proteins reported to be present in lipid rafts like Lyn and caveolin-1, a membrane protein Na ϩ /K ϩ ATPase, which is excluded from DIGs and tubulin as a representative soluble counterpart (see Fig. 1, A and E). We have extended these studies to other gene products, which belong to other DNA repair pathways besides NHEJ. Among those, the mismatch repair genes MSH2 and MSH6, which have been localized in soluble fractions (data not shown) (45). Furthermore, the helicase BLM-1, which has been identified as the mutated target responsible for Bloom's syndrome in humans, is restricted to soluble fractions as well (data not shown). Collectively, these results show that the localization of DNA-PK complex is not a ubiquitous event of DNA repair components.
The choice of detergent solubilization conditions is critical to the identification of DIG components. Retention of prominin in microvilli has been recently reported by using another nonionic detergent, Lubrol WX, at 4°C, revealing a distinct cholesterolbased lipid microdomain (46). When experiments were repeated using Lubrol WX as the solubilizing detergent, results were indistinguishable from the ones achieved by using Triton X-100 (data not shown).
To fulfill one of the criteria of the operational definition of DIGs, detergent extraction was performed instead at 37°C, and the gradient was run at room temperature. Results in Fig. 2 show, as expected, a shift to soluble fractions consistent with the identification of a protein as a component of lipid rafts.
In addition, if cholesterol is extracted by treatment with methyl-␤-cyclodextrin (MCBD), raft domains are perturbed, and raft proteins usually, but not always, become detergentsoluble. Because MCBD is very toxic to the cells, treatment conditions are restricted to low concentration of this chemical to maintain cell viability. However, lipid rafts are first assembled in the Golgi, and movement out of the Golgi seems to be mainly toward the plasma membrane. Thus, mild conditions for in vivo treatment might be insufficient to compete with the intracellular source of rafts contributed by Golgi vesicles. To circumvent this problem, we fractionated cells and isolated microsome fractions and treated them with a higher concentration of MCBD than the one normally used for in vivo studies. As shown in Fig. 3 (A-C), MCBD treatment under these conditions shifted the pattern of all of the subunits of DNA-PK toward the soluble fraction of the gradient.
Finally, the localization of DNA-PK components in lipid rafts was confirmed in at least three independent preparations of the human cell line MO57 K and J and extended to other human cells (lymphocytes and tonsil), mouse embryonic fibroblasts, and Chinese hamster ovary (CHO) cell lines.
Independent Localization of DNA-PK Subunit into Lipid Rafts-DNA-PK is a heterotrimeric protein complex composed of Ku70, Ku86, and DNA-PKcs, which are assembled at DSBs. It has been previously shown that DNA-PKcs and Ku do not associate in the absence of DNA ends (47,48); however, a recent report has challenged this notion (49).
In an attempt to understand the mechanism by which the DNA-PK complex is localized in lipid rafts, we prepared DIGs from human cells defective in DNA-PKcs, MO57J, and from controls MO57K (50), from CHO cell line xrs-6, which we characterized previously as carrying a mutation in the Ku86 gene (51,52), and from its parental cell line CHO-K1. We and others demonstrated that the absence of Ku86 impairs the recruitment of DNA-PKcs to DNA ends. Thus, xrs-6 cells lack kinase activity, which in turn renders the cell line radiosensitive and defective in V(D)J recombination. In addition, the absence of Ku86 affects the stability of Ku70, and reduced levels of Ku70 are seen in xrs-6 when compared with parental control CHO-K1 (53).
As shown in Fig. 4, the absence of DNA-PKcs does not affect localization of Ku70 (A and B) or Ku86 (not shown) into lipid rafts. Similarly, mutation in Ku86 does not impact on the localization of DNA-PKcs into lipid rafts (D-E). It is worth noting that the level of the DNA-PK complex in rodents is nearly 20-fold less than in humans, explaining the reduced signal in Fig. 4, D-E, even after concentration of the sample. These results show that the DNA-PK complex needs not be preassembled to be localized into DIGs and suggest an independent interaction of the subunits directly or indirectly with the plasma membrane proteins.
Finally, one of the most important properties of lipid rafts is that they can include or exclude proteins to a variable extent in response to a stimulus. This regulatory aspect of DIGs was initially reported in studies on IgE receptors (for a review, see Ref. 43). To investigate whether the response to low dose irra-diation promotes relocation of any DNA-PK component into DIGs, we prepared lipid rafts from cells treated with IR and nonirradiated controls and compare the protein distribution pattern in representative fractions of the gradient. As shown in Fig. 4 (A-C), relocation of any of the subunits of the DNA-PK was not identified after 2 h of recovery postirradiation (5 grays).
DNA-PK Components Are Mainly Membrane-associated Proteins-None of the subunits of the DNA-PK complex contain a trans-membrane domain that might explain their anchoring in the plasma membrane.
To test the relative strength of association between DNA-PK protein subunits and cell membranes, we extracted the microsome fraction with 0.1 M sodium carbonate, pH 11.5, and with a strong chaotropic salt, KBr (2 M). Supernatants and extracted membrane pellets were analyzed for the presence of component proteins. We organized the resulting immunoblots from least extractable to most tightly bound (left to right in pellets and right to left in supernatant) (see Fig. 5). As negative controls we included three membrane proteins, epidermal growth factor receptor (EGFR), calnexin, and Lyn, which, as expected, showed little signal in the supernatant of sodium carbonatetreated membrane.
Because a significant portion of the DNA-PK subunits (DNA-PKcs and Ku86 panel A and Ku70 panel B) can be extracted with sodium carbonate, we concluded that these subunits are peripheral membrane proteins that are tightly associated with lipids and/or integral membrane proteins.
Ku86 Is Also Localized in Lipid Rafts Isolated from Yeast-DNA-PKcs homologues have not been identified in nonvertebrates. Since the complete Saccharomyces cerevisiae genome sequence is now known, it can be concluded that a homologue does not exist. The presence of DNA-PKcs in different species correlates with the ability of the species to carry out V(D)J recombination. Thus, DNA-PKcs may be an evolutionary late addition to enhance this rejoining process. However, homologues of Ku70 and Ku86 have been identified in yeast, and a similar function to its mammalian counterpart has been reported (54,55).
To investigate whether the localization of Ku86 is conserved in lower eukaryotes, studies were performed in lipid rafts isolated from yeast. As shown in Fig. 6, a signal corresponding to yeast Ku86 is developed using a specific Ab raised against the C terminus of yeast Ku86. To control for the specificity of this Ab, whole cell extract prepared from a yeast mutant in which the Ku86 gene has been disrupted by homologous recombination was included (54). It is worth noting that this process of mutagenesis generated a truncated version of the protein that lacks approximately the first 130 amino acids, is stable, and can be visualized by this Ab (see Fig. 6). We interpreted this result as an indication that the N terminus fragment of the yeast Ku86 protein is dispensable for the localization in lipid rafts.
Differential DNA-PKcs-dependent Phosphorylation Pattern of Proteins Resident in Lipid Raft in Response to IR Exposure-In an attempt to understand the physiological role that DNA-PK complex might have in the plasma membrane, we compared the protein phosphorylation pattern of lipid rafts isolated from a human cell line defective in DNA-PKcs (MO57J) and control (MO57K) after irradiation. This was achieved by in vivo labeling with [ 32 P]orthophosphate and lipid rafts isolated and separated by two-dimensional PAGE. Signals are detected by autoradiography, and differences are shown in Fig. 7A (see arrows). To control for equivalent protein loading, silver-stained pattern of two-dimensional gels showed equivalent protein loading (see Fig. 7B).
We interpreted these results as an activation of DNA-PK directly or indirectly in response to IR treatment that leads to a phosphorylation of raft proteins. This, in turn, leads us to speculate that a novel role that this complex may have is in signal transduction of the IR response.
Experiments are in progress to identify these substrates present in lipid rafts by mass spectroscopy. In the same context, a differential phosphorylation pattern has been observed from rafts isolated from the CHO cell line V-3, defective in DNA-PK activity, when compared with its parental control AA8 (data not shown).

DISCUSSION
Increased sensitivity to IR has been shown to be due to defects in DSBR and mutations in the proteins that detect DNA damage. However, it is now recognized that the cellular radiation response is complex, and radiosensitivity may be regulated at different levels.
Radiation also induced rapid activation of acid sphingomyelinase rapid sphingomyelin, hydrolysis, and ceramide production in a nuclear free cell lysate, indicating a direct effect of radiation on cytoplasmic membrane, independent of the effect on nuclei. In addition, sphingomyelinase activities have also been observed in sphingomyelin-rich plasma membrane microdomains, caveolae, which are a specialized subset of lipid rafts (for reviews, see Refs. 38 and 56). A direct relationship The two lanes on the left were loaded with whole extract prepared from wild type and yeast Ku86 mutant, and the arrows indicate the major bands detected by the Ku86 Ab, the upper one being the wild type (thick arrow) and the lower one the truncated mutant (thin arrow). B, fractions isolated from mutant yeast strain. M, molecular mass marker (in kDa); mut., yeast Ku80 mutant generated by homologous recombination (described under "Experimental Procedures"); wt, wild type. between resistance to radiation-induced apoptosis and defective ceramide signaling has been established.
In addition, a common product of the exposure to IR is the generation of reactive oxygen species, which in turn generates hydrogen peroxide (H 2 O 2 ). Many receptors (EGFR, B cell receptor, and insulin receptor kinase to name a few) start to signal in a ligand-independent manner when cells are treated with either H 2 O 2 or even stronger oxidants. This indicates that H 2 O 2 can mimic the ligand engagement process, and a list of possible alternatives to explain this activation phenomenon may include direct oxidation of the receptor that leads to aggregation, crosslinking, or conformational changes. Alternatively, H 2 O 2 could activate intracellular protein kinases involved in the signal transduction pathway associated with this receptor. A third possibility, not mutually exclusive, involves the inactivation of protein phosphatases by H 2 O 2 (for a review, see Ref. 57).
It is universally recognized that lipid rafts play a crucial role in the initiation and organization of signaling cascades, since they spatially concentrate or exclude components of the signaling machinery. Of note, another protein involved in the response to IR, ATM, has also been reported to localize outside the nucleus in membrane-associated vesicles. Furthermore, lymphoblast cells isolated from patients carrying a mutation in the ATM gene were defective in the IR-induced activation of a raft-resident protein-tyrosine kinase p53/56 Lyn (44).
To unravel the potential role of the DNA-PKcs in signal transduction, we searched for its localization in lipid rafts (shown in Figs. 1-3). Results from this study show that the holoenzyme DNA-PK, but not another member of the NHEJ machinery, XRCC4, is localized in these membrane-associated compartments. In addition, this localization in DIGs is not a general characteristic of proteins associated with DNA repair protein complexes, because representatives of other DNA repair pathways such as MSH2 (45), MSH6 (data not shown), or BLM-1 (data not shown) do not localize to lipid rafts. This in turn suggests that DNA-PK may be involved in the response to IR independent of its role in DSBR. These results warrant further experiments to determine the degree to which DNA-PK activity is dependent on DNA damage.
One of the most important properties of lipid rafts is that they can variably include or exclude proteins in response to a stimulus. However, as shown in Fig. 4 (A-E), relocation of any of the subunits of the DNA-PK was not dependent on irradiation.
Although DNA-PKcs is absent in yeast, functional homologues of Ku70 and Ku86 exist. Results from this study show the localization of Ku86 in lipid rafts isolated from yeast. Lipid rafts in yeast are similar to mammalian rafts but contain ergosterol instead of cholesterol (see Fig. 6). In an attempt to understand the physiological role that the presence of DNA-PK complex might have in the plasma membrane, we compared the protein phosphorylation pattern of lipid rafts isolated from a human cell line defective in DNA-PKcs, MO57J, and a control, MO57K, after irradiation. This was achieved by in vivo labeling with [ 32 P]orthophosphate and subsequent separation of lipid raft fractions using two-dimensional PAGE. Clear differences can be seen in Fig. 7A. These phosphorylation changes appear to be mediated directly or indirectly by DNA-PK activation. Efforts are under way to identify by mass spectroscopy putative targets.
At present, the mechanism of activation of DNA-PK in lipid rafts has not been determined, and a signal from DNA damage may be involved. In fact, this alternative activation pathway might be Ku-dependent as well. In response to IR, an allosteric modification in the DNA-PKcs subunit may lead to a direct activation of its kinase activity in the absence of DNA and/or to an alteration of its ability to interact with other proteins present in DIGs. Although DNA-PK is known to be activated by DNA ends, recent compelling evidence suggests that its kinase activity can also be stimulated by interacting proteins such as thyroid hormone receptor-binding protein or even certain Ab against DNA-PKcs (60) or the C1B protein (61). Layered on these simple outcomes are the roles of modified lipids after exposure to IR. The effect this treatment has on the regulation of DNA-PK activity is presently unknown. Experiments are in progress to dissect these possibilities.
In closing, an attractive possibility might be that an alternative DNA-independent activation pathway for the DNA-PK exists and that it is involved in a novel signal transduction pathway in response to IR exposure.
How Can This Complex Be Targeted to DIGs?-None of the subunits of the DNA-PK has any canonical signature of membrane proteins. However, the presence of Ku70 and Ku86 on the cell surface and their participation in cell-cell interaction have been reported (62)(63)(64)(65)(66). Furthermore, physical interaction between EGFR, a receptor activated after cells are exposed to IR, and DNA-PK has been described (32). Similarly, interaction between Ku70 with p40 phox in mammalian cells has also been identified (67,68).
As mentioned above, the assembly of the holoenzyme DNA-PK has been reported to occur in the absence of DNA. As shown in Fig. 4 (A-E), the lack of one component of the DNA-PK complex, either DNA-PKcs or Ku86, does not preclude the localization of other subunits into lipid rafts. This result suggests that there is no need for a complete preassembled complex to occur and favors the hypothesis of independent interaction directly or indirectly with membrane proteins. This in turn might bring these components to a common location, and further interaction might occur between them. In this context, immunoprecipitation of Ku86 from lipid raft fractions (fractions 3 and 4) brings down Ku70 but not DNA-PKcs, suggesting a physical interaction of at least the Ku complex in the absence of DNA (data not shown). However, it is still poorly understood whether the different subunits of the DNA-PK interact in DIG or whether they act independently of one another and only co-exist in lipid rafts.
Proteins incorporated into DIGs fall into different categories: glycophosphatidylinositol-anchored, trans-membrane, or cholesterol linked proteins (43). Unfortunately, none of these characteristics are present in any of the components of the DNA-PK complex that could explain its localization in lipid rafts. Curiously, examination of DNA-PKcs and Ku86 sequence using the Prosite software package revealed intriguing N-terminal myristoylation consensus, but in vivo data are lacking. However, results obtained from yeast Ku86 might suggest a dispensable role of the proximal N terminus.
It was recently shown that inositol hexakisphosphate binds to mammalian Ku and that this process is independent of its ability to bind to DNA (58,59). To date, a number of distinct, highly conserved phosphoinositol binding motifs have been identified (69). For example, the PX domain was initially identified as a conserved motif within p40 phox , and other subunits of the NADPH complex and a significant number of PX proteins are localized to membranes or vesicular structures within cells. Although none of the known domains bind inositol hexakisphosphate, there may be a link between the presence of an inositol phosphate intermediate and its membrane localization.
Another way in which DNA-PK subunits may be recruited into lipid rafts is by tethering directly or indirectly with plasma-resident proteins. In agreement with this, results shown in Fig. 5 demonstrate that the DNA-PK complex subunits are peripheral proteins associated with DIGs. For example, PDZ domains are protein-protein recognition modules that play a central role in organizing diverse cell signaling assemblies. They do this by localizing interacting protein partners that contain acceptor motifs, to specific regions on the cell membrane. Although PDZ domains recognize specific C-terminal sequence motifs, a growing number of examples also show internal peptide motifs as acceptors (70). Provocative evidence of such internal motifs is present in all of the subunits of the DNA-PK complex but in vivo evidence is awaiting further investigation.
A precedent for the second scenario may be represented by a protein named KIP/C1B, which has been identified to bind to the leucine zipper motif of the DNA-PKcs in a two-hybrid system search (61,71). In addition, DNA-PK can be activated by C1B in the absence of DSBs but in the presence of Ku. This in turn suggests an allosteric function of C1B that allows the binding of DNA-PKcs to Ku in the absence of DSBs (61). In addition, C1B is a calcium binding protein that is N-terminally myristoylated and has been identified to bind to two plasma membrane proteins: integrin ␣ IIb (72) and presenilin 2 (73). Interestingly, the C-terminal portion of either mouse or human C1B harbors a consensus signature of proteins that interact with PDZ domains. In addition, an x-ray-induced Ku70-binding protein associated with cell death, clusterin/XIP8/apolipoprotein J, is known to be present in plasma membrane and is involved in membrane lipid recycling, apoptosis, and other functions (74). Although the original report associated Ku70 with only the nuclear form of clusterin, the membrane-associated isoform may be present at low concentrations undetectable without an enrichment of plasma membrane proteins.
It is worth noting that a recent report provides evidence for the role of DNA-PK in innate immune response. In this study, in vitro activation of bone marrow-derived macrophages with bacterial DNA isolated from DNA-PKcs Ϫ/Ϫ mice was impaired in interleukin-6 and -12 production, and this pathway is dependent on activation of IKK and NF-B (75). The authors speculated about the possibility of a DNA-PK-dependent signaling cascade, but its internal (nuclear/cytoplasmic) localization does not support this hypothesis. Thus, the localization of DNA-PKcs in lipid rafts reported in this study might help to reconcile this conundrum. Additionally, NF-B has been demonstrated to be an in vitro substrate for DNA-PK and also to be involved in IR response (76). Consistent with their role in signal transduction, we 2 and others (77) localized these factors in DIGs.
Equally intriguing is the involvement of membrane signaling 2 H. Lucero, D. Gae, G. E. Taccioli, unpublished data.
in the bystander effect in irradiated cells and its impairment by destabilization of lipid raft after cholesterol sequestration (78). Experiments are under way to address the possibility that DNA-PK might be involved in this process.
In closing, we have provided compelling evidence that shows a novel localization of the DNA-PK complex in lipid rafts and evidence that this recruitment is not a generalized property of factors associated with the NHEJ machinery. This localization is probably an interaction with resident membrane proteins. Results from this study provide provocative evidence for the existence of an additional role of the DNA-PKcs, or perhaps the holoenzyme DNA-PK, to the one known in DSBR.
The implications of these findings in signal transduction warrant further investigation to unravel the role of DNA-PK in the response to IR at the membrane level and its dependence on DNA damage. These results may shed some light on understanding the mechanism of some DNA-PKcs-dependent plasma membrane phenomena, which do not reconcile with the known nuclear and cytoplasmic localization of this protein complex.