- Email: [email protected]
Immunoliposome labeling: A sensitive and specific method for cell surface labeling
Journal of Immunological Methods, 46 ( 1981 ) 141--151 Elsevier/North-Holland Biomedical Press
141
IMMUNOLIPOSOME LABELING: A SENSITIVE A N D SPECIFIC METHOD FOR CELL S U R F A C E LABELING
ANTHONY
HUANG
I, S T E P H E N
J. K E N N E L
I and L E A F H U A N G
2
I Biology Division, Oak Ridge National Laboratory, and 2 Department of Biochemistry,
University of Tennessee, Knoxville, TN 37916, U.S.A. (Received 20 March 1981, accepted 10 June 1981)
A simple, one-step procedure for fluorescence labeling of cultured cells with high sensitivity and specificity is described. We term this method immunoliposome labeling. Monoclonal IgG antibody was first covalently coupled with palmitic acid. The palmitoylIgG was then incorporated into unilamellar liposomes (about 100 nm diameter) containing either N-(4-nitrobenzo-2-oxa-l,3-diazole)- or N-(fluorescein isothiocyanyl)-phosphatidylethanolamine by a detergent-dialysis procedure. A monoclonal antibody to the mouse major histocompatibility antigen, H-2k, was tested as a model system. Fluorescent liposomes with covalently coupled anti-H-2k specifically labeled the mouse L-929 cells (H-2 k type), but not the A-31 cells (H-2 d type). The degree of labeling was quantitated by a microscope photometer. Cells labeled with fluorescent liposomes showed 4--6-fold stronger fluorescence than cells labeled with either fluorescein-conjugated antibody, or with primary antibody followed by fluorescein-conjugated secondary antibody (indirect immunofluorescence). Since different types of label (fluorescent, radioactive, etc.) can be incorporated into liposomes, this specific and sensitive method is potentially very versatile.
INTRODUCTION
Specific fluorescence labeling of cell surface antigens or receptors has become useful in several aspects of cell biology. Standard methods employ primary or secondary antibody or lectins which have been conjugated with proper fluorophores, such as fluorescein or rhodamine. Conventionally, indirect immunofluorescence is required to enhance the sensitivity of the method. However, these amplification steps can be time-consuming and can also increase background fluorescence significantly. They are therefore often unsuitable for many specific applications, such as the fluorescence activated Send correspondence to: Dr. Leaf Huang, Department of Biochemistry, University of Tennessee, Knoxville, TN 37916, U.S.A. Abbreviations: FITC, fluorescein isothiocyanate; IgG, immunoglobulin G; NBD, 4-nitrobenzo-2~axa-l,3-diazole; PBS, phosphate-buffered saline (137 mM NaC1, 2.7 mM KC1, 1.5 mM KH2PO4, 1 mM Na2HPO4, pH 7.8.); PC, phosphatidylcholine; PE, phosphatidylethanolamine. 0022-1759/81/0000--0000/$02.50 © 1981 Elsevier/North-Holland Biomedical Press
142
cell sorting which requires a rapid, sensitive, and specific labeling of living cells. Based on a previous finding in our laboratories, we have developed a one-step, sensitive technique which specifically labels the cell surface antigenic sites. The method involves the covalent coupling of antibody to liposomes and the subsequent binding of liposomes to the cells. We have named this method immunoliposome labeling. This method is quite different from that described by Leserman et al. (1979) which used liposomes containing haptens or antigens to label only cells bearing specific antibodies. Our method is therefore considerably more versatile. A preliminary report of this work has been presented elsewhere (Huang et al., 1980b). MA TER I ALS AND METHODS
Materials Anti-H-2 k IgG was purified from the culture medium of a hybridoma 114.1 cell line (Oi et al., 1978) by protein A affinity, chromatography (Herrmann and Mescher, 1979). Normal mouse IgG was similarly purified from normal mouse serum or ascites fluid induced with complete Freund's adjuvant. Goat antibody to mouse IgG was enriched from goat antiserum using immunoabsorbent technique (Kennel and Feldman, 1976). N-hydroxysuccinimide ester of palmitic acid was synthesized as described (Lapidot et al., 1967). Egg yolk phophatidylcholine (PC) and phosphatidylethanolamine (PE) were purified as described (Huang and Pagano, 1975). N-(4-nitrobenzo2-oxa-l,3
Coupling of IgG with palmitic acid The previously described method (Huang et al., 1980a) was followed with minor modifications. The molar ratio of N-hydroxysuccinimide ester of palmitic acid to IgG in the reaction mixture was 10:1 which gave a coupling efficiency of about 3--4 palmitoyl chains per IgG without significant decrease in antigen binding capacity (A. Huang, unpublished observation). The excess free palmitic acid was removed by Sephadex G-75 column chromatography as described (Huang et al., 1980a). Both anti-H-2 k and normal mouse IgG were treated identically.
Preparation of fluorescent liposomes A detergent
143 was used with minor modifications. Briefly, palmitoyl IgG and phospholipids (weight ratio 1:20) were mixed in phosphate-buffered saline (PB8) containing 0.7% deoxycholate. The phospholipids contained approximately 95 mole % egg yolk PC and 5 mole % NBD-PE or FITC-PE. The mixture was dialyzed for 40 h at r o o m temperature against PBS using Spectro-Pore 2 dialysis tubing (Spectrum). The turbid, yellowish dialysate was chromatographed on a Sepharose 4B column to remove unincorporated IgG. Liposomes eluted in the void volume were diluted with McCoy's medium and used for the cell labeling experiments.
Electron microscopy Liposomes containing antibody and fluorescent phospholipids were negatively stained with phosphotungstate as described (Chang and Huang, 1979). Samples were viewed in a Hitachi 600 electron microscope.
Preparation of FITC-labeled antibodies Labeling of anti-H-2 k or goat anti-mouse IgG with FITC was carried o u t as follows. Protein (4 mg in 0.4 ml PBS) was dialyzed against 60 ml carbonate buffer (0.025 M Na carbonate, pH 9.4) containing 6 mg FITC for 27 h at 4°C. The labeled antibody was then dialyzed extensively against PBS. The extent of labeling was estimated b y measuring absorbance at 495 nm and calculated according to the following equation (Fothergill, 1969): mole FITC/mole IgG = 2.8 X A49s/[IgG] in mg/ml. The degree of labeling was a b o u t 2 FITC/IgG for anti-H-2 k and 3 FITC/IgG for goat anti-mouse IgG.
Preparation of NBD-labeled antibodies Two mg anti-H-2 k or goat anti-mouse IgG (9.5 mg/ml in PBS) and 15 #1 N B D ~ l (4 mg/ml in dimethylsulfoxide) were added to 1.27 ml 0.025 M Na carbonate buffer, pH 9.4. The molar ratio of protein to N B D ~ I was approximately 1:24. Incubation was for 4 h in the dark at r o o m temperature. The mixture was applied to a Sephadex G-75 column equilibrated and eluted with PBS. Protein eluted in the void volume s h o w e d an absorption peak at 480 nm. The degree o f labeling, estimated b y using an extinction coefficient of 1.62 X 104/mol/cm (Cantley and Hammes, 1975), was a b o u t 3 NBD/IgG for anti-H-2 k and 4 NBD/IgG for goat anti-mouse IgG.
Fluorescence labeling of cells Mouse L-929 and A-31 cells were grown on glass coverslips (22 mm X 11 mm) in McCoy's medium with 10% d o n o r calf serum {Flow Laboratory). Coverslips were transferred to a 6-well 35 m m Linbro plate (2 coverslips per well) after washing once in PBS. Fluorescent liposome or antibody at 5 ~tg protein/ml was added and culture plates were gently rocked at r o o m temperature for 1 h. Coverslips were then washed 5 times in PBS and stored in PBS until observed under a microscope. For indirect immunofluorescence labeling, the incubation with unlabeled primary antibody (anti-H-2 k, 5 ~g/
144 ml) was for 1 h at r o o m temperature. After washing, the fluorescent secondary antibody (goat anti-mouse IgG) was added at 50 ~g/ml in McCoy's medium. After 1 h at r o o m temperature, coverslips were washed 5 times in PBS and observed under a microscope.
Fluorescence microscopy and quantitation Coverslips were placed with cell-side d o w n on glass slides. A Leitz fluorescence microscope equipped with epiluminescence was used. FITC and NBD fluorescences were observed with appropriate filters. Photomicrographs were taken by an automated camera. The quantitation of fluorescence was done immediately after the micrographs were taken and was measured on a new field of the same coverslip to minimize bleaching. A microscope p h o t o m e t e r (Leitz MPV-2) with an adjustable orifice was used. A small circle (3 pm diameter) which covered only a portion of the cell periphery was measured for relative fluorescence intensity. Only one cell in a given microscope field was measured, because prolonged illumination of a .given field produced significant bleaching. This is especially true for FITC-labeled liposomes with which the fluorescence bleached with a half time of a b o u t 15 sec. A measurement by the p h o t o m e t e r usually t o o k less than 1 sec, so that there was no significant bleaching during the measurement. All p h o t o m e t e r readings were expressed in arbitrary units after subtracting a background reading measured at an area with no cells. Other analytical methods Lipid phosphorus content was measured by the m e t h o d of Ames and Dubin (1960). Absorption was measured with a Cary 15 spectrophotometer and fluorescence with a Aminco-Bowman fluorospectrophotometer. RESULTS
Preparation of antibody-coated, fluorescence-labeled liposomes by the deoxycholate
145
Fig. 1. Negatively stained liposomes containing covalently coupled anti-H-2 k and NBDPE. Bar is 1 gin.
Liposome labeling of cells was compared with conventional immunofluorescence labeling techniques. When FITC-labeled anti-H-2 k was incubated with L-929 cells at a concentration identical to those in the liposomes (5 pg/ ml), only very weak fluorescence was observed at the cell periphery. With an indirect immunofluorescence technique, i.e., incubation with anti-H-2 k followed b y FITC-labeled goat anti-mouse IgG, stronger fluorescence was seen (Fig. 2a). The distribution of fluorescence at the cell surface was somewhat punctate in this case. This was in contrast to the liposome labeling which showed a more diffused distribution (Fig. 2c). Labeling of cells by either direct or indirect antibody m e t h o d was also specific for L-929 cells, since A-31 cells were not labeled under identical conditions (Fig. 3a). FITClabeled goat anti-mouse IgG alone did n o t label either cell t y p e (micrograph not shown). When NBD-labeled primary and secondary antibodies were tested for binding to cells, no significant fluorescence on either cell t y p e was
146
! ~ -
¸ ¸~i~
~
¸
~
~
"
i
Fig. 2. Phase-contrast (left panels) and fluorescence (right panels) photomicrographs of mouse L-929 cells labeled with: (a) anti-H-2 k followed by FITC-labeled goat anti-mouse IgG, photographic exposure time 2 rain 28 sec; (b) anti-H-2 k liposomes containing FITCPE, exposure time 3 rain 30 sec; (c) anti-H-2 k liposomes containing NBD-PE, exposure time 2 rain 2 sec; and (d) mouse IgG-liposomes containing NBD-PE, exposure time 2 rain. Bar is 10/zm.
147
$
"
q
Fig. 3. Phase-contrast (left panels) and fluorescence (right panels) p h o t o m i c r o g r a p h s of m o u s e A-31 cells. T r e a t m e n t s were the same as in Fig. 2, e x c e p t the p h o t o g r a p h i c exposure times were: (a) 5 rain; (b) 4 rain; (c) 5 rain; and (d) 2 rain. Bar is 10/~m.
148
observed. The reason for this failure of cell labeling b y NBD-antibody was unclear. We have estimated that each IgG molecule contained a b o u t 3--4 NBD molecules on the average. Labeling of antibody with NBD could have resulted in some denaturation o f the protein and the subsequent loss o f the antigen-binding activity. It should be emphasized that the fluorescence intensities of cells shown in Figs. 2 and 3 are somewhat misleading for t w o reasons. Firstly, these photographs were not taken with the same exposure time (ranging from 2 to 5 min). Secondly, bleaching o f the fluorophore during the exposure period significantly reduced the apparent fluorescence intensity. The latter problem was particularly serious for the FITC-PE label which was usually bleached within 30 sec (Fig. 2b). The degree o f fluorescence labeling o f cells was therefore quantitated by a microscope p h o t o m e t e r m e t h o d . The relative fluorescence intensities for b o t h L-929 and A-31 cells labeled b y various types of reagents are shown in Table 1. Although all cells were labeled, the degree of labeling by liposome or antibody varied substantially from cell to cell, or from one area o f a cell to another. It was therefore necessary to make a large number of measurements and analyze the data statistically. For the FITC label, liposome labeling gave a cell surface fluorescence which was a b o u t 5--6-fold higher than the direct immunofluorescence labeling and 4--5-fold higher than the indirect immunofluorescence labeling. The labeling was specific for the antigen-bearing cells, since only background fluorescence was obtained with A-31 cells. Similar results were obtained with the NBD label, except the NBD-labeled direct or indirect immunofluorescence was negative. NBD-labeled liposomes gave a b o u t the same degree of labeling intensity as the FITC-labeled liposomes under identical conditions, except the NBD label was much more resistant to
TABLE 1 Relative fluorescence intensitites o f labeled cells. Fluorophore
Cell
FITC
L-929 A-31
NBD
L-929 A-31
Anti.H.2 k a
3.2 -+ 1.9 (35) b --0.5 _+ 1.9 (22) 0.6 -+ 1.4 c (20) 0.3 -+ 1.4 (20)
Anti-H-2 k + goat × m o u s e IgG
Anti-H-2 kliposome
G o a t x m o u s e IgG
4.2 -+ 2.4 (46) 0.4 _+ 1.8 (30)
18 -+ 7.7 (26) --0.7 _+ 1.7 (26)
0.6 -+ 2.0 (34) 0.3 + 0.9 (2O)
0.8 + 1.3 c (20) ---0.2 _+ 1.8 (20)
18 -+ 9.6 (47) --0.2 _+ 0.9 (20)
0 . 2 -+ 3.7 (22) --0.9 _+ 1.0 (12)
a L o c a t i o n o f the f l u o r o p h o r e is indicated by parentheses. b N u m b e r of measurements. c L o w degree o f labeling p r o b a b l y due to N B D - a n t i b o d y d e n a t u r a t i o n ; see text.
149 photo-bleaching than FITC (bleaching half-time for NBD-PE liposomes was 1.5 min with the same illumination intensity). DISCUSSION We have previously shown that liposomes with covalently coupled antibody can bind specifically to cell surfaces of the antigen-bearing cells (Huang et al., 1980a). On the basis of this finding, we have designed an immunoliposome technique which can specifically label cell surface antigens with a fluorescence probe. Since liposomes are much larger than IgG, the amount of label that each liposome can carry is theoretically much greater than can be carried by the IgG molecule. We thus expect an improved labeling sensitivity over the conventional immunofluorescence technique. Using monoclonal antibody to the H-2 k antigen as a model system, we have shown a 4--6-fold stronger fluorescence with liposomes than with direct or indirect immunofluorescence. We have used 5 pg/ml anti-H-2 k, either free or liposome-bound, which produces a saturation binding on L-929 cells. Under these conditions, we expect to bind 1.1 X 104 liposomes per cell (A. Huang, unpublished data), assuming that there are approximately 8.6 X 104 phospholipid molecules per 100 nm liposome (Enoch and Strittmatter, 1979). This level of binding would bring about 4.8 X 107 fluorescent lipid molecules/cell at the cell surface. On the other hand, about 1.2 X 106 free anti-H-2 k molecules/cell can be bound to the L-929 cell surface at 5 pg/ml antibody concentration (P. Naylor, personal communication). This amounts to about 2.4--3.6 X 106 fluorescence molecules/cell, since there are 2--3 fluorescence molecules per IgG. This simple estimation would predict a 1 3 20-fold higher labeling by the liposome technique. In reality, we have only observed 4--5-fold increase over the direct antibody labeling. We can list at least 3 explanations for the apparent discrepancy. Firstly, the fluorescence quantum yield of the fluorophore on the liposome might be lower than that on the protein (Chang and Huang, 1979). Secondly, the amount of fluorescent phospholipid per liposome (about 5 mole %) might be high enough to generate self~luenching. Finally, the microscope photometer might give erroneously high readings for the weak fluorescence from the immunofluorescence labeling. This last possibility is quite plausible since we have noticed that the Leitz MPV-2 system consistently overestimates weak fluorescence. Despite these problems, the immunoliposome labeling has demonstrated a higher sensitivity than the conventional immunofluorescence techniques. The use of monoclonal antibody for the immunoliposome labeling is theoretically not essential. However, the use of conventional antibody preparations would result in lower amounts of fluorescence. This is because the number of IgG molecules per liposome (about 8--10) in the present study is not very large. Since most immune IgG preparations contain only 1--2% high affinity antibody, many liposomes would contain no antibody.
150
We are currently improving the IgG-liposome coupling method to obtain a higher amount of IgG per liposome. Until such method is available, the use of monoclonal antibody or some enriched immune IgG prepared by immunoabsorbent technique is still desirable. Cell surface labeling requires not only high sensitivity, but also high specificity. The immunoliposome labeling procedure is very specific for the antigen-bearing cells. We have previously shown that the binding of liposome to cell surface can be specifically blocked by pretreatment of cells with the unlabeled, free antibody (Huang et al., 1980a). Thus the binding is indeed mediated by specific antigen-antibody interactions. We have purposely chosen two fluorescent phospholipids with the fluorophores located at the head group position of the PE. It has been shown that the N-modified fluorescent PE does not undergo exchange reactions with the cellular phospholipids, a process that might contribute significantly to the background nonspecific labeling of the cell surface (Struck and Pagano, 1980). We have also chosen to use liposomes of about 100 nm in diameter, instead of the small unilamellar type (about 25 nm diameter} which are more prone to fuse with the cultured cells (Pagano and Huang, 1975). Non-specific fusion of liposome with cells would also increase the background labeling. In experiments shown here, fusion of bound liposomes with cells apparently did not occur, since significant amounts of the bound liposomes could be dissociated by a brief protease treatment (unpublished observation). Thus, the specificity of this cell-labeling method is also favorably compared to the conventional immunofluorescence methods. Since the phospholipids do not exchange between liposomes in the absence of exchange proteins (Hellings et al., 1974), the present method can be easily adapted for double fluorescence labeling. Such labeling would only require a one-step procedure using a mixture of two types of immunoliposome, each containing a different fluorescent phospholipid (e.g. rhodamine-PE and FITC-PE). Theoretically, the current method is not limited to the fluorescence labeling. Radioactive lipids or water-soluble molecules can be incorporated into the lipid or aqueous phase of liposomes, respectively. With a proper choice of the labeling molecule to avoid non-specific labeling or leakage, autoradiography of cells should also be possible. Certain limitations of the liposome labeling technique are worth mentioning. Since the liposomes used here are about 100 nm in diameter, any attempt to visualize the cell-bound liposomes at the electron microscopic level will be limited to a spatial resolution of, at best, about 100 nm, which is greater than the established techniques, such as immunoferritin and immunochemocyanin techniques. Furthermore, the bound liposomes may not survive the harsh treatments in sample preparation for electron microscopy, even after fixation. At the light microscopy level, however, the size of the liposomes is still within the limiting spatial resolution of the optics. Actually, we consider the ultimate use of the liposome labeling technique
151 p r o b a b l y to be in exper i m ent s which require high sensitivity, high specificity labeling o f living cells, e.g., fluorescence activated cell sorting. We are c u r r en tly exploring these possibilities. ACKNOWLEDGEMENTS This work was jointly s uppor t e d by NIH Grant CA 24553 and by the Office o f Health and Environmental Research, D e p a r t m e n t of Energy under Contract W-7405-eng-26 with the Union Carbide Corporation. The fluorescence microscope system was f unde d in part by NIH Grant, 1-P41-GM 27545. A. Huang is a p o s t d o c t o r a l Investigator under Union Carbide Subc o n t r a c t 3322. REFERENCES Ames, B.N. and P.T. Dubin, 1960, J. Biol. Chem. 235,769. Cantley, L.C. and G.G. Hammes, 1975, Biochemistry 14, 2976. Chang, B.C. and L. Huang, 1979, Biochim. Biophys. Acta 565, 52. Enoch, H.G. and P. Strittmatter, 1979, Proc. Natl. Acad. Sci. U.S.A. 76, 145. Fothergill, J.E., 1969, in: Fluorescent Protein Tracing, ed. R.C. Nairn (Livingstone, Edinburgh) Ch. 3, p. 35. Hellings, J.A., H.H. Kamp, K.W.A. Wirtz and L.L.M. Van Deenen, 1974, Eur. J. Biochem. 47,601. Herrmann, S.H. and M.F. Mescher, 1979, J. Biol. Chem. 254, 8713. Huang, L. and R.E. Pagano, 1975, J. Cell Biol. 67, 38. Huang, A., L. Huang and S.J. Kennel, 1980a, J. Biol. Chem. 255, 8015. Huang, A., L. Huang and S.J. Kennel, 1980b, J. Cell Biol. 87, 85a. Kates, M., 1972, Techniques of Lipodology (North-Holland, Amsterdam) Ch. 5, p. 436. Kennel, S.J. and J.D. Feldman, 1976, Cancer Res. 36,200. Lapidot, Y., S. Pappoport and Y. Wolman, 1967, J. Lipid Res. 8,142. Leserman, L.D., J.N. Weinstein, R. Blumenthal, S.O. Sharrow and W.D. Terry, 1979, J. Immunol. 122,585. Oi, V.T., P.P. Jones, J.W. Goding, L.A. Herzenberg and L.A. Herzenberg, 1978, Curr. Top. Microbiol. Immunol. 81,115. Pagano, R.E. and L. Huang, 1975, J. Cell Biol. 67, 49. Struck, D.K. and R.E. Pagano, 1980, J. Biol. Chem. 255, 5404.
141
IMMUNOLIPOSOME LABELING: A SENSITIVE A N D SPECIFIC METHOD FOR CELL S U R F A C E LABELING
ANTHONY
HUANG
I, S T E P H E N
J. K E N N E L
I and L E A F H U A N G
2
I Biology Division, Oak Ridge National Laboratory, and 2 Department of Biochemistry,
University of Tennessee, Knoxville, TN 37916, U.S.A. (Received 20 March 1981, accepted 10 June 1981)
A simple, one-step procedure for fluorescence labeling of cultured cells with high sensitivity and specificity is described. We term this method immunoliposome labeling. Monoclonal IgG antibody was first covalently coupled with palmitic acid. The palmitoylIgG was then incorporated into unilamellar liposomes (about 100 nm diameter) containing either N-(4-nitrobenzo-2-oxa-l,3-diazole)- or N-(fluorescein isothiocyanyl)-phosphatidylethanolamine by a detergent-dialysis procedure. A monoclonal antibody to the mouse major histocompatibility antigen, H-2k, was tested as a model system. Fluorescent liposomes with covalently coupled anti-H-2k specifically labeled the mouse L-929 cells (H-2 k type), but not the A-31 cells (H-2 d type). The degree of labeling was quantitated by a microscope photometer. Cells labeled with fluorescent liposomes showed 4--6-fold stronger fluorescence than cells labeled with either fluorescein-conjugated antibody, or with primary antibody followed by fluorescein-conjugated secondary antibody (indirect immunofluorescence). Since different types of label (fluorescent, radioactive, etc.) can be incorporated into liposomes, this specific and sensitive method is potentially very versatile.
INTRODUCTION
Specific fluorescence labeling of cell surface antigens or receptors has become useful in several aspects of cell biology. Standard methods employ primary or secondary antibody or lectins which have been conjugated with proper fluorophores, such as fluorescein or rhodamine. Conventionally, indirect immunofluorescence is required to enhance the sensitivity of the method. However, these amplification steps can be time-consuming and can also increase background fluorescence significantly. They are therefore often unsuitable for many specific applications, such as the fluorescence activated Send correspondence to: Dr. Leaf Huang, Department of Biochemistry, University of Tennessee, Knoxville, TN 37916, U.S.A. Abbreviations: FITC, fluorescein isothiocyanate; IgG, immunoglobulin G; NBD, 4-nitrobenzo-2~axa-l,3-diazole; PBS, phosphate-buffered saline (137 mM NaC1, 2.7 mM KC1, 1.5 mM KH2PO4, 1 mM Na2HPO4, pH 7.8.); PC, phosphatidylcholine; PE, phosphatidylethanolamine. 0022-1759/81/0000--0000/$02.50 © 1981 Elsevier/North-Holland Biomedical Press
142
cell sorting which requires a rapid, sensitive, and specific labeling of living cells. Based on a previous finding in our laboratories, we have developed a one-step, sensitive technique which specifically labels the cell surface antigenic sites. The method involves the covalent coupling of antibody to liposomes and the subsequent binding of liposomes to the cells. We have named this method immunoliposome labeling. This method is quite different from that described by Leserman et al. (1979) which used liposomes containing haptens or antigens to label only cells bearing specific antibodies. Our method is therefore considerably more versatile. A preliminary report of this work has been presented elsewhere (Huang et al., 1980b). MA TER I ALS AND METHODS
Materials Anti-H-2 k IgG was purified from the culture medium of a hybridoma 114.1 cell line (Oi et al., 1978) by protein A affinity, chromatography (Herrmann and Mescher, 1979). Normal mouse IgG was similarly purified from normal mouse serum or ascites fluid induced with complete Freund's adjuvant. Goat antibody to mouse IgG was enriched from goat antiserum using immunoabsorbent technique (Kennel and Feldman, 1976). N-hydroxysuccinimide ester of palmitic acid was synthesized as described (Lapidot et al., 1967). Egg yolk phophatidylcholine (PC) and phosphatidylethanolamine (PE) were purified as described (Huang and Pagano, 1975). N-(4-nitrobenzo2-oxa-l,3
Coupling of IgG with palmitic acid The previously described method (Huang et al., 1980a) was followed with minor modifications. The molar ratio of N-hydroxysuccinimide ester of palmitic acid to IgG in the reaction mixture was 10:1 which gave a coupling efficiency of about 3--4 palmitoyl chains per IgG without significant decrease in antigen binding capacity (A. Huang, unpublished observation). The excess free palmitic acid was removed by Sephadex G-75 column chromatography as described (Huang et al., 1980a). Both anti-H-2 k and normal mouse IgG were treated identically.
Preparation of fluorescent liposomes A detergent
143 was used with minor modifications. Briefly, palmitoyl IgG and phospholipids (weight ratio 1:20) were mixed in phosphate-buffered saline (PB8) containing 0.7% deoxycholate. The phospholipids contained approximately 95 mole % egg yolk PC and 5 mole % NBD-PE or FITC-PE. The mixture was dialyzed for 40 h at r o o m temperature against PBS using Spectro-Pore 2 dialysis tubing (Spectrum). The turbid, yellowish dialysate was chromatographed on a Sepharose 4B column to remove unincorporated IgG. Liposomes eluted in the void volume were diluted with McCoy's medium and used for the cell labeling experiments.
Electron microscopy Liposomes containing antibody and fluorescent phospholipids were negatively stained with phosphotungstate as described (Chang and Huang, 1979). Samples were viewed in a Hitachi 600 electron microscope.
Preparation of FITC-labeled antibodies Labeling of anti-H-2 k or goat anti-mouse IgG with FITC was carried o u t as follows. Protein (4 mg in 0.4 ml PBS) was dialyzed against 60 ml carbonate buffer (0.025 M Na carbonate, pH 9.4) containing 6 mg FITC for 27 h at 4°C. The labeled antibody was then dialyzed extensively against PBS. The extent of labeling was estimated b y measuring absorbance at 495 nm and calculated according to the following equation (Fothergill, 1969): mole FITC/mole IgG = 2.8 X A49s/[IgG] in mg/ml. The degree of labeling was a b o u t 2 FITC/IgG for anti-H-2 k and 3 FITC/IgG for goat anti-mouse IgG.
Preparation of NBD-labeled antibodies Two mg anti-H-2 k or goat anti-mouse IgG (9.5 mg/ml in PBS) and 15 #1 N B D ~ l (4 mg/ml in dimethylsulfoxide) were added to 1.27 ml 0.025 M Na carbonate buffer, pH 9.4. The molar ratio of protein to N B D ~ I was approximately 1:24. Incubation was for 4 h in the dark at r o o m temperature. The mixture was applied to a Sephadex G-75 column equilibrated and eluted with PBS. Protein eluted in the void volume s h o w e d an absorption peak at 480 nm. The degree o f labeling, estimated b y using an extinction coefficient of 1.62 X 104/mol/cm (Cantley and Hammes, 1975), was a b o u t 3 NBD/IgG for anti-H-2 k and 4 NBD/IgG for goat anti-mouse IgG.
Fluorescence labeling of cells Mouse L-929 and A-31 cells were grown on glass coverslips (22 mm X 11 mm) in McCoy's medium with 10% d o n o r calf serum {Flow Laboratory). Coverslips were transferred to a 6-well 35 m m Linbro plate (2 coverslips per well) after washing once in PBS. Fluorescent liposome or antibody at 5 ~tg protein/ml was added and culture plates were gently rocked at r o o m temperature for 1 h. Coverslips were then washed 5 times in PBS and stored in PBS until observed under a microscope. For indirect immunofluorescence labeling, the incubation with unlabeled primary antibody (anti-H-2 k, 5 ~g/
144 ml) was for 1 h at r o o m temperature. After washing, the fluorescent secondary antibody (goat anti-mouse IgG) was added at 50 ~g/ml in McCoy's medium. After 1 h at r o o m temperature, coverslips were washed 5 times in PBS and observed under a microscope.
Fluorescence microscopy and quantitation Coverslips were placed with cell-side d o w n on glass slides. A Leitz fluorescence microscope equipped with epiluminescence was used. FITC and NBD fluorescences were observed with appropriate filters. Photomicrographs were taken by an automated camera. The quantitation of fluorescence was done immediately after the micrographs were taken and was measured on a new field of the same coverslip to minimize bleaching. A microscope p h o t o m e t e r (Leitz MPV-2) with an adjustable orifice was used. A small circle (3 pm diameter) which covered only a portion of the cell periphery was measured for relative fluorescence intensity. Only one cell in a given microscope field was measured, because prolonged illumination of a .given field produced significant bleaching. This is especially true for FITC-labeled liposomes with which the fluorescence bleached with a half time of a b o u t 15 sec. A measurement by the p h o t o m e t e r usually t o o k less than 1 sec, so that there was no significant bleaching during the measurement. All p h o t o m e t e r readings were expressed in arbitrary units after subtracting a background reading measured at an area with no cells. Other analytical methods Lipid phosphorus content was measured by the m e t h o d of Ames and Dubin (1960). Absorption was measured with a Cary 15 spectrophotometer and fluorescence with a Aminco-Bowman fluorospectrophotometer. RESULTS
Preparation of antibody-coated, fluorescence-labeled liposomes by the deoxycholate
145
Fig. 1. Negatively stained liposomes containing covalently coupled anti-H-2 k and NBDPE. Bar is 1 gin.
Liposome labeling of cells was compared with conventional immunofluorescence labeling techniques. When FITC-labeled anti-H-2 k was incubated with L-929 cells at a concentration identical to those in the liposomes (5 pg/ ml), only very weak fluorescence was observed at the cell periphery. With an indirect immunofluorescence technique, i.e., incubation with anti-H-2 k followed b y FITC-labeled goat anti-mouse IgG, stronger fluorescence was seen (Fig. 2a). The distribution of fluorescence at the cell surface was somewhat punctate in this case. This was in contrast to the liposome labeling which showed a more diffused distribution (Fig. 2c). Labeling of cells by either direct or indirect antibody m e t h o d was also specific for L-929 cells, since A-31 cells were not labeled under identical conditions (Fig. 3a). FITClabeled goat anti-mouse IgG alone did n o t label either cell t y p e (micrograph not shown). When NBD-labeled primary and secondary antibodies were tested for binding to cells, no significant fluorescence on either cell t y p e was
146
! ~ -
¸ ¸~i~
~
¸
~
~
"
i
Fig. 2. Phase-contrast (left panels) and fluorescence (right panels) photomicrographs of mouse L-929 cells labeled with: (a) anti-H-2 k followed by FITC-labeled goat anti-mouse IgG, photographic exposure time 2 rain 28 sec; (b) anti-H-2 k liposomes containing FITCPE, exposure time 3 rain 30 sec; (c) anti-H-2 k liposomes containing NBD-PE, exposure time 2 rain 2 sec; and (d) mouse IgG-liposomes containing NBD-PE, exposure time 2 rain. Bar is 10/zm.
147
$
"
q
Fig. 3. Phase-contrast (left panels) and fluorescence (right panels) p h o t o m i c r o g r a p h s of m o u s e A-31 cells. T r e a t m e n t s were the same as in Fig. 2, e x c e p t the p h o t o g r a p h i c exposure times were: (a) 5 rain; (b) 4 rain; (c) 5 rain; and (d) 2 rain. Bar is 10/~m.
148
observed. The reason for this failure of cell labeling b y NBD-antibody was unclear. We have estimated that each IgG molecule contained a b o u t 3--4 NBD molecules on the average. Labeling of antibody with NBD could have resulted in some denaturation o f the protein and the subsequent loss o f the antigen-binding activity. It should be emphasized that the fluorescence intensities of cells shown in Figs. 2 and 3 are somewhat misleading for t w o reasons. Firstly, these photographs were not taken with the same exposure time (ranging from 2 to 5 min). Secondly, bleaching o f the fluorophore during the exposure period significantly reduced the apparent fluorescence intensity. The latter problem was particularly serious for the FITC-PE label which was usually bleached within 30 sec (Fig. 2b). The degree o f fluorescence labeling o f cells was therefore quantitated by a microscope p h o t o m e t e r m e t h o d . The relative fluorescence intensities for b o t h L-929 and A-31 cells labeled b y various types of reagents are shown in Table 1. Although all cells were labeled, the degree of labeling by liposome or antibody varied substantially from cell to cell, or from one area o f a cell to another. It was therefore necessary to make a large number of measurements and analyze the data statistically. For the FITC label, liposome labeling gave a cell surface fluorescence which was a b o u t 5--6-fold higher than the direct immunofluorescence labeling and 4--5-fold higher than the indirect immunofluorescence labeling. The labeling was specific for the antigen-bearing cells, since only background fluorescence was obtained with A-31 cells. Similar results were obtained with the NBD label, except the NBD-labeled direct or indirect immunofluorescence was negative. NBD-labeled liposomes gave a b o u t the same degree of labeling intensity as the FITC-labeled liposomes under identical conditions, except the NBD label was much more resistant to
TABLE 1 Relative fluorescence intensitites o f labeled cells. Fluorophore
Cell
FITC
L-929 A-31
NBD
L-929 A-31
Anti.H.2 k a
3.2 -+ 1.9 (35) b --0.5 _+ 1.9 (22) 0.6 -+ 1.4 c (20) 0.3 -+ 1.4 (20)
Anti-H-2 k + goat × m o u s e IgG
Anti-H-2 kliposome
G o a t x m o u s e IgG
4.2 -+ 2.4 (46) 0.4 _+ 1.8 (30)
18 -+ 7.7 (26) --0.7 _+ 1.7 (26)
0.6 -+ 2.0 (34) 0.3 + 0.9 (2O)
0.8 + 1.3 c (20) ---0.2 _+ 1.8 (20)
18 -+ 9.6 (47) --0.2 _+ 0.9 (20)
0 . 2 -+ 3.7 (22) --0.9 _+ 1.0 (12)
a L o c a t i o n o f the f l u o r o p h o r e is indicated by parentheses. b N u m b e r of measurements. c L o w degree o f labeling p r o b a b l y due to N B D - a n t i b o d y d e n a t u r a t i o n ; see text.
149 photo-bleaching than FITC (bleaching half-time for NBD-PE liposomes was 1.5 min with the same illumination intensity). DISCUSSION We have previously shown that liposomes with covalently coupled antibody can bind specifically to cell surfaces of the antigen-bearing cells (Huang et al., 1980a). On the basis of this finding, we have designed an immunoliposome technique which can specifically label cell surface antigens with a fluorescence probe. Since liposomes are much larger than IgG, the amount of label that each liposome can carry is theoretically much greater than can be carried by the IgG molecule. We thus expect an improved labeling sensitivity over the conventional immunofluorescence technique. Using monoclonal antibody to the H-2 k antigen as a model system, we have shown a 4--6-fold stronger fluorescence with liposomes than with direct or indirect immunofluorescence. We have used 5 pg/ml anti-H-2 k, either free or liposome-bound, which produces a saturation binding on L-929 cells. Under these conditions, we expect to bind 1.1 X 104 liposomes per cell (A. Huang, unpublished data), assuming that there are approximately 8.6 X 104 phospholipid molecules per 100 nm liposome (Enoch and Strittmatter, 1979). This level of binding would bring about 4.8 X 107 fluorescent lipid molecules/cell at the cell surface. On the other hand, about 1.2 X 106 free anti-H-2 k molecules/cell can be bound to the L-929 cell surface at 5 pg/ml antibody concentration (P. Naylor, personal communication). This amounts to about 2.4--3.6 X 106 fluorescence molecules/cell, since there are 2--3 fluorescence molecules per IgG. This simple estimation would predict a 1 3 20-fold higher labeling by the liposome technique. In reality, we have only observed 4--5-fold increase over the direct antibody labeling. We can list at least 3 explanations for the apparent discrepancy. Firstly, the fluorescence quantum yield of the fluorophore on the liposome might be lower than that on the protein (Chang and Huang, 1979). Secondly, the amount of fluorescent phospholipid per liposome (about 5 mole %) might be high enough to generate self~luenching. Finally, the microscope photometer might give erroneously high readings for the weak fluorescence from the immunofluorescence labeling. This last possibility is quite plausible since we have noticed that the Leitz MPV-2 system consistently overestimates weak fluorescence. Despite these problems, the immunoliposome labeling has demonstrated a higher sensitivity than the conventional immunofluorescence techniques. The use of monoclonal antibody for the immunoliposome labeling is theoretically not essential. However, the use of conventional antibody preparations would result in lower amounts of fluorescence. This is because the number of IgG molecules per liposome (about 8--10) in the present study is not very large. Since most immune IgG preparations contain only 1--2% high affinity antibody, many liposomes would contain no antibody.
150
We are currently improving the IgG-liposome coupling method to obtain a higher amount of IgG per liposome. Until such method is available, the use of monoclonal antibody or some enriched immune IgG prepared by immunoabsorbent technique is still desirable. Cell surface labeling requires not only high sensitivity, but also high specificity. The immunoliposome labeling procedure is very specific for the antigen-bearing cells. We have previously shown that the binding of liposome to cell surface can be specifically blocked by pretreatment of cells with the unlabeled, free antibody (Huang et al., 1980a). Thus the binding is indeed mediated by specific antigen-antibody interactions. We have purposely chosen two fluorescent phospholipids with the fluorophores located at the head group position of the PE. It has been shown that the N-modified fluorescent PE does not undergo exchange reactions with the cellular phospholipids, a process that might contribute significantly to the background nonspecific labeling of the cell surface (Struck and Pagano, 1980). We have also chosen to use liposomes of about 100 nm in diameter, instead of the small unilamellar type (about 25 nm diameter} which are more prone to fuse with the cultured cells (Pagano and Huang, 1975). Non-specific fusion of liposome with cells would also increase the background labeling. In experiments shown here, fusion of bound liposomes with cells apparently did not occur, since significant amounts of the bound liposomes could be dissociated by a brief protease treatment (unpublished observation). Thus, the specificity of this cell-labeling method is also favorably compared to the conventional immunofluorescence methods. Since the phospholipids do not exchange between liposomes in the absence of exchange proteins (Hellings et al., 1974), the present method can be easily adapted for double fluorescence labeling. Such labeling would only require a one-step procedure using a mixture of two types of immunoliposome, each containing a different fluorescent phospholipid (e.g. rhodamine-PE and FITC-PE). Theoretically, the current method is not limited to the fluorescence labeling. Radioactive lipids or water-soluble molecules can be incorporated into the lipid or aqueous phase of liposomes, respectively. With a proper choice of the labeling molecule to avoid non-specific labeling or leakage, autoradiography of cells should also be possible. Certain limitations of the liposome labeling technique are worth mentioning. Since the liposomes used here are about 100 nm in diameter, any attempt to visualize the cell-bound liposomes at the electron microscopic level will be limited to a spatial resolution of, at best, about 100 nm, which is greater than the established techniques, such as immunoferritin and immunochemocyanin techniques. Furthermore, the bound liposomes may not survive the harsh treatments in sample preparation for electron microscopy, even after fixation. At the light microscopy level, however, the size of the liposomes is still within the limiting spatial resolution of the optics. Actually, we consider the ultimate use of the liposome labeling technique
151 p r o b a b l y to be in exper i m ent s which require high sensitivity, high specificity labeling o f living cells, e.g., fluorescence activated cell sorting. We are c u r r en tly exploring these possibilities. ACKNOWLEDGEMENTS This work was jointly s uppor t e d by NIH Grant CA 24553 and by the Office o f Health and Environmental Research, D e p a r t m e n t of Energy under Contract W-7405-eng-26 with the Union Carbide Corporation. The fluorescence microscope system was f unde d in part by NIH Grant, 1-P41-GM 27545. A. Huang is a p o s t d o c t o r a l Investigator under Union Carbide Subc o n t r a c t 3322. REFERENCES Ames, B.N. and P.T. Dubin, 1960, J. Biol. Chem. 235,769. Cantley, L.C. and G.G. Hammes, 1975, Biochemistry 14, 2976. Chang, B.C. and L. Huang, 1979, Biochim. Biophys. Acta 565, 52. Enoch, H.G. and P. Strittmatter, 1979, Proc. Natl. Acad. Sci. U.S.A. 76, 145. Fothergill, J.E., 1969, in: Fluorescent Protein Tracing, ed. R.C. Nairn (Livingstone, Edinburgh) Ch. 3, p. 35. Hellings, J.A., H.H. Kamp, K.W.A. Wirtz and L.L.M. Van Deenen, 1974, Eur. J. Biochem. 47,601. Herrmann, S.H. and M.F. Mescher, 1979, J. Biol. Chem. 254, 8713. Huang, L. and R.E. Pagano, 1975, J. Cell Biol. 67, 38. Huang, A., L. Huang and S.J. Kennel, 1980a, J. Biol. Chem. 255, 8015. Huang, A., L. Huang and S.J. Kennel, 1980b, J. Cell Biol. 87, 85a. Kates, M., 1972, Techniques of Lipodology (North-Holland, Amsterdam) Ch. 5, p. 436. Kennel, S.J. and J.D. Feldman, 1976, Cancer Res. 36,200. Lapidot, Y., S. Pappoport and Y. Wolman, 1967, J. Lipid Res. 8,142. Leserman, L.D., J.N. Weinstein, R. Blumenthal, S.O. Sharrow and W.D. Terry, 1979, J. Immunol. 122,585. Oi, V.T., P.P. Jones, J.W. Goding, L.A. Herzenberg and L.A. Herzenberg, 1978, Curr. Top. Microbiol. Immunol. 81,115. Pagano, R.E. and L. Huang, 1975, J. Cell Biol. 67, 49. Struck, D.K. and R.E. Pagano, 1980, J. Biol. Chem. 255, 5404.