Cell separation based on the reversible interaction between calmodulin and a calmodulin-binding peptide

JOUMAL OF I-ICAL YElHOUS

EISEVIER

Journal of Immunological Methods 2 12 ( 1998) 69-78

Cell separation based on the reversible interaction between calmodulin and a calmodulin-binding peptide Robert J. Colinas The Wadsworth Cmter, New York State Department

*, Anne C. Walsh ofHealth, P.O. Box 509, Albany. NY 12201-0509. USA

Received 13 August 1997; revised 5 December 1997; accepted 10 December 1997

Abstract A cell separation system based on the calcium-dependent interaction of calmodulin (CM) with a calmodulin-binding peptide (CBP) has been developed. The prototype of this system utilizes an indirect method to label the target cell population, Cells are first labeled with a primary monoclonal antibody directed to a specific cell surface antigen, then with a secondary affinity reagent, consisting of a polyclonal goat anti-mouse IgG (GAM-IgG) that has been cross-linked to a CBP derived from the sequence of the rabbit skeletal muscle myosin light chain kinase. In the presence of Ca’+, the CBP on the cells labeled with CAM-IgG-CBP binds to biotinylated calmodulin (CM-Biotin) with high affinity. The target cells are then captured with a solid-phase streptavidin. The unbound non-target cells are washed away and the immobilized target cells are released by chelating Cal+ with EGTA. The specificity of the GAM-IgG-CBP and CM-Biotin and the feasibility of using this system to separate cells was demonstrated using the KG-l human acute myelogenous leukemia cell line. KG-l cells were fractionated on the basis of cell surface expression of HLA-DR. The cell selection reagents and the cell separation process did not affect KG- 1 cell viability while cells selected by this procedure were 90% pure with a yield of 75%. This cell separation system also was used for rare cell isolation from normal human peripheral blood mononuclear cells. T cells expressing the VP5 T cell receptor. which represent < 5% of the unfractionated cells, were isolated with 89% viability, 72% purity, 80% yield, and retained the ability to respond to activation signals as measured by blast transformation. The results from this study show that a cell selection system based on the reversible interaction between CM and a CBP can be applied to gently and efficiently isolate cells from a heterogeneous starting population that are free of the solid matrix without exposure to the stresses of mechanical or enzymatic release. 0 1998 Elsevier Science B.V. All rights reserved. K~JwYw~~: Cell separation;

Calmodulin:

Calmodulin-binding

peptide

1. Introduction Abbreviations: HPC, hematopoietic stem/progenitor cells: CM. calmodulin: CM-Biotin. biotinylated calmodulin: CM-FITC. fluoresceinated calmodulin: CBP, calmodulin-binding peptide; GAM-IgG. Goat anti-mouse IgG; MPI, mean fluorescence intenantibody; PBMC, peripheral blood sity: Mab. monoclonal mononuclear cells: SA-FITC, fluoresceinated streptavidin; TCR, T cell receptor * Corresponding author. Tel.: + l-5 18-474-6509: 3742007; e-mail: [email protected].

fax: + l-5 18.

The isolation of a phenotypically unique subpopulation of cells from a heterogenous cell population based on the differential expression of ceil surface markers is an essential step in many medical and biological research investigations. Purification of a desired cell type directly from a tissue source offers

0022-l 759/98/$19.00 0 1998 Elsevier Science B.V. All rights reserved. PII soo221759(98)00009-x

70

R.J. Colinas. A.C. Walsh/Journal

of lmmunologicul

the basic researcher alternatives to using cell lines for experiments requiring homogeneous populations of cells. In addition, several protocols for the treatment of hereditary or acquired disease that depend on the efficient selection of subpopulations of cells either are currently being employed or are in clinical trials. For example, in bone marrow (BM) or peripheral blood stem cell transplantation, hematopoietic stem/progenitor cells (HPC) are being purified from harvested BM or leukophoresis products to eliminate contaminating tumor cells or potentially alloreactive T cells (reviewed in Gee, 1994; Berenson et al., 1996). Also, tumor cells are being isolated from biopsy specimens to assess minimal residual disease (Naume et al., 1997) and stem cell gene therapy protocols are under development that rely on HPC selection (Kohn, 1995). Thus, it is clear that cell separation technologies are important in basic biological and medical research and in clinical medicine and the importance of this technology is likely to grow in the future. At present, the available cell separation media employ monoclonal antibodies (Mab) that bind to specific cell surface antigens and are linked to solid

Methods 212 (IY98169-78

(Lebkowski et al., 1992; Bensinger et al., 1990) or magnetic (Egeland et al., 1993; Miltenyi et al., 1990) matrices through covalent means or via biotin/avidin or immunoglobulin/anti-immunoglobulin interactions. The cells are either used without detachment from the matrix or are released by mechanical agitation, enzymatic digestion or binding of an anti-mouse Fab antibody (Shpall et al., 1994). As an alternative to these methodologies, we have developed a cell selection system based on the reversible interaction between calmodulin (CM) and a CM-binding peptide (CBP). CM is an ubiquitous 17 kDa cytosolic calcium (Ca2’ )-binding protein. Each CM molecule binds four Ca2+ ions with association constants (K) of 10-6-10-7 M (Linse et al., 1991). When the Ca*+-binding sites of CM are occupied, CM binds to a specific peptide sequence motif termed the CBP present in a number of cellular proteins through high affinity (K = 1O-8-1O-9 Ml interactions (Klee and Vanaman, 1982; O’Neil and DeGrado, 1990). However, when Ca’+ is removed or chelated away from CM, the affinity of CM for the CBP is greatly reduced and the CBP is released by CM. A schematic diagram of the CM/CBP-based

_A Target Cell

I

2’ mnibodycBP

4

f

calcillm

Binding

_ Release

Chelation -

StreptavidinSolid support

/ Fig. 1. Schematic

diagram of the prototype

CM/CBP

cell separation

Solid support

\

system. See the text for details

R.J. Colinas. A.C. Walsh/Journal

of Immunological

cell selection system is shown in Fig. 1. In the prototype of this system, target cells, labeled with a primary antibody specific to a cell surface antigen, are recognized by a secondary goat anti-mouse IgG (GAM-IgGl-CBP conjugate. In the presence of Ca’+, the GAM-IgG-CBP is recognized by biotinylated CM (CM-Biotin) and a streptavidin (SA)-solid support. Removal of Ca’+ by a chelating agent such as EGTA reduces the affinity of CM for the CBP and results in the release of the bound target cell. Furthermore, because the CBP-binding ability of CM can be regenerated by removal of the chelator and cell separare-addition of Ca”, a CM-CBP-based tion system utilizing solid-phase CM has the potential for reusability. Furthermore, the CM/CBP-based cell separation system can be used in conjunction with other commercially available cell separation technologies, making it possible to purify a subpopulation of cells by sequential positive selection for multiple cell surface markers.

2. Materials and methods

2.1. Reagents Goat anti-mouse IgG (GAM-IgG) was from Zymed Laboratories (So. San Francisco, CA) and unlabeled anti-HLA-DR (clone TU36, IgG,,), antihuman VP5 T cell receptor (TCR) (clone MH3-2, IgG,,), anti-human CD3-cychrome (clone HIT3a, IgG,,) and isotype controls were obtained from Pharmingen (San Diego, CA). Fluoresceinated (FITC) anti-HLA-DR (clone H279, IgG,,) was obtained from Coulter (Hialeah, FL). Anti-human CD3 (clone OKT3, IgG,,) used for T cell activation was from American Type Culture Collection (ATCC) (Rockville, MD); Dulbecco’s phosphate buffered saline (PBS) was from Biowhittiker (Walkersville, MD); bovine brain CM, CM-FITC, human gamma globulins, Cohn fraction II (HuIgG), CaCl?, propidium iodide and 2-(4-hydroxyazobenzene) benzoic acid (HABA) were from Sigma (St. Louis, MO). NaN, was from Fisher (Springfield, NJ); Sephadex G-25 was from Pharmacia (Piscataway, NJ) and sulfosuccinimidyl-4-( N-maleimidomethyl)cyclohexane-1 -carboxylate (S-SMCC), sulfosuccinimidyl-6(biotinamido)hexanoate (S-NHS-LC-Biotin) and

Methods 212 (19981 69-78

71

Coomassie Plus Protein Assay Reagent were from Pierce (Rockford, IL). The CBP was synthesized by the Wadsworth Center peptide synthesis core facility and was based on the sequence of the rabbit skeletal muscle myosin light chain kinase CM-binding domain (Takio et al., 1986) to which an amino-terminal cysteine residue was added. M-280 Streptavidin (SA) Dynabeads were obtained from Dynal (Oslo, Norway); MACS SA microbeads were from Miltenyi Biotec (Auburn, CA); SA-FITC was obtained from TAG0 (Burlingame, CA); EGTA was from Fluka Chem. (Ronkonkoma, NJ); Ficoll-Hypaque was from Pharmacia (Uppsala, Sweden) and Immulan goat anti-human IgG polystyrene beads were from Biotecx Laboratories (Houston, TX). 2.2. Cells The human acute myelogenous leukemia cell line KG-l (Koeffler and Golde, 1978) was obtained from ATCC. KG-l cells were passed weekly in Iscove’s modified Dulbecco’s medium containing 4 mM Lglutamine, 20 pg/ml gentamicin-SO, (complete IMDM) and 20% fetal bovine serum (FBS) (Hyclone, Logan UT) and incubated at 37°C in 7% CO,. KG-l cells in log-phase growth were used in all experiments. Human peripheral blood was obtained from healthy volunteers following informed consent according to an Institutional Review Board-approved protocol. Peripheral blood mononuclear cells (PBMC) were prepared by density gradient centrifugation over Ficoll-Hypaque cushions. Fractions of PBMC enriched or depleted of T cells were prepared using Immulan columns of goat anti-human IgG polystyrene beads using the procedure provided by the manufacturer (T cell-enriched fraction 82% CD3+, < 2% CD14+ or CD19+). For use as accessory cells in T cell activation, the T cell-depleted PBMC fraction was recovered and y-irradiated with 3000 cGy at 4°C using a ‘37Cs source (Isomedix, Parsippany, NJ) at a dose rate of 300 cGy/min and a cell density of 5 X lo6 cells/ml. Following irradiation, the cells were diluted to 3 X 105/ml in RPM1 1640 containing 1 X MEM nonessential amino acids, 1 X pyruvate, 0.9% NaHCO,, 2.4 mM L-gln, 10 pg/ml gentamicin-SO, and 10% FBS (complete RPM1 1640/10% FBS). In some experiments, 2 X lo4 irradiated accessory cells, in 0.5 ml complete

12

R.J. Colinas. A.C. Walsh/Journal

of Immunological

RPM1 1640/10% FBS, were added to wells of 24 well plates previously coated with 100 ,ug/ml antiCD3 (OKT3) in PBS for 2 h at 22°C and washed 3 X with PBS. Cells were incubated overnight at 37°C in 7% CO, before use. Aliquots of PBMC, from each fractionation step, were added to OKT3coated wells _t irradiated accessory cells and cultured for 4 days at 37°C 7% CO, and then 1 additional day in uncoated wells to allow TCR expression to recover as described (Kappler et al., 1989). 2.3. Chemical syntheses To maximize versatility of CM/CBP-based cell selection reagents, the CBP was cross-linked to a secondary anti-immunoglobulin reagent using the heterobifunctional crosslinker S-SMCC. Briefly, 90 pg of freshly prepared S-SMCC (18 ~1 of a 5 mg/ml stock in PBS) was added to 830 ~1 0.75 mg/ml GAM-IgG (5O:l molar ratio) PBS, pH 7.4 and rocked for 1 h at 22°C. Uncoupled S-SMCC was removed by passage over a 0.5 X 25 cm G-25 column equilibrated with PBS, pH 6.8 at room temperature and the GAM-IgG-S-SMCC conjugate-containing fractions were detected using the Coomassie Plus Protein Assay Reagent and pooled. Then, 33 ~1 of 25 mg/ml CBP in PBS, pH 6.8, was added to the pooled GAM-IgG-S-SMCC fractions at a molar ratio of 65 CBP to I GAM-IgG-S-SMCC and rotated overnight at 4°C. Excess CBP was removed by dialysis against PBS/O.O2% NaN,. CM was modified by biotinylation (CM-Biotin) to make it compatible with existing SA-based target cell separation media. CM-Biotin was synthesized by adding 29 ~1 of 5 mg/ml freshly prepared S-NHS-LC-Biotin to 860 pug CM in 400 ~1 ice cold 50 mM HEPES/l mM CaCl,/0.25 M NaCl, pH 7.5 and rotating for 2 h at 4°C. Uncoupled S-NHS-LC-Biotin was removed by passage of the reaction mixture over a 0.5 X 25 cm G-25 column equilibrated with D-PBS/l mM CaCl,/0.02% NaN, at 4°C. CM-Biotin-containing fractions were pooled and the CM to biotin molar ratio was calculated to be 1.6 by the spectrophotometric determination of the protein concentration and the displacement of HABA from avidin as described (Green, 1965). CM-Biotin:SA-FITC was made by combining CM-Biotin and SA-FITC at a molar ratio

Methods 212 (1998) 69-78

of 4 CM-Biotin to 1 SA-FITC for more than 1 h before use.

and incubating

on ice

2.4. Flow cytometry Unless otherwise indicated, all labeling procedures were conducted at 4°C and incubations were for 30 min in PBS/OS% heat inactivated HuIgG/0.02% NaN,/ 1 mM CaCl, (Ca’+-buffer) and the cells were washed twice with the same buffer or with PBS/O.O2%NaN,/l mM EGTA (EGTA-buffer) and pelleted by centrifugation for 7.5 min at 400 X g. Staining of the cells was performed with 1 ,ug of each primary antibody/lo6 cells. GAM-IgG-CBP was titrated to determine the optimal concentration for labeling of target cells using unlabeled or anti-HLA-DR-labeled KG-l cells. Following GAM-IgG-CBP labeling of the KG-l cells, a molar excess of CM-FITC was added and the samples were analyzed by flow cytometry. The GAMIgG-CBP concentration of 1 pg/lO” cells gave the greatest difference in mean fluorescence intensity (MFI) between unlabeled and anti-HLA-DR-labeled KG-l cells and was used in all subsequent experiments. CM and CM-FITC were used at 0.2 pg/lOh cells and CM-Biotin:SA-FITC was used at 1 pg/ lo6 cells. These concentrations were in molar excess relative to GAM-IgG-CBP. To demonstrate reversibility of CM-Biotin:SA-FITC binding, some samples were split and washed in either Ca’+- or EGTA-buffer. Because the GAM-IgG-CBP is bivalent and any unoccupied binding sites on GAM-IgGCBP on labeled cells can potentially bind a second FITC-conjugated anti-HLA-DR antibody, the cells were restained with CM-FITC instead. Alternatively, as with the V/35 selected cells described below, the cells could also be cultured to remove the bound antibodies before analysis. It would also be possible to use a Mab of an IgM isotype if available. To restain fractions of selected cells, cells were washed twice with EGTA-buffer, once with Ca’+-buffer and restained with CM-FITC or CM-biotin:SA-FITC. Alternatively, PBMC were cultured for 4 days with solid-phase anti-CD3 and irradiated accessory cells, transferred to uncoated wells for an additional 18 h and then stained with anti-V/35-FITC and anti-CD3 cychrome. Samples were stained with propidium iodide to assess viability and analyzed using a FAC-

R.J. Colinas. A.C. Walsh/Journal

SCAN flow cytometer CA).

(Becton Dickinson,

of Immunological

San Jose,

Methods 212

i 1998169-78

73

mined using a Coulter ZM cell counter to assess cell recovery and yield.

2.5. Cell selection

3. Results

KG-l cells were labeled with anti-HLA-DR while PBMC were labeled with anti-V/35 followed by GAM-IgG-CBP and CM-Biotin as described in Section 2.4. Using the bead:target cell ratio recommended by the manufacturers, cells were either captured with SA-coated magnetic Dynabeads beads (20 beads/cell, previously washed 3 X with PBS/l mM CaC12/0.02% NaN, and resuspended in 75 ~1 Ca’+-buffer) or MACS SA microbeads (IO ~1 suspension/lo’ cells in 90 ~1 Ca”-buffer) with very similar results. MACS SA microbeads were chosen for the VP5 TCR T cell selection for reasons of economy. Cell selection with Dynabeads was performed as follows: After the final wash, the labeled cells were resuspended in 125 ~1 Ca*+-buffer and 8.2 X 10’ cells were gently mixed with the washed SA Dynabeads. The bead/cell suspension was rotated axially at approximately 25 rpm for 1 h at 4°C. The target cells captured by the beads were separated from the unbound cells using a magnet. Unattached cells trapped in the magnetic beads were removed by washing twice with 200 ~1 Ca’+-buffer and combined with the target-depleted fraction. The cells bound to the beads were released by gentle agitation with four 200 ~1 aliquots of EGTA-buffer. Prior to labeling with the MACS SA microbeads, cells were passed over a miniMACS column to remove adherent cells and cell aggregates, Cells were labeled with MACS SA microbeads, washed, passed over a second miniMACS column twice and washed with 3 X 500 ~1 Ca’+-buffer. Column-bound target cells were released and collected in a separate tube containing 0.5 ml PBS/O.5% human IgG/0.02% NaN, as follows. The column was rinsed with 3 X 250 /.~l EGTA-buffer. incubated with the EGTA-buffer for 5 min at room temperature, then 1 ml of EGTA-buffer was pushed through with a miniMACS column piston at approximately 12 ml/min. Aliquots of preselection and target-depleted or -enriched cells were restained for flow cytometry as described above to assess the purity of the separated cells. In addition, the cell numbers in the cell fractions were deter-

3.1. Specificity

and rellersibility of CM-Biotin GAM- IgG-CBP cell selection reagents

and

KG-l cells were used to test the specificity of the cell selection reagents. The KG-l AML cell line, in log-phase growth, is heterogeneous with respect to cell surface expression of the major histocompatibility complex HLA-DR molecule. Direct staining of KG-l cells with anti-HLA-DR-FITC showed that 80% of the cells were HLA-DR+ relative to the FITC-conjugated isotype control antibody (Fig. 2A). Similar percentages of HLA-DR+ cells we,re detected by indirect staining in Ca’+-buffer with unconjugated anti-HLA-DR followed by GAM-IgGCBP and either CM-Biotin:SA-FITC or CM-FITC (Fig. 2C and D, 77% and 78%. respectively). The differences between the MFIs correlate with the differences in FITC/protein (F/P) ratios of HLA-DRFITC, SA-FITC and CM-FITC (F/P = 10, 5.5 and 0.8, respectively). The specificity of the GAM-IgGCBP and CM-Biotin:SA-FITC reagents is shown in Fig. 2. In the absence of anti-HLA-DR, the MFI of GAM-IgG-CBP/CM-Biotin:SA-FITC stained cells were reduced by an order of magnitude (Fig. 2C, grey histogram). Similar reductions in MFIs were observed if GAM-IgG-CBP was replaced with GAM-IgG (Fig. 2B, black histogram) or CM-Biotin was replaced with CM (Fig. 2B, grey histogram). Human PBMC also exhibited low background binding of CM-Biotin or GAM-IgG-CBP (data not shown). The reversibility of CM binding to GAMIgG-CBP when Ca’+ was chelated was demonstrated using a sample of KG-l cells stained with anti-HLA-DR, GAM-IgG-CBP and CM-FITC where half of the cells were washed in Ca”-buffer and the other half in EGTA-buffer. The specific HLA-DR staining observed in the presence of Ca’+ (Fig. 2D, black histogram) was reduced to background levels following the EGTA wash (Fig. 2D. grey histogram). Washing with EGTA-buffer did not reduce cell viability and had a comparable reversing effect on CM-Biotin:SA-FITC staining (data not shown).

R.J. Colinas, A.C. Walsh/ Journal of Immunological Methods 212 (1998) 69-78

h

E

‘.

100

IO’

-. 10’

IO”

104

P

Fluorescence intensity

Fig. 2. Specificity and reversibility of CM/CBP cell selection reagents and selection of HLA-DR+ KG-l cells. KG-I cells were stained. analyzed or selected as described in Section 2. Specific HLA-DR staining is shown as follows: anti-HLA-DR-FITC (A, black histogram). anti-HLA-DR, GAM-IgG-CBP and CM-Biotin:SA-FITC (C, black histogram) or anti-HLA-DR, GAM-IgG-CBP and CM-FITC (D, black histogram). Reagent specificity controls were mouse IgG,,-FITC (A, grey histogram), anti-HLA-DR. GAM-IgG-CBP, unbiotinylated CM and SA-FITC (B, grey histogram), GAM-IgG-CBP and CM-Biotin:SA-FITC without anti-HLA-DR (B, black histogram) and anti-HLA-DR. unconjugated GAM-IgG and CM-Biotin:SA-FITC (C. grey histogram). The grey histogram in (D) was the result of washing cells stained with anti-HLA-DR, GAM-IgG-CBP and CM-Biotin:SA-FITC with EGTA-buffer.

3.2. Selection of HLA-DR+

KG-I cells

The feasibility of using the CM/CBP cell separation system to fractionate KG-l cells into HLA-DRdepleted and -enriched subpopulations was tested. KG- 1 cells were labeled with anti-HLA-DR followed by GAM-IgG-CBP, CM-Biotin and SA-magnetic beads. The target cells were immobilized on a magnet and nontarget cells were removed by washing in the presence of Ca 2+. Without removing the column containing immobilized cells from the magnetic field, the target cells were released by chelating Ca’+ with EGTA-buffer. Analysis of restained cells by flow cytometry showed that the CM/CBP selection reagents and EGTA were not cytotoxic and the HLA-DR-enriched fractions averaged a purity of 90% HLA-DR+ while the HLA-DR-depleted fraction was reduced to 13% HLA-DR+ (Table 1). Using the absolute numbers of HLA-DR’ cells in the HLA-

DR-enriched and Pre-selection fractions, a 75% yield was calculated. To ensure that the target cells were

Table 1 Section of HLA-DR+ KG-1 cells using calmodulin/calmodulinbinding peptide affinity reagenta Cell fraction

Viability

Percentage

Pre-selection HLA-DR-depleted

94 (1) 95 (2)

76 (1) 13 (8)

HLA-DR+

N.A.’ N.A.

Yieldb

HLA-DR-enriched

96 (5)

90 (3)

75 (2)

aCells were labeled selected, re-labeled and analysed as described in the text. The data shown represent the means and standard deviations (in parentheses) from the results of two independent experiments. ‘Yield = (number of anti HLA-DRGAM-IgG-CBPCM-FITC positive cells in the HLA-DR-enriched fraction)/{number of antiHLA-DR:GAM-IgG-CBPCM-FITC positive cells in the Preselection fraction) X 100. ‘N.A.-not applicable.

R.J. Colinas, A.C. Walsh/Journal

of Immunological Methods 212 (1998) 69-78

actually dissociated from the MACS magnetic particles, an aliquot of target cells was reapplied to the magnet. Less than 10% of the released cells were capable of re-binding to the magnet following EGTA treatment (data not shown). Reversibility of the CM/CBP interaction was also demonstrated. HLADR+ cells in the Pre-selection (Fig. 2D, black his-

75

togram) and HLA-DR-enriched and -depleted (Fig. 2E, black and grey histograms, respectively) fractions could be restained with CM-Biotin:SA-FITC following EGTA wash and re-equilibration with Ca”. These results showed that the reversible CM/CBP interaction could be used to gently and efficiently separate cells.

A

Fig. 3. Flow cytometric analysis of human T cell fractions before and after labeling with the CM/CBP cell separation reagents or VPS-targeted fractionation. Unlabeled cells of the T cell-enriched fraction were analyzed for V/35 and CD3 expression prior to (A) or following (B) 4 days of culture with solid phase anti-CD3 and irradiated accessory cells. Cells, labeled with the CM/CBP selection reagents before fractionation (C) or from the VPS-depleted (D) or -enriched (E) fractions were all analyzed post-culture with anti-CD3 and irradiated accessory cells.

76

R.J. Colinas. A.C. Walsh/

3.3. Rare cell separation selection system

Journal of Immunological

using the CM/CBP

cell

Methods 212 (19981 69-78

Table 3 Effect of fractionation

on T cell blast transformationa

Percentage

The capabilities of the CM/CBP cell selection system were more rigorously tested by targeting a relatively rare subpopulation of cells from normal human PBMC for isolation. T cells expressing TCR p subunits derived from the V/35S2 or VpSS3 gene segments represent two of the approximately 52 possible VP gene segments (Akolkar et al., 1997) and were chosen as a test subpopulation for purification. These cells, which represent approximately 3-5% of the lymphocytes in PBMC (Kappler et al., 1989) are recognized by the V/35S2/S3-specific monoclonal antibody, MH3-2 (Posnett et al., 1996; Peyrat et al., 1996). Nonspecifically adherent cells were removed by first enriching for T cells and then passing the T cell-enriched fraction, labeled with anti-Vfi5, GAM-IgG-CBP and CM-Biotin, over a magnetic column in Ca’+-buffer. The cells were then labeled with MACS SA microbeads and loaded onto a second magnetic column and washed with Ca’+buffer. Selected cells bound to the column were eluted in EGTA-buffer. It has been previously reported that the relative proportion of Vp5’ T cells present in freshly isolated PBMC were unaffected by culturing (Kappler et al., 1989); our results shown in Fig. 3A and B and Table 2 confirmed this observation. In addition, labeling cells with the cell separation reagents prior to 4 days of culture with solid-

Table 2 Effects of labeling with selection reagents,

fractionation

and culturing

Pre-culture Post-culture

of lymphoblastsb

PBMC

Pre-selection

V/35-depleted

VPS-enriched

13 91

2 79

2 71

2 71

‘Cells were fractionated as described in the text. The percentage of lymphoblasts in each fraction was determined by flow cytometry before or after 4 days of culture with irradiated accessory cells and solid-phase antiCD3. bRepresentative data from one experiment are shown.

phase anti-CD3 + irradiated accessory cells did not significantly influence the percentages of CD3+ or Vp5’ cells (Fig. 3B and C and Table 2). Fig. 3 also illustrates the ability of this prototype CM/CBPbased cell separation system to yield a PBMC subpopulation highly enriched for Vfi5’ target cells. The performance of the CM/CBP-based cell selection system is also summarized in Table 2. VP5 selection had no significant effect on cell viability determined immediately following fractionation and, on average, Vp5’ cells were enriched by 20-fold to a purity of 72% with an 80% yield. Furthermore, in the presence of irradiated accessory cells and solidphase anti-CD3, unlabeled PBMC unfractionated cells labeled with the CM/CBP cell selection reagents and VPS-depleted or -enriched cells exhibited similar abilities to undergo blast transformation typical of T cell activation (Table 3). Taken together,

on CD3+

and Vfi5+

cell frequencies”

T cell-enriched Pre-culture

Viability CD3+’ !Jps+ Yieldd Enrichment

84 (7) 82 (4) 4.1 (1) N.A.e N.A.

unlabeled

Post-culture Unlabeled

Labeled

V/35-depleted

VPS-enriched

89 (3) 88 (8) 4.5 (1) N.A N.A.

95 (2) 90 (7) 3.9 (1) N.A. N.A.

93 (4) 88 (4) 2.3 (1) N.A. N.A.

89 (10) 96b 72 (12) 80 (7) 20 (3)

“Where indicated, cells were labeled with selection reagents, fractionated, cultured. stained with anti-VP5 and anti-CD3 and analyzed by flow cytometry as described in the text. bResult from one experiment only. All other results represent the means k (standard deviations) from three experiments. ‘Paired t-test analyses of CD3+ and Vp5’ cell frequencies between the pre- and post-culture or the unlabeled and labeled groups demonstrated no statistically significant differences (P > 0.06). dYield = (number of Vp5’ cells in the VPS-enriched fraction)/{number of Vp5’ cells in the Pm-selection fraction) X 100. “N.A.-not applicable.

R.J. Colinus, A.C. Walsh/Journal

ofImmunologiculMethods

these results show that the CM/CBP cell selection system can generate a functional high purity target cell population without significant cytotoxicity.

4. Discussion The need to gently and efficiently isolate specific cell types is gaining importance in basic research and clinical medicine. As an alternative to the cell separation technologies that are already available, we have developed a cell separation system based on the reversible interaction between CM and a CBP. Our results show that nonspecific binding of the CM and CBP affinity reagents to nontarget cells was low and did not exhibit significant cytotoxicity. Fractionation of cells based on cell surface expression of HLA-DR on a human AML cell line and VP5 TCR on normal human peripheral blood T cells using the CM/CBP cell selection reagents in combination with antigenspecific monoclonal antibodies and commercially available magnetic solid supports was performed. While the purities and yields of target cells were similar to those reported for other cell selection systems already available (Shpall et al., 19941, the primary advantage of the CM/CBP-based cell selection system is the reversibility of the CM/CBP interaction without the need for additional specialized reagents or potentially damaging target cells. Thus, by chelating Ca”+, cells are gently released without the need for mechanical agitation or enzymatic digestion used by other cell separation systems yielding both target cell-enriched and -depleted fractions. Our results also showed that neither the cell separation reagents nor the EGTA-buffer caused significant cytotoxicity. Moreover, preliminary analysis suggested that the VPS’ T cells retained their ability to be activated. However, additional studies are necessary to further evaluate the functionality of cells separated by this method. The CM/CBP cell separation system that we have developed utilizes an indirect method to fractionate cells, making it highly versatile as well as compatible with existing cell separation technologies. Because this prototype of the CM/CBP cell separation system utilizes an indirect labeling method, it is more labor-intensive than directly labeling the cells for isolation. Nevertheless, cells can be

212

(IYYR)69-78

77

isolated by this method in less than 4 h. It may be possible to reduce the time necessary to separate cells using the CM/CBP method by using a pre-assembled labeling complex consisting of an antigenspecific Mab. GAM-IgG-CBP, CM-Biotin and MACS SA microbeads. Other approaches to simplify cell separation and analysis might involve preparation of CBP-conjugated, fluorochrome-labeled primary antibodies to make labeling of cells less time consuming and allow assessment of target cell purity without additional washing and staining steps. Moreover, the CBP could be conjugated to ligands other than monoclonal antibodies, such as lectins, cytokines, peptides or drugs, that differentially recognize a cell surface determinant and define a specific subpopulation of cells. It also would be possible to incorporate a CM-binding domain into the coding regions for affinity reagents generated by phage display (Griffiths, 1993) or transgenic immunoglobulin (Lonberg and Huszar, 1995) technologies. Expression of a CBP domain in recombinant phage or transgenic immunoglobulins would be useful for both purification and their adaptation to cell separation. Additional improvements in the CM/CBP cell separation system would include direct coupling of CM to a solid matrix while retaining the ability to bind and release CBP-labeled cells. Our initial attempts to produce a solid phase CM capable of binding CBPlabeled cells have demonstrated a low target cell capacity. This is most likely due to adsorption of CM to the solid matrix which inhibits the conformational changes that have been shown to be necessary for CM to bind the CBP (Meador et al., 1992). It is also possible that the relatively small CM molecule may have entered the solid matrix making it inaccessible to the CBP on the labeled cells. Ideally, the solid matrix would exhibit low porosity and adsorptivity in combination with a chemical functionality to which CM could be covalently linked to the surface of the solid phase through a single amino acid residue. A solid matrix with these physical properties would allow attachment of CM molecules that have retained their Ca’+-dependent ability to bind CBP on labeled cells. As a result, the cell separation procedure would be less time consuming and the matrixbound CM would be reusable. The results presented above show that a CM/CBP-based cell selection system is fully capable of gentle, highly specific and

78

R.J. Colinas, A.C. Walsh/Journal

reversible selection of target cells another cell separation alternative.

and offers

of Immunological Methods 212 (1998) 69-78

yet

Acknowledgements The authors wish to thank Dr. David A. Lawrence for his advice and discussion of these results. This study was aided by the Wadsworth Center Molecular Immunology and Peptide Synthesis Core Facilities. Financial support for this project was provided by grants ES05020 and ES03778 from the NIH.

References Akolkar, P.N., Gulwani-Akolkar, B., Silver, J., 1997. Methods for analysis of the human T-cell receptor (TCR) repertoire in health and disease. In: Leffel, MS., Donnenberg, A.D., Rose, N.R. (Eds.), Handbook of Human Immunology. CRC Press, Boca Raton, FL, p. 567. Bensinger, W.I., Berenson, R.J., Andrews, R.G., Kalamasz, D.F., Hill, R.S., Bernstein, I.D., Lopez, J.G., Buckner, C.D., Thomas, E.D., 1990. Positive selection of hematopoietic progenitors from marrow and peripheral blood for transplantation. J. Clin. Apheresis 5, 74. Berenson, R.J., Shpall, E.J., Auditore-Hargreaves, K., Heimfeld, S., Jacobs, C., Krieger, M.S., 1996. Transplantation of CD34+ hematopoietic progenitor cells. Cancer Invest. 14, 589. Egeland, T., Tjonnfjord, G., Steen, R., Gaudemack, G., Thorsby. E., 1993. Positive selection of bone marrow-derived CD34 positive cells for possible stem cell transplantation. Transplant. Proc. 25, 1261. Gee, A.P., 1994. Immunologically based methods for the elimination of tumor cells from autologous stem cell grafts. lmmunomethods 5, 232. Green, N.M., 1965. A spectrophotometric assay for avidin and biotin based on binding of dyes by avidin. Biochem. J. 94, 23~. Griffiths, A.D., 1993. Production of human antibodies using bacteriophage. Curr. Opin. lmmunol. 5, 263. Kappler, J., Kotzin. B., Herron, L., Gelfand, E.W., Bigler, R.D.,

Boylston, A., Carrel, S., Posnett, D.N., Choi, Y., Marrack, P., 1989. V/3-specific stimulation of human T cells by staphylococcal toxins. Science 244. 811. Klee, C.B., Vanaman, T.C., 1982. Calmodulin. Adv. Protein Chem. 35, 213. Koeffler, H.P., Golde, D.W., 1978. Acute myelogenous leukemia: a human cell line responsive to colony-stimulating activity. Science 200. 1153. Kohn, D.B., 1995. The current status of gene therapy using hematopoietic stem cells. Curr. Opin. Pediatr. 7. 56. Lebkowski, J.S., Schain, L.S., Okrongly, D., Levinsky, R.. Hatvey, M., Okarma, T.B., 1992. Rapid isolation of human CD34 hematopoietic stem cells: Purging of human tumor cells. Transplantation 53, 1011. Linse, S., Helmersson, A., Forsen, S., 1991. Calcium binding to calmodulin and its globular domains. J. Biol. Chem. 266, 8050. Lonberg, N., Huszar, D., 1995. Human antibodies from transgenic mice. lnt. Rev. lmmunol. 13, 65. Meador, W.E., Means, A.R., Quiocho, F.A., 1992. Target enzyme recognition by calmodulin: 2.4. A structure of a calmodulinpeptide complex. Science 257, 125 1, Miltenyi, S., Muller, W., Weichel, W.. Radbruch, A., 1990. High gradient magnetic separation with MACS. Cytometry 11,23 1. Naume, B., Borgen, E.. Beiske, K.. Herstad. T.K.. Ravnas, G.. Renolen, A., Trachsel, S., Thrane-Steen, K., Funderud, S., Kvalheim, G., 1997. lmmunomagnetic selection for the enrichment and detection of isolated breast carcinoma cells in bone marrow and peripheral blood. J. Hematother. 6. 103. O’Neil, K.T., DeGrado, W.F., 1990. How calmodulin binds its targets: sequence independent recognition of amphiphilic (Yhelices. Trends Biochem. Sci. 15, 59. Peyrat, M.A., Gaschet. J., Vivien, R., Vie. H., Bonneville, M., 1996. Clustering of the TcR workshop mABs by FACS analysis of polyclonal T-cell lines. Immunologist 4, 9. Posnett, D.N., Romagne, F., Necker, A., Kotzin, B.L., Sekaly. R.P., 1996. First human TcR monoclonal antibody workshop. Immunologist 4, 5. Shpall, E.J., Gehling, U., Cagnoni, P., Purdy, M., Hami. L., Gee, A.P., 1994. Isolation of CD34-positive hematopoietic progenitor cells. lmmunomethods 5, 197. Takio. K.. Blumenthal, D.K.. Walsh, K.A., Titani, K., Krebs, E.G., 1986. Amino acid sequence of rabbit skeletal muscle myosin light chain kinase. Biochemistry 25. 8049.