ANTISENSE RESEARCH AND DEVELOPMENT 2:17-25 (1992) Mary Ann Liebert, Inc., Publishers

Analysis of Oligonucleotide Binding, Internalization, and Intracellular Trafficking Utilizing a Novel Radiolabeled Crosslinker

DANIEL A. GESELOWITZ and LEONARD M. NECKERS

ABSTRACT

Although antisense oligonucleotides have been widely used to inhibit gene expression, their mechanism of entry into cells and their site of action are still in some doubt. In this report, we describe a novel technique for kinetically analyzing oligonucleotide association with living cells as well as intracellular compartmentalization. The technique utilizes a photoactivatable, radiolabeled crosslinker, the DennyJaffe reagent. Oligonucleotides containing pendant amine groups were conjugated to this reagent, added to HL60 cells in culture, and photocrosslinked to associated proteins, which were analyzed electrophoretically. We find that several proteins are labeled, predominantly a 75 kD one that appears to be membrane-associated. Our results suggest that the majority of intracellular oligonucleotide is associated in vesicles with the same protein to which it bound on the cell surface, but only a small percentage of non-protein-bound cytosolic oligonucleotide can be detected. Additionally, oligonucle¬ otides are readily accumulated by nuclei, and by treating whole nuclei, a unique set of nuclear binding proteins is detected. INTRODUCTION intracellular mRNAs are added exogenously to cells, be specifically inhibited. This is the basis of "antisense oligonucleotide" technology, but little is yet known about the movement of oligonucleotide from the extracellular milieu to intracellular sites of action. We previously showed that oligonucleotides accumulate intracellularly in a

When

oligonucleotides complementary to

mRNA translation

can

manner. Additionally, uptake of oligonucleotides by mamma¬ lian cells demonstrates characteristics associated with endocytosis, including a punctate intracellular distribution, sensitivity to sodium azide, and recycling to the cell surface (Loke et al., 1989). Endocytic internalization of oligonucleotides can be mediated by nonspecific association with plasma membrane proteins, as demonstrated for poly-L-lysine-conjugated oligonucleotides (Leonetti et al., 1990) or by association with a specific cell surface receptor structure or structures. The strongly anionic nature of unmodified phosphodiester oligonucleotides makes their nonspecific association with the largely anionic plasma membrane unlikely.

saturable, temperature- and energy-dependent

Clinical

Pharmacology Branch,

NCI, N1H, Bethesda,

Maryland. 17

GESELOWITZ AND NECKERS

In

previous report (Loke et al., 1989) we identified an 80 kD plasma membrane protein, isolated by oligonucleotide affinity chromatography, which we proposed as a candidate cell surface oligonucleotide receptor. Oligonucleotide binding proteins were isolated by passing radiolabeled cell lysate over an affinity column comprised of a homopolymer of oligo(dT) residues. Similar results are also obtained with oligo(dC) and oligo(dG) affinity columns (Loke and Neckers, 1989). Although yielding valuable information, this method suffers from the twin drawbacks that the binding characteristics of oligonucleotide homopolymers may not be representative of those of mixed-sequence oligonucleotides and that protein affinity isolation from a lysed cell preparation may not truly reflect the in situ situation. Even if mixed-sequence columns were prepared, the kinetics of association of oligonucleotide with cell membrane proteins, as well as intracellular trafficking of internalized oligonucleotide, cannot be visualized with this procedure. Yakubov et al. (1989) addressed some of these issues by incubating intact cells with a 5' radioactively end-labeled oligonucleotide that had been derivatized with a chemical crosslinking group capable of modifying nucleophilic centers in both nucleic acids and proteins. Although potentially well suited for receptor studies, this method relies on uncontrolled reactivity of the modified oligonucleotide with cellular constituents. In addition, detection of oligonucleotide-associated proteins depends on the 5'-radiolabeled phosphate group of the oligonucleotide; 5'-exonucleases, phosphatases, and transferases can rapidly remove the radiolabel from the oligonucleotidecrosslinker complex, resulting in decreased sensitivity, and perhaps specificity, of this method. We modified the approach of Yakubov et al. by coupling mixed-sequence oligonucleotides to a photoactivatable, radioiodinated crosslinking reagent, termed the Denny-Jaffe reagent (see Fig. 1). We chemically conjugated this reagent to oligonucleotides containing either a 5' or 3' pendant amine group (the amine displaces the W-hydroxysuccinimide group). Figure 1 schematically demonstrates the utility of this reagent. When exposed to ultraviolet (UV) light, the aryl azide group crosslinks to any cellular component in close proximity (within approximately 16 A). If desired, the complex can be chemically cleaved at the azo linkage and the iodinated portion of the crosslinker remains with the cellular component, not with the oligonucleotide. Since the photocrosslinking step can be performed quickly and efficiently at any desired time, one can obtain a "snapshot" of oligonucleotide distribution and protein binding status in intact cells. Furthermore, the use of the cleavable crosslinker allows the target molecules to be labeled and analyzed by gel electrophoresis without interference from the oligonucleotide. We report here studies using this reagent in which we confirm the presence of an approximately 75 kD cell surface oligonucleotide binding protein in several human cell lines. In addition, we observed that oligonucleotides can associate with several unique nuclear proteins and appear to do so within several hours of addition to intact cells. a

MATERIALS AND METHODS

Synthesis of oligonucleotides Oligodeoxynucleotides were prepared using a DNA synthesizer (Model 380B; Applied Biosystems, Foster City, CA). ß-Cyanoethylphosphoramidites and auxiliary reagents were commercially obtained. To add a pendant hexylamine phosphate group to the 5' end of the oligonucleotide, either Aminolink2 (Applied Biosystems) or 5'-Aminomodifier C6 (Glen Research) was used in the synthesis after the last DNA base. To incorporate a 3'-pendant amine group, a 3'-Amine On CPG column (Glen Research) was used in the synthesis. The oligonucleotides were purified by ethanol precipitation. The structures of the oligonucleotides used in this study are

NH2-(CH2)6-p-d[GATCATGCCCGGCAT] 1

d[GATCATGCCCGGCAT]pCH2CH(OH)CH2NH2 2 18

INTRACELLULAR TRAFFICKING OF OLIGONUCLEOTIDE The

Denny-Jaffe Reagent:

Principle of Crosslinking: DJ-reagent

Crosslinking (UV light)

SDS-PAGE

-Protein-Oligo Conjugate -Protein Alone

FIG. 1.

Denny-Jaffe reagent and crosslinking of a Denny-Jaffe-oIigonucleotide conjugate to an associated protein.

Conjugation of oligonucleotide

to

Denny-Jaffe reagent

All

manipulations of the Denny-Jaffe reagent were performed under safelight (Kodak OC filter). The was obtained as a benzene solution (NEN/Dupont). The pendant amine-modified oligonucleotide (typically 0.5-4 nmol) was dissolved in 25 µ of 0.1 M sodium carbonate buffer, pH 8.5. The benzene was evaporated from 100 µ of the Denny-Jaffe reagent using a stream of nitrogen, and the oligonucleotide solution was added and allowed to react for approximately I h. Glycine (250 µ ; 0.2 M, adjusted to pH 8.5) was added to quench unreacted reagent. The reaction solution was loaded on a gel permeation column (PD10, Pharmacia) and the column was eluted with phosphate-buffered saline (PBS). The eluate was collected for about 1.0 ml after the first detection of radioactivity and stored in a light-tight container. Specific activity of crosslinked oligonucleotide was approximately 8 Ci/mmol. Polyacrylamide gel electrophoresis (PAGE) confirmed that the radioactivity in the sample was due to Denny-Jaffe reagent conjugated to oligonucleotide. reagent

Cell culture HL60 (human promyelocytic leukemia), U937, or CEM cells were grown in RPMI 1640 medium supplemented with glutamine, penicillin, streptomycin, HEPES buffer (15 mM), and 10% heat-inactivated fetal bovine serum (FBS). Cells were incubated at 37°C in a 95% air and 5% C02 atmosphere. 19

GESELOWITZ AND NECKERS

h-J 1. Resuspend HL60 pellet in RPMI/1%

&

-

3. Irradiate with long-wave UV

2. Add Denny-Jaffe ollgo complex.

BSA

Incubate 2-3 h.

5.

6. Resuspend In NP40 solution. Centrifuge and wash.

Split 4. Pellet and wash

with PBS.

Nuclei. Lyse and by SDS-PAGE

analyze

7.

Homogenize in

sucrose

buffer.

10. Ultraconcentrate

Ü --u supernatant.

Pellet is Nuclei and

unhomogenjzed

cells. Phenol-extract for DNA

Cytosol fraction Analyze by SDS-PAGE

analysis. Pellet is membrane fraction.

Analyze by

SDS-PAGE

FIG. 2.

Procedures used to study uptake subsequent cellular subfractionation.

and

photocrosslinking

Uptake of Denny-Jaffe-conjugated oligonucleotide by

of

Denny-Jaffe-oligonucleotide conjugate

and

HL60 cells

90 106 cells were washed several times with a solution of 1% bovine serum albumin (BSA) in PBS and were then resuspended in 1.0 ml of this solution in 1 well of a 12-well plate. Under safelight, 0.2 ml (approximately 8 µ ) of the Denny-Jaffe-crosslinked oligonucleotide was added to the cell suspension and the plate was incubated in the dark at 4 or 37°C. The well was then exposed to long-wave UV light (366 nm) from a hand-held U V lamp (Model UVGL25 ; U VP, Ine., San Gabriel, C A) mounted a few cm above the well for 10 min. The cell suspension was then transferred to a 15 ml centrifuge tube and washed several times with cold PBS.

Approximately

Cellular

subfractionation

Figure 2 shows schematically the procedure for oligonucleotide uptake and fractionation of cells. Cell pellets were either lysed directly in lysis buffer (300 mM NaCl, 50 mM Tris, and 0.5% NP-40, adjusted to pH 7.5) or processed for isolation of membrane and soluble cytosolic proteins. To isolate the membrane and cytosol fractions, the cell pellet was resuspended in 5.0 ml ice-cold 0.25 M sucrose, 0.15 M NaCl, 5 mM Na2-EDTA, and 1 mM phenylmethylsulfonyl fluoride (PMSF) and homogenized with 50 strokes in a Dounce homogenizer. The homogenate was centrifuged at 1000 x g for 10 min at 4°C, and the pellet (containing intact cell and nuclei) was discarded. The supernatant was removed and centrifuged at 105,000 x g for 60 min at 4°C. The supernatant from the ultracentrifugation contains soluble cytosolic protein; the pellet contains membrane-associated, nonnuclear proteins. The cytosol fraction was concentrated by spin ultrafiltration 20

INTRACELLULAR TRAFFICKING OF OLIGONUCLEOTIDE

75

64 52

40

34 28

FIG. 3. Autoradiograph of two-dimensional gel analysis of proteins from HL60 cells treated with Denny-Jaffeoligonucleotide 1 conjugate for 2 h at 37°C before photolysis. The lysate was electrophoresed in an 8% polyacrylamideSDS tube gel (horizontal dimension), treated with hydrosulfite to cleave the oligonucleotide from crosslinked protein, and electrophoresed again the the vertical dimension. Molecular weights of major bands (kD) are indicated.

(Centricon 10, Amicon) and mixed with an equal volume of 2X lysis buffer. The membrane pellet was lysed in 100 µ lysis buffer. The protease inhibitors PMSF, aprotinin, and leupeptin were added to the lysis buffer immediately before use. Lysates were analyzed on 8% gels by polyacrylamide-SDS (sodium dodecyl sulfate)

electrophoresis. In some experiments, lysates were electrophoresed in 8% tube gels; the tube gels were soaked three times for 15 min in 0.1 M sodium hydrosulfite to cleave the crosslinked oligonucleotide-protein complex, reequilibrated with buffer, and then were carefully placed horizontally on a second 8% gel and electrophoresed a second time. Gels were dried and autoradiographed with XOMAT XAR-5 film for several days to weeks. Incubation

of intact nuclei

with

Denny-Jaffe-conjugated oligonucleotides

In some experiments, intact HL60 nuclei were isolated and treated with the Denny-Jaffe-oligonucleotide conjugate. Cell pellets were resuspended in 4 ml cold PBS, made 0.1% in NP-40, and incubated for 10 min at 4°C. The presence of naked but intact nuclei was confirmed both microscopically and by electronic volume determination using a Coulter Counter (Model ZBI). Following a final wash in PBS, nuclei were suspended at 90 X 106 per ml in PBS containing 1% BSA and were incubated with Denny-Jaffe-crosslinked oligonucleotide as described for intact cells. Following incubation and UV exposure, nuclei were washed twice in cold PBS, pelleted, and lysed in lysis buffer. Lysates were analyzed as described.

RESULTS Association

of Denny-Jaffe-conjugated oligonucleotide

with intact cells

Denny-Jaffe-conjugated oligonucleotide 1 was incubated with HL60 cells at 37°C for 2 h before photocrosslinking, the whole-cell lysate was found to have several prominently labeled proteins. When lysates were analyzed following crosslinker cleavage, a prominent 75 kD protein and several smaller proteins can be seen (Fig. 3). If the lysate is run without the chemical cleavage step, the prominent band runs at Mr approximately 79 kD and is sometimes seen as two or three bands. We attribute this altered migration to different attachment sites of the aryl azide, or possibly to multiple binding. Nearly identical results are When the

21

GESELOWITZ AND NECKERS obtained when oligonucleotide 2 is used (data not shown). Moreover, very similar bands are seen with the monocytic leukemia cell line U937 or the cell acute lymphoblastic leukemia cell lineCEM (data not shown). We previously determined that at 4°C oligonucleotide binding to cells occurs, but intracellular accumula¬ tion of oligonucleotides is nearly completely abrogated (Loke and Neckers, 1989; Loke et al., 1989). We observed that incubation of cells with Denny-Jaffe-derivatized oligonucleotide at 4°C for 2 h before photocrosslinking results in the labeling of a 75 kD band, but the overall amount of labeling is greatly reduced compared to results obtained at 37°C (data not shown). These data suggest that p75 is a cell surface protein that continues to associate with oligonucleotide upon internalization. By lysing oligonucleotide-exposed intact cells and extracting nucleic acids with phenol-chloroform, we were able to analyze intracellular non-protein-bound oligonucleotide. Based on radioactivity measurements, the amount of free intracellular oligonucleotide observed is less than 10% of the total cell-associated material. When run on a denaturing polyacrylamide gel, the recovered free oligonucleotide appears little different from the starting material, suggesting that intracellular oligonucleotide remains largely intact after 2 h (data not

shown). In other experiments, we examined a variety of Denny-Jaffe-conjugated phosphodiester oligonucleotides of different lengths and sequences, including one with fluorescein on the 5' end. Although in certain cases some unique proteins are found to be weakly labeled, p75 is uniformly the predominant band detected.

Association

of Denny-Jaffe-conjugated oligonucleotide

with subcellular fractions

unlikely that the non-p75 bands detected in Fig. 3 represent intracellularly processed proteolytic fragments of p75 because, since the crosslink had been broken before the gel was electrophoresed in the second dimension, all radioactively labeled protelytic fragments would have had to contain the oligonucle¬ otide binding site. Instead, we reasoned that these novel bands may represent distinct intracellular oligonucleotide binding proteins that become associated with oligonucleotides only following oligonucleotide internalization. To test this hypothesis, we subfractionated cells that had been incubated with oligonucleotides for 2 h at 37°C and determined the presence of oligonucleotide binding proteins in nonnuclear membrane (including endosomal vesicles) and soluble cytosolic fractions (Fig. 4). When the treated HL60 cells were subfractionated into membrane and cytosol, the membrane fraction was found to contain the 75 kD band (reproducibly running at 79 kD uncleaved), as well as a second band of Mr 28 kD. The soluble cytosolic fraction contained only a faint band at 64 kD. The p64, p52, p40, and p34 bands observed in Fig. 3 were not It is

detected in either fraction.

Association

of Denny-Jaffe-crosslinked oligonucleotide with

intact nuclei

The unidentified bands that appear upon prolonged incubation of intact cells with oligonucleotide at 37°C clearly do not represent soluble cytosolic oligonucleotide binding proteins (compare Figs. 3 and 4). To determine if these labeled bands represent nuclear oligonucleotide binding proteins, we isolated intact nuclei and incubated them with Denny-Jaffe-conjugated oligonucleotide 1. Several bands were strongly labeled by this procedure, including bands of 63, 48, 38, 32, 29, and 20 kD (Fig. 5). A 75-79 kD band was clearly not present. Upon nuclear lysis followed by phenol-chloroform extraction to remove protein, fully 70% of the radioactivity appeared to be protein associated.

DISCUSSION describe a novel method of studying cell membrane and intracellular oligonucleotide The method requires only that an oligonucleotide contain a pendant amine group and thus binding proteins. lends itself to analysis of protein associations and intracellular trafficking of various modified oligonucleIn this report

we

22

INTRACELLULAR TRAFFICKING OF OLIGONUCLEOTIDE

M

C

79

28 Comparison of membrane and cytosol fractions from HL60 cells treated with Denny-Jaffe-oligonucleotide 1 conjugate for 2 h at 37°C before photolysis. Samples were electrophoresed in an 8% polyacrylamide-SDS gel. Molecular weights of major band (kD) are indicated. FIG. 4.

otides, including phosphorothioates, methylphosphonates, anomers, and even chimeric oligonucleotides. By cellular subfractionation, one can localize oligonucleotide association with intracellular proteins. We did not determine the optimum photocrosslinking time, but there are reports that a short flash may be sufficient (Ji, 1979); thus one could conceivably follow the kinetics of intracellular trafficking by photocrosslinking at different times. We observed that phosphodiester oligonucleotides of various lengths and sequences bind to a common 75 kD cell membrane protein present in at least three different cell lines. This protein appears to be identical to the oligonucleotide binding protein previously identified by our laboratory and the 79 kD protein identified by Yakubov et al., (1989). We also identified a second membrane-associated oligonucleotide binding protein, p28. These findings correlate very well with the southwestern blotting identification of DNA binding cell surface proteins of 30 kD by Bennett and coworkers (Bennett et al., 1985) and of 79 and 28 kD by Gasparro et al. (1990). Additionally. in examining subcellular fractions we detected a soluble cytosolic oligonucleotide binding protein (p64), as well as several unique nuclear proteins capable of associating with oligonucleotides. A recent report has identified a similar set of nuclear oligonucleotide binding proteins (Leonetti et al., 1991 ). Our data suggest that the majority of cell-associated oligonucleotide is bound to p75, but after a 2 h incubation, this material is primarily associated with internalized vesicles rather than the cell surface. These findings suggest that oligonucleotides remain associated with the 75 kD protein upon endocytosis. Approximately 90% of internalized oligonucleotide is protein-associated. It is not clear if the "free" 10% is truly free or simply represents failure of the aryl azide group to crosslink. Consistent with two recent studies (Leonetti et al., 1991; Chin et al., 1990), however, it seems likely that, during and following lysosomal fusion, some oligonucleotide is freed from protein association and promptly migrates to the nucleus, whereupon it can reassociate with a series of unique nuclear proteins. We estimate that fully 70% of nuclear

23

GESELOWITZ AND NECKERS

FIG. 5. Proteins labeled by treatment of intact HL60 nuclei with Denny-Jaffe-oligonucleotide 1 conjugate for 2 h at 37°C before photolysis. Lysates were electrophoresed in an 8% polyacrylamide-SDS gel. Molecular weights of major bands (kD) are indicated.

oligonucleotide is protein-bound. As hypothesized by Leonetti et al., (1991), if this protein association were weak, oligonucleotides might be free to reassociate with target RNA or DNA, but if the association with nuclear protein is too strong it would certainly hinder antisense or triplex effectiveness. Perhaps pulse-chase studies on intact nuclei will determine the stability of oligonucleotide-protein interactions. Our results appear to corroborate a number of reports that exogenous oligonucleotide can enter the cytoplasm via binding to cell surface receptors. It is interesting to speculate on the biologic purpose of these receptors. Gasparro and coworkers (1990) suggest that these DNA binding proteins could be the proteins that naturally maintain cell membrane DNA (cmDNA). cmDNA appears to be distinct from chromosomal and mitochondrial DNA (Lerneret al., 1971) and may have an immunologie role (Aggarwal et al., 1975). If so, the natural exchange of cmDNA between the interior and surface of the cell may be the mechanism that allows antisense oligonucleotides to enter the cell. A substantial literature now suggests that once in the cytoplasm, oligonucleotides are transported to the nucleus. If oligonucleotide tends to reside in the nucleus rather than in the cytoplasm, pre-mRNA targets, such as splice donor-acceptor sites, may prove more amenable to antisense inhibition than the more commonly utilized translation initiation site region of mature mRNA. Furthermore, triplex inhibition of gene transcrip¬ tion appears ideally suited to take advantage of the tendency of endosomally escaped oligonucleotides to concentrate

in the nucleus.

REFERENCES AGGARWAL, S.K., WAGNER, R.W., MCALLISTER, P.K., and ROSENBERG, B. (1975). Cell-surface-associated nucleic acid in tumorigenic cells made visible with Nati. Acad. Sci. U S A 72, 928-932.

platinum-pyrimidine complexes by electron microscopy. 24

Proc.

INTRACELLULAR TRAFFICKING OF OLIGONUCLEOTIDE

BENNETT, R.M., GABOR, G.T., and MERRITT, M. (1985). DNA binding to human leukocytes. J. Clin. Invest. 76,

2182-2190. CHIN, D.J., GREEN, G.A., ZON, G., SZOKA, F.C.J., and STRAUBINGER, R.M. (1990). Rapid nuclear accumulation of injected

oligodeoxyribonucleotides.

New Biol.

2, 1091-1100.

GASPARRO, F.P., DALL'AMICO, R., O'MALLEY, M., HEALD, P.W., and EDELSON, R.L. (1990). Cell membrane DNA: A new target forpsoralen photoadduct formation. Photochem. Photobiol. 52, 315-321. JI, T.H. (1979). The application of chemical crosslinking for studies on cell membranes and the identification of surface reports. Biochim. Biophys. Acta 559, 39-69. LEONETTI, J.-P., DEGOLS, G., and LEBLEU, B. (1990). Biological activity of oligonucleotide-poly (L-lysine)

conjugates: Mechanism of cell uptake. Bioconj. Chem. 1, 149-153. LEONETTI, J.P., MECHTI, N., DEGOLS, G., GAGNOR, C, and LEBLEU, B. (1991). Intracellular distribution of microinjected antisense oligonucleotides. Proc. Nati. Acad. Sci. U S A 88, 2702-2706. LERNER, R. ., MEINKE,W., and GOLDSTEIN, D.A. (1971). Membrane associated DNA in the cytoplasm of diploid human

lymphocytes.

Proc. Nati. Acad. Sci. U S A

68,

1212-1216.

LOKE, S.L., and NECKERS, L. (1989). Unpublished observations. LOKE, S.L., STEIN, CA., ZHANG, X.H., MORI, K., NAKANISHI, M., SUBASINGHE, C, COHEN, J.S., and NECKERS, L.M. (1989). Characterization of oligonucleotide transport into living cells. Proc. Nati. Acad. Sci. U S A

86, 3474-3478.

YAKUBOV, L.A., DEEVA, E.A., ZARYTOVA, V.F., IVANOVA, E.M., RYTE, A.S., YURCHENKO, V.L., and VLASSOV, V.V. (1989). Mechanism of oligonucleotide uptake by cells: Involvement of specific receptors? Proc. Nati. Acad. Sci. U S A

86, 6454-6458.

Address reprint requests to: Leonard M. Neckers Clinical Pharmacology Branch NCI, N1H

Building 10, 12N236 Bethesda, MD 20892 Received

September 24, 1991;

in revised form October 16, 1991;

25

accepted

for

publication

October 21, 1991.