Proc. Nati. Acad. Scf. USA Vol. 75, No. 6, pp. 2844-2848, June 1978 Cell Biology
"Panning" for lymphocytes: A method for cell selection (antibodies/plastic dishes/cell separation)
L. J. WYSOCKI AND V. L. SATO Cell and Developmental Biology, The Biological Laboratories, Harvard University, Cambridge, Massachusetts 02138
Communicated by Walter Gilbert, April 10, 1978
ABSTRACT We have developed a simple method for the fractionation of T and B lymphocytes. Plastic dishes coated with antibodies specific for mouse Ig selectively bind splenic B lymphocytes. The adherent cells are easily removed by gentle pipetting; both adherent and nonadherent populations retain immunologic function. In a typical experiment, when 3 X 107 splenic Iymphocytes were added to a 100 X 15 mm plastic dish coated with microgram quantities of anti-Ig, 98% of the nonadherent cells were Ig negative and 97% of the adherent cells were Ig positive. The method is sufficiently sensitive to allow detection and separation of cell types comprising as little as 2% of the total population and can be modified to allow the selection of cells by a double-antibody procedure. We believe that the plastic dish method will be generally useful for fractionating cells on the basis of their cell surface antigens. The ability to fractionate lymphocytes on the basis of surface phenotype has been a major technical advance in the study of the functional diversity of these cells. Many of the methods currently used exploit antibody specificity to separate cells of one type from a mixed population. Cell lysis with antibody and complement is useful only for cell elimination. Separation by affinity column chromatography (1) or fluorescence-activated cell sorting (2) offers the advantage of positive selection by allowing the recovery of both enriched and depleted cell populations. We report a simple and inexpensive method for the separation of B and T lymphocytes from the heterogeneous population found in murine spleen. It is based on the observation by Catt and Tregear (3) that antibody molecules adsorb onto polystyrene surfaces and bind antigen. We coat polystyrene dishes with antibody specific for cell surface antigens and permit cells to bind to such dishes. We show that this method can be used for either specific cell selection or specific cell elimination, that the fractionated cells are recovered in high yield with low levels of contamination by other cell types, and that the recovered cells retain immunologic function. MATERIALS AND METHODS Animals. BALB/cCRL and C57BL/6 mice were purchased from Charles River Laboratories. (BALB/c X C57BL/6)F1 mice were bred in the animal rooms of The Biological Laboratories, Harvard University. All mice were maintained in filter-top cages and given food and water ad lib. New Zealand White rabbits were purchased from Pine Acre Rabbitry and White Pine Rabbitry. Antisera. Rabbit anti-thymocyte antiserum was raised by subcutaneous injection of 1 X 108 BALB/c thymocytes in isotonic saline followed by four identical boosting injections at 2-week intervals. Crude serum was precipitated with ammonium sulfate at 45% saturation (4) and adsorbed with BALB/c liver cells to eliminate anti-H-2 reactivity from the serum. The The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
T cell specificity of the serum was determined by complement-dependent lysis of thymocytes and specific immunoprecipitation of T25 (Thy 1 antigen) from thymocyte membranes (5). The cytotoxic titer of this serum was very low (95% of thymocytes were lysed at an antibody dilution of 1:5). Rabbit anti-mouse Ig was raised by repeated subcutaneous injection of DEAE-cellulose-purified mouse Ig in complete Freund's adjuvant. The Ig fraction was purified by ammonium sulfate precipitation (4) and affinity chromatography on Sepharose 4B conjugated with purified mouse Ig (6). Goat anti-rabbit IgG was purchased as serum from SeraSource, Inc., and purified by ammonium sulfate precipitation and affinity chromatography on Sepharose 4B conjugated with DEAE-cellulose-purified rabbit IgG. Normal rabbit IgG was prepared by ion exchange chromatography on Whatman DE-52 ion exchange resin (7). F(ab')2 fragments were prepared by pepsin digestion and gel filtration through Sephadex G-150 (4). Antibodies conjugated with rhodamine were prepared by the method of Cebra and Goldstein (7). Purity of all antibody reagents was determined by sodium dodecyl sulfate/polyacrylamide gel electrophoresis (8) under reducing conditions. All immunoglobulin preparations were centrifuged at 80,000 X g for 1 hr to remove aggregates, sterilized by Millipore filtration, and stored at 4°. Fetal Calf Serum. Fetal calf serum, purchased from Grand Island Biological Co., was heat inactivated at 560 for 2 hr and tested for mitogenic activity on splenic lymphocytes. Batches with low endogenous mitogenic activity were used in experiments involving [3H]thymidine uptake and batches with high endogenous activity were selected for Mishell-Dutton cultures. Membrane Immunofluorescence. Pelleted cells (5 X 106) were resuspended in 0.1 ml of a solution containing tetramethylrhodamine isothiocyanate-conjugated F(ab')2 rabbit anti-mouse Ig at a concentration of 100 ,ug/ml and stained at 40 for 20 min. The cell suspension was centrifuged through a layer of fetal calf serum to remove unreacted antibody and washed again in phosphate-buffered saline (Pi/NaCl) containing 5% (vol/vol) fetal calf serum. Immunofluorescence was determined by using a Zeiss fluorescence microscope with epi-illumination. Functional Assays. Mitogen responses to phytohemagglutinin (2.5 ,g/ml), concanavalin A (2.5 Lg/ml), and lipopolysaccharide (25 ,g/ml) were determined as described (9). Mixed lymphocyte responses were measured in microculture using 5 X 105 F1 unseparated splenic lymphocytes as stimulator cells and either 2.5 X 105 or 5 X 105 C57BL/6 lymphocytes as responder cells. Cultures were pulsed at 72 hr with 1 liCi of [3H]thymidine and harvested 12 hr later on an automatic harvester (Bioplastics). Abbreviations: P,/NaCI, phosphate-buffered saline; PFC, plaque forming cells. 2844
Cell Biology:
Nati. Acad. Sato andnSatonProc. Wysocki and
Both mitogen and mixed lymphocyte responses were performed in triplicate and reported as the mean cpm. Antibody responses were assayed by Mishell-Dutton culture (10). Mice were primed with 1 X 108 sheep erythrocytes given intravenously 17 days before use; 1 X 106 lymphocytes were plated in each well and boosted with 2 X 105 sheep erythrocytes. Cultures were harvested at day 5 and the number of plaqueforming cells was determined by the Cunningham modification of the Jerne plaque assay (11). Cell Separation. Fisher 100 X 15 mm polystyrene bacteriological petri dishes (cat. no. 8-757-12) were used in all experiments. (We have found that use of tissue culture grade plastic dishes increases the levels of nonspecific cell sticking and do not recommend their use.) Antibodies to be adhered to the plates were diluted in 0.05 M Tris, pH 9.5, and 10 ml was poured immediately onto the plates. The solution was swirled on each plate until the bottom surface was evenly covered. After 40 min at room temperature, the buffer was decanted, and the dishes were washed four times with Pi/NaCl and once with a few milliliters of Pi/NaCl made 1% in fetal calf serum. For direct binding, 2-3 X 107 nucleated cells were suspended in 3 ml of Pi/NaCl/5% fetal calf serum and poured onto each antibody-coated dish, with care taken to avoid bubbles. If more than 2 X 107 nucleated cells were to be poured per dish, the cell preparation was cleared of erythrocytes by ammonium chloride treatment. The plates were incubated on a level surface at 40 for 70 min. After 40 min, unattached cells were redistributed by tilting and swirling the plate. At the end of 30 more min, the nonadherent cells were removed by again swirling the dish and decanting the supernatant. Plates were washed by gently pouring 5-10 ml of Pi/NaCl/1% fetal calf serum down the side wall of the dish and then swirling, tilting, and decanting. If only the nonadherent population was to be used (as in a cell elimination experiment), two additional washes were sufficient. If cells were to be recovered from the plates (positive selection), five washes were performed. To recover the bound cells, the plate was filled with 25-30 ml of Pi/NaCl/1% fetal calf serum and the entire surface of the plate was flushed by using a pasteur pipette. For indirect binding, 3 X 107 cells were treated with 0.5 ml of rabbit antibody for 20 min. The cells were washed twice in
Pi/NaCl/5% fetal calf serum and poured onto dishes previously coated with pure goat anti-rabbit IgG at 10 jtg/ml. The remainder of the procedure was identical to that described for direct binding. RESULTS Conditions for Specific Cell Depletion. Polystyrene petri dishes coated with rabbit anti-mouse Ig or its F(ab')2 fragment bound splenic Ig+ cells. When a suspension containing approximately 50% Ig+ cells was incubated on these dishes, the effluent population contained only 1-3% Ig+ cells (Table 1). More than 90% of the Ig- cells added to the dishes were recovered in the effluent. Because the cell recoveries were so high, it was possible to repeat the selection without significant cell loss; the last line of Table 1 shows that two cycles of selection resulted in very low levels of contamination by Ig+ cells. Plates coated with normal rabbit IgG or its F(ab')2 fragment gave cell recoveries ranging from 97 to 99%, indicating that cell adherence depends on the presence of specific antibody on the dish. Although the whole rabbit anti-mouse Ig antibody was slightly more efficient than its F(ab')2 fragment, it is clear that the binding of antibody to the plates is not strictly dependent on the Fc portions of the molecule. The * time necessary for coating dishes with antibody varied
Sci. USA 75 (1978)
2845
Table 1. Depletion of Ig+ cells by using rabbit anti-mouse Igcoated plastic dishes % cells recovered
Plates coated with NR IgG RAM IgI F(ab')2 RAM Ig§ RAM Ig (2 cycles)
in effluent
% Ig+ *
98 + 1.Ot 41.1 + 3.6 38.3 i 1.7
50.4 + 4.5t 1.8 : 0.54 2.6 + 1.1 95% of the Ig+ cells from a splenic lymphocyte population. Table 3 shows that, if a longer coating time is allowed, antibody solutions as dilute as 1 tig/ml can be used to effect strong selection. We find that no more than 3 X 107 lymphocytes can be put on a dish without sacrificing the quality of the recovered population (data not shown). To assess the effect of cold storage on antibody-coated plates, we coated two dishes with rabbit anti-mouse Ig, washed them with tap water, and subjected them to five cycles of freezethawing over a period of 2 days. This treatment did not affect the cell selection; effluent cell populationswere recovered with high yield and contained less than 1% Ig+ cells (data not shown). Modifications for Positive Cell Selection. Ig+ cells attached to plates that had been coated with affinity-purified rabbit anti-mouse Ig were bound so tightly that they could not be recovered. Vigorous pipetting, incubation at 370, and even competition with normal mouse Ig or rabbit anti-mouse Ig in solution did not release the cells. To recover cells in a positive selection procedure, we decreased the number of specific associations between the cells and the dish by using a mixture of pure rabbit anti-mouse Ig and normal rabbit IgG for coating. When the pure rabbit anti-mouse Ig was diluted 1:10 with normal IgG, a gentle stream of medium impinging on the plate released the cells (Table 4). We achieved high cell recovery and excellent enrichment (97% Ig+) even at normal:anti-mouse ratios of 40:1. The 1-3% of Ig- cells found in the selected population is presumably that fraction (0.4-4%) of cells that binds to plates coated with normal rabbit IgG. No attempt was made to remove intrinsically adherent cells such as macrophages prior to cell selection. When F(ab')2 fragments of normal rabbit IgG and rabbit Table 2. Time required for coating dishes with antibody* % cells recovered % Ig+* t in effluent Time, min Uncoated 100 52.2 5 10 20 40
70 51.2 44 40.3
34.8 11.6 4.4 1.9
All spleen cell suspensions used in these experiments were treated with ammonium chloride to eliminate erythrocytes. Ig+ cells were detected by fluorescence microscopy. * Antibody solution, 5 gg/ml. t t least 500 cells were counted.
Cell Biology: Wysocki and Sato
2846
Proc. Nati. Acad. Sci. USA 75 (1978)
Table 3. Antibody concentration required for coating dishes* Protein % cells concentration, recovered Ug/ml in effluent % Ig+ t 0 100 43 5 57 1.5 2.5 46 1.6 1.0 52 4.4 0.5 54 8.7 0.25 65 15.2 0.10 100 46.2 All spleen cell suspensions used in these experiments were treated with ammonium chloride to eliminate erythrocytes. Ig+ cells were detected by fluorescence microscopy. * Dishes were coated with antibody solutions for 90 min. t At least 500 cells were counted.
Table 4. Recovery of Ig+ cells from plates coated with rabbit anti-mouse Ig diluted with normal rabbit IgG % cells IgG, Ag/ml Rabbit recovered Normal anti-mouse from effluent % Ig+* 10 10 10
0.0 1.0 0.5 0.25
10
0.0625
10
Average cpm/culture X
Avg. cpm/ _
TZ
Cell source | B _
_
_
_
TFOb
__
Mitogen _
_
_
| unsep.
+
*
_
+
CONA
+
LPS
108,101 120,069
PHA CONA
1,490 156,601 69,999
PHA CONA
+
LPS -
2,593
+
PHA
284
+
CONA
427
+
LPS
+
+
+
PHA
+
+
CONA
+
+
LPS
+
+
-
+
+
PHA
+
+
+
+
CONA _
LPS
4
6
8
10-4 12
10
14
16
18
3,328 1,952 109,542 86,631
+
+
_2
7,327 I1 151,776
+
+
_
_
PHA
LPS +
_
+ +
.9 +
culture
97.1 + 0.4 95.6 1 1.4
was determined by the state of the anti-Ig and not by that of the diluent antibody. It is possible that whole Ig molecules bind to the dishes at a greater density than F(ab')2 fragments or that Fc portions on intact molecules allow the display of more binding sites to the cells. Modification for Cell Selection with Double Antibody (Indirect Selection) For indirect selection, cells are coated with specific rabbit antibody as for immunofluorescence. They are then allowed to bind to dishes coated with affinity-purified goat anti-rabbit IgG. Table 5 shows that cells pretreated with rabbit
nonspecific to specific antibody should not exceed 10:1; at higher ratios, cells did not bind efficiently to the dish. This difference between the whole antibodies and their F(ab')2 fragments is not due to the Fc receptor on Ig+ cells: in experiments using mixtures of either whole rabbit anti-mouse Ig and F(ab')2 of normal rabbit IgG or whole normal rabbit IgG and F(ab')2 of rabbit anti-mouse Ig, the efficiency of cell binding
_
44.2 ± 5.5t 97.3 :: 0.3 97.5 + 0.4
Splenic lymphocytes (2 X 107) were bound to the dishes. The number of Ig+ cells was determined by fluorescence microscopy. * At least 500 cells were counted. t i average deviation from the mean.
anti-mouse Ig were used in a positive selection, the ratio of
__
1.8 ± 1.2t 37.2 ± 4.2 34.0 + 0.1 31.0 4:2.1 9.8 ± 6.5
9,422
162,160 7,095 156,283
3-
B
86,314 107,479 8,064 171,088 92,771 113,726
H
L
_
__
_
II
FIG. 1. Mitogen responsiveness of separated populations; 3 X 105 cells were cultured in each well. "B," "Tbg," and "TF(ab')2"V fractions were 96.4,2.5, and 2.4% Ig+, respectively, by immunofluorescence. Mitogens: PHA, phytohemagglutinin; CONA, concanavalin A; LPS, lipopolysaccharide.
Cell Biology:
Proc. Natl. Acad. Sci. USA 75 (1978)
Wysocki and Sato
Table 6. Recovery of a minor population of Ig+ cells from a population of thymocytes % cells % cells recovered recovered % Ig+ t % Ig+ * from dishes in effluent
Table 5. Indirect selection on antibody coated dishes % cells recovered from Antibody bound Antibody bound plates to plate to cells, Ag/ml
100 97 100
Rabbit anti-thymocyte: 45.6 Goat anti-rabbit IgG 100 44.3 50 34.6 25 Normal rabbit:* 0.8 Goat anti-rabbit IgG 100 2.6 50 2.0 25 1.8 Normal goat Ig* Untreated 2 X 107 splenic lymphocytes were coated with rabbit anti-thymocyte antibody and bound to the dishes. * Ig fraction was isolated from whole serum by ammonium sulfate precipitation only.
unseparated
-
2.6
unseparated
+
42.4
B
-
0.4
B
+
T, g
+
0.0 2.6
TFd b
+
5.2
B + T
+
43.2
B +
+
46.5
TFab2
93.7 91.8 92.6
1.4 1.7 1.6
"Panning" for lymphocytes: A method for cell selection (antibodies/plastic dishes/cell separation)
L. J. WYSOCKI AND V. L. SATO Cell and Developmental Biology, The Biological Laboratories, Harvard University, Cambridge, Massachusetts 02138
Communicated by Walter Gilbert, April 10, 1978
ABSTRACT We have developed a simple method for the fractionation of T and B lymphocytes. Plastic dishes coated with antibodies specific for mouse Ig selectively bind splenic B lymphocytes. The adherent cells are easily removed by gentle pipetting; both adherent and nonadherent populations retain immunologic function. In a typical experiment, when 3 X 107 splenic Iymphocytes were added to a 100 X 15 mm plastic dish coated with microgram quantities of anti-Ig, 98% of the nonadherent cells were Ig negative and 97% of the adherent cells were Ig positive. The method is sufficiently sensitive to allow detection and separation of cell types comprising as little as 2% of the total population and can be modified to allow the selection of cells by a double-antibody procedure. We believe that the plastic dish method will be generally useful for fractionating cells on the basis of their cell surface antigens. The ability to fractionate lymphocytes on the basis of surface phenotype has been a major technical advance in the study of the functional diversity of these cells. Many of the methods currently used exploit antibody specificity to separate cells of one type from a mixed population. Cell lysis with antibody and complement is useful only for cell elimination. Separation by affinity column chromatography (1) or fluorescence-activated cell sorting (2) offers the advantage of positive selection by allowing the recovery of both enriched and depleted cell populations. We report a simple and inexpensive method for the separation of B and T lymphocytes from the heterogeneous population found in murine spleen. It is based on the observation by Catt and Tregear (3) that antibody molecules adsorb onto polystyrene surfaces and bind antigen. We coat polystyrene dishes with antibody specific for cell surface antigens and permit cells to bind to such dishes. We show that this method can be used for either specific cell selection or specific cell elimination, that the fractionated cells are recovered in high yield with low levels of contamination by other cell types, and that the recovered cells retain immunologic function. MATERIALS AND METHODS Animals. BALB/cCRL and C57BL/6 mice were purchased from Charles River Laboratories. (BALB/c X C57BL/6)F1 mice were bred in the animal rooms of The Biological Laboratories, Harvard University. All mice were maintained in filter-top cages and given food and water ad lib. New Zealand White rabbits were purchased from Pine Acre Rabbitry and White Pine Rabbitry. Antisera. Rabbit anti-thymocyte antiserum was raised by subcutaneous injection of 1 X 108 BALB/c thymocytes in isotonic saline followed by four identical boosting injections at 2-week intervals. Crude serum was precipitated with ammonium sulfate at 45% saturation (4) and adsorbed with BALB/c liver cells to eliminate anti-H-2 reactivity from the serum. The The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
T cell specificity of the serum was determined by complement-dependent lysis of thymocytes and specific immunoprecipitation of T25 (Thy 1 antigen) from thymocyte membranes (5). The cytotoxic titer of this serum was very low (95% of thymocytes were lysed at an antibody dilution of 1:5). Rabbit anti-mouse Ig was raised by repeated subcutaneous injection of DEAE-cellulose-purified mouse Ig in complete Freund's adjuvant. The Ig fraction was purified by ammonium sulfate precipitation (4) and affinity chromatography on Sepharose 4B conjugated with purified mouse Ig (6). Goat anti-rabbit IgG was purchased as serum from SeraSource, Inc., and purified by ammonium sulfate precipitation and affinity chromatography on Sepharose 4B conjugated with DEAE-cellulose-purified rabbit IgG. Normal rabbit IgG was prepared by ion exchange chromatography on Whatman DE-52 ion exchange resin (7). F(ab')2 fragments were prepared by pepsin digestion and gel filtration through Sephadex G-150 (4). Antibodies conjugated with rhodamine were prepared by the method of Cebra and Goldstein (7). Purity of all antibody reagents was determined by sodium dodecyl sulfate/polyacrylamide gel electrophoresis (8) under reducing conditions. All immunoglobulin preparations were centrifuged at 80,000 X g for 1 hr to remove aggregates, sterilized by Millipore filtration, and stored at 4°. Fetal Calf Serum. Fetal calf serum, purchased from Grand Island Biological Co., was heat inactivated at 560 for 2 hr and tested for mitogenic activity on splenic lymphocytes. Batches with low endogenous mitogenic activity were used in experiments involving [3H]thymidine uptake and batches with high endogenous activity were selected for Mishell-Dutton cultures. Membrane Immunofluorescence. Pelleted cells (5 X 106) were resuspended in 0.1 ml of a solution containing tetramethylrhodamine isothiocyanate-conjugated F(ab')2 rabbit anti-mouse Ig at a concentration of 100 ,ug/ml and stained at 40 for 20 min. The cell suspension was centrifuged through a layer of fetal calf serum to remove unreacted antibody and washed again in phosphate-buffered saline (Pi/NaCl) containing 5% (vol/vol) fetal calf serum. Immunofluorescence was determined by using a Zeiss fluorescence microscope with epi-illumination. Functional Assays. Mitogen responses to phytohemagglutinin (2.5 ,g/ml), concanavalin A (2.5 Lg/ml), and lipopolysaccharide (25 ,g/ml) were determined as described (9). Mixed lymphocyte responses were measured in microculture using 5 X 105 F1 unseparated splenic lymphocytes as stimulator cells and either 2.5 X 105 or 5 X 105 C57BL/6 lymphocytes as responder cells. Cultures were pulsed at 72 hr with 1 liCi of [3H]thymidine and harvested 12 hr later on an automatic harvester (Bioplastics). Abbreviations: P,/NaCI, phosphate-buffered saline; PFC, plaque forming cells. 2844
Cell Biology:
Nati. Acad. Sato andnSatonProc. Wysocki and
Both mitogen and mixed lymphocyte responses were performed in triplicate and reported as the mean cpm. Antibody responses were assayed by Mishell-Dutton culture (10). Mice were primed with 1 X 108 sheep erythrocytes given intravenously 17 days before use; 1 X 106 lymphocytes were plated in each well and boosted with 2 X 105 sheep erythrocytes. Cultures were harvested at day 5 and the number of plaqueforming cells was determined by the Cunningham modification of the Jerne plaque assay (11). Cell Separation. Fisher 100 X 15 mm polystyrene bacteriological petri dishes (cat. no. 8-757-12) were used in all experiments. (We have found that use of tissue culture grade plastic dishes increases the levels of nonspecific cell sticking and do not recommend their use.) Antibodies to be adhered to the plates were diluted in 0.05 M Tris, pH 9.5, and 10 ml was poured immediately onto the plates. The solution was swirled on each plate until the bottom surface was evenly covered. After 40 min at room temperature, the buffer was decanted, and the dishes were washed four times with Pi/NaCl and once with a few milliliters of Pi/NaCl made 1% in fetal calf serum. For direct binding, 2-3 X 107 nucleated cells were suspended in 3 ml of Pi/NaCl/5% fetal calf serum and poured onto each antibody-coated dish, with care taken to avoid bubbles. If more than 2 X 107 nucleated cells were to be poured per dish, the cell preparation was cleared of erythrocytes by ammonium chloride treatment. The plates were incubated on a level surface at 40 for 70 min. After 40 min, unattached cells were redistributed by tilting and swirling the plate. At the end of 30 more min, the nonadherent cells were removed by again swirling the dish and decanting the supernatant. Plates were washed by gently pouring 5-10 ml of Pi/NaCl/1% fetal calf serum down the side wall of the dish and then swirling, tilting, and decanting. If only the nonadherent population was to be used (as in a cell elimination experiment), two additional washes were sufficient. If cells were to be recovered from the plates (positive selection), five washes were performed. To recover the bound cells, the plate was filled with 25-30 ml of Pi/NaCl/1% fetal calf serum and the entire surface of the plate was flushed by using a pasteur pipette. For indirect binding, 3 X 107 cells were treated with 0.5 ml of rabbit antibody for 20 min. The cells were washed twice in
Pi/NaCl/5% fetal calf serum and poured onto dishes previously coated with pure goat anti-rabbit IgG at 10 jtg/ml. The remainder of the procedure was identical to that described for direct binding. RESULTS Conditions for Specific Cell Depletion. Polystyrene petri dishes coated with rabbit anti-mouse Ig or its F(ab')2 fragment bound splenic Ig+ cells. When a suspension containing approximately 50% Ig+ cells was incubated on these dishes, the effluent population contained only 1-3% Ig+ cells (Table 1). More than 90% of the Ig- cells added to the dishes were recovered in the effluent. Because the cell recoveries were so high, it was possible to repeat the selection without significant cell loss; the last line of Table 1 shows that two cycles of selection resulted in very low levels of contamination by Ig+ cells. Plates coated with normal rabbit IgG or its F(ab')2 fragment gave cell recoveries ranging from 97 to 99%, indicating that cell adherence depends on the presence of specific antibody on the dish. Although the whole rabbit anti-mouse Ig antibody was slightly more efficient than its F(ab')2 fragment, it is clear that the binding of antibody to the plates is not strictly dependent on the Fc portions of the molecule. The * time necessary for coating dishes with antibody varied
Sci. USA 75 (1978)
2845
Table 1. Depletion of Ig+ cells by using rabbit anti-mouse Igcoated plastic dishes % cells recovered
Plates coated with NR IgG RAM IgI F(ab')2 RAM Ig§ RAM Ig (2 cycles)
in effluent
% Ig+ *
98 + 1.Ot 41.1 + 3.6 38.3 i 1.7
50.4 + 4.5t 1.8 : 0.54 2.6 + 1.1 95% of the Ig+ cells from a splenic lymphocyte population. Table 3 shows that, if a longer coating time is allowed, antibody solutions as dilute as 1 tig/ml can be used to effect strong selection. We find that no more than 3 X 107 lymphocytes can be put on a dish without sacrificing the quality of the recovered population (data not shown). To assess the effect of cold storage on antibody-coated plates, we coated two dishes with rabbit anti-mouse Ig, washed them with tap water, and subjected them to five cycles of freezethawing over a period of 2 days. This treatment did not affect the cell selection; effluent cell populationswere recovered with high yield and contained less than 1% Ig+ cells (data not shown). Modifications for Positive Cell Selection. Ig+ cells attached to plates that had been coated with affinity-purified rabbit anti-mouse Ig were bound so tightly that they could not be recovered. Vigorous pipetting, incubation at 370, and even competition with normal mouse Ig or rabbit anti-mouse Ig in solution did not release the cells. To recover cells in a positive selection procedure, we decreased the number of specific associations between the cells and the dish by using a mixture of pure rabbit anti-mouse Ig and normal rabbit IgG for coating. When the pure rabbit anti-mouse Ig was diluted 1:10 with normal IgG, a gentle stream of medium impinging on the plate released the cells (Table 4). We achieved high cell recovery and excellent enrichment (97% Ig+) even at normal:anti-mouse ratios of 40:1. The 1-3% of Ig- cells found in the selected population is presumably that fraction (0.4-4%) of cells that binds to plates coated with normal rabbit IgG. No attempt was made to remove intrinsically adherent cells such as macrophages prior to cell selection. When F(ab')2 fragments of normal rabbit IgG and rabbit Table 2. Time required for coating dishes with antibody* % cells recovered % Ig+* t in effluent Time, min Uncoated 100 52.2 5 10 20 40
70 51.2 44 40.3
34.8 11.6 4.4 1.9
All spleen cell suspensions used in these experiments were treated with ammonium chloride to eliminate erythrocytes. Ig+ cells were detected by fluorescence microscopy. * Antibody solution, 5 gg/ml. t t least 500 cells were counted.
Cell Biology: Wysocki and Sato
2846
Proc. Nati. Acad. Sci. USA 75 (1978)
Table 3. Antibody concentration required for coating dishes* Protein % cells concentration, recovered Ug/ml in effluent % Ig+ t 0 100 43 5 57 1.5 2.5 46 1.6 1.0 52 4.4 0.5 54 8.7 0.25 65 15.2 0.10 100 46.2 All spleen cell suspensions used in these experiments were treated with ammonium chloride to eliminate erythrocytes. Ig+ cells were detected by fluorescence microscopy. * Dishes were coated with antibody solutions for 90 min. t At least 500 cells were counted.
Table 4. Recovery of Ig+ cells from plates coated with rabbit anti-mouse Ig diluted with normal rabbit IgG % cells IgG, Ag/ml Rabbit recovered Normal anti-mouse from effluent % Ig+* 10 10 10
0.0 1.0 0.5 0.25
10
0.0625
10
Average cpm/culture X
Avg. cpm/ _
TZ
Cell source | B _
_
_
_
TFOb
__
Mitogen _
_
_
| unsep.
+
*
_
+
CONA
+
LPS
108,101 120,069
PHA CONA
1,490 156,601 69,999
PHA CONA
+
LPS -
2,593
+
PHA
284
+
CONA
427
+
LPS
+
+
+
PHA
+
+
CONA
+
+
LPS
+
+
-
+
+
PHA
+
+
+
+
CONA _
LPS
4
6
8
10-4 12
10
14
16
18
3,328 1,952 109,542 86,631
+
+
_2
7,327 I1 151,776
+
+
_
_
PHA
LPS +
_
+ +
.9 +
culture
97.1 + 0.4 95.6 1 1.4
was determined by the state of the anti-Ig and not by that of the diluent antibody. It is possible that whole Ig molecules bind to the dishes at a greater density than F(ab')2 fragments or that Fc portions on intact molecules allow the display of more binding sites to the cells. Modification for Cell Selection with Double Antibody (Indirect Selection) For indirect selection, cells are coated with specific rabbit antibody as for immunofluorescence. They are then allowed to bind to dishes coated with affinity-purified goat anti-rabbit IgG. Table 5 shows that cells pretreated with rabbit
nonspecific to specific antibody should not exceed 10:1; at higher ratios, cells did not bind efficiently to the dish. This difference between the whole antibodies and their F(ab')2 fragments is not due to the Fc receptor on Ig+ cells: in experiments using mixtures of either whole rabbit anti-mouse Ig and F(ab')2 of normal rabbit IgG or whole normal rabbit IgG and F(ab')2 of rabbit anti-mouse Ig, the efficiency of cell binding
_
44.2 ± 5.5t 97.3 :: 0.3 97.5 + 0.4
Splenic lymphocytes (2 X 107) were bound to the dishes. The number of Ig+ cells was determined by fluorescence microscopy. * At least 500 cells were counted. t i average deviation from the mean.
anti-mouse Ig were used in a positive selection, the ratio of
__
1.8 ± 1.2t 37.2 ± 4.2 34.0 + 0.1 31.0 4:2.1 9.8 ± 6.5
9,422
162,160 7,095 156,283
3-
B
86,314 107,479 8,064 171,088 92,771 113,726
H
L
_
__
_
II
FIG. 1. Mitogen responsiveness of separated populations; 3 X 105 cells were cultured in each well. "B," "Tbg," and "TF(ab')2"V fractions were 96.4,2.5, and 2.4% Ig+, respectively, by immunofluorescence. Mitogens: PHA, phytohemagglutinin; CONA, concanavalin A; LPS, lipopolysaccharide.
Cell Biology:
Proc. Natl. Acad. Sci. USA 75 (1978)
Wysocki and Sato
Table 6. Recovery of a minor population of Ig+ cells from a population of thymocytes % cells % cells recovered recovered % Ig+ t % Ig+ * from dishes in effluent
Table 5. Indirect selection on antibody coated dishes % cells recovered from Antibody bound Antibody bound plates to plate to cells, Ag/ml
100 97 100
Rabbit anti-thymocyte: 45.6 Goat anti-rabbit IgG 100 44.3 50 34.6 25 Normal rabbit:* 0.8 Goat anti-rabbit IgG 100 2.6 50 2.0 25 1.8 Normal goat Ig* Untreated 2 X 107 splenic lymphocytes were coated with rabbit anti-thymocyte antibody and bound to the dishes. * Ig fraction was isolated from whole serum by ammonium sulfate precipitation only.
unseparated
-
2.6
unseparated
+
42.4
B
-
0.4
B
+
T, g
+
0.0 2.6
TFd b
+
5.2
B + T
+
43.2
B +
+
46.5
TFab2
93.7 91.8 92.6
1.4 1.7 1.6
