www.elsevier.com/locate/jim

Journal of Immunological Met

Research paper

ACSD labelling and magnetic cell separation: a rapid method of

separating antibody secreting cells from non-secreting cells

Silvia Carroll, Mohamed Al-Rubeai*

Department of Chemical Engineering, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK

Received 17 August 2004; received in revised form 13 October 2004; accepted 11 November 2004

Available online 8 December 2004

Abstract

Several new ways of selecting cells have recently been developed. These include magnetic separation of cells by labelling

with magnetic beads against the recombinant product, gel microdrop technology which encapsulates the cells in gelatine beads

and matrix-based secretion assays. Affinity capture surface display (ACSD) is a matrix-based assay for the enrichment of high

producing cells and relies on the strong affinity between biotin and avidin derivatives. Matrix-based assays have previously

only been used for the enrichment of recombinant cells. Here, we have optimised this assay and developed a method of

separating antibody producing cells from non-producing cells in a recombinant myeloma cell line using ACSD combined with

MACS magnetic separation. The method is rapid, simple enough to become routine and adaptable to many different secreted

products from recombinant mammalian cells.

D 2004 Elsevier B.V. All rights reserved.

Keywords: Affinity capture surface display (ACSD); Secretion assay; Magnetic cell separation; Flow cytometry; NS0; Cell culture

1. Introduction

A fast and effective way to select for high

producing cell lines is a major goal in industry.

Traditionally, the selection of high producing cell

lines is time and labour consuming and only a few

clones with the ability to produce large amounts of

recombinant protein are selected for expansion. Since

0022-1759/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.jim.2004.11.007

* Corresponding author. Tel.: +44 121 414 3888; fax: +44 121

414 3888.

E-mail address: m.al-rubeai@bham.ac.uk (M. Al-Rubeai).

recombinant protein production stability is an impor-

tant issue several rounds of screening for high

producers have to be performed. Moreover, many of

the high producing cells are frequently overgrown by

the faster growing low or non-producing cells. Thus, it

is important to select for stable high producers

accurately, efficiently and at an early stage of the cell

line development process.

Flow cytometry analysis allows a fast qualitative

and quantitative analysis of cell surface expression of

the recombinant protein, rendering the method useful

for screening high producing cells. However, the only

commonly used application of this technology has been

hods 296 (2005) 171–178

S. Carroll, M. Al-Rubeai / Journal of Immunological Methods 296 (2005) 171–178172

the selection of hybridoma cells for high IgG produc-

tivity due to the correlation found in this cell line

between secretion rate and surface associated protein

(McKinney et al., 1991; Chuck and Palsson, 1992).

For other cell lines several methods based on the

affinity of biotin and avidin have been described in

which it is possible to capture the secreted protein

on the cell surface or in the vicinity of the secreting

cell (Manz et al., 1995; Holmes and Al-Rubeai,

1999; Borth et al., 2001; for a review, see Carroll

and Al-Rubeai, 2004). Biotin/avidin matrix systems

have been described for the enrichment of popula-

tions of cells with relatively high producers and the

removal of relatively low producing cells. However,

there is no literature available in which these

techniques have been used for the separation of

producers from non-producers.

Miltenyi et al. (1990) developed a method for

separating rare cells using a simultaneous combined

magnetic and fluorescent tagging of the staining

antibody, which recognized one characteristic antigen

of the rare cell. Thus, the rare cells could be physically

isolated (by magnetic columns), optimally identified

by the fluorescent dye and analysed for any other

parameter, either by immunofluorescence or by flow

cytometry. In this approach, the cells can be labelled

with the magnetic antibody either directly or indi-

rectly, depending on the available magnetic antibodies

and the separation can be either positive or negative.

Positive selection magnetically labels the wanted

target cells which are retained in the magnetic column,

while negative selection labels the unwanted cells and

thus the positive cells pass through the magnetic

column. Magnetic cell separation has been widely

used to isolate and identify subsets of T cells and B

cells in many immunological studies. Waldrop et al.

(1997) compared the frequency of cytomegalovirus-

specific CD4+ memory/effector T cells and, in a later

study, their costimulatory requirements (Waldrop et

al., 1988). Brosterhus et al. (1999) used this technique

to isolate antigen specific CD4+ and CD8+ T

lymphocytes, while Ouyang et al. (2000) used the

technique to find the source of Th2 inducement.

Here, we describe a novel way to separate

producers from non-producers from a mixed popula-

tion of NS0 cells by combining affinity capture

surface display (ACSD) labelling with magnetic cell

separation.

2. Materials and methods

2.1. Cell lines

NS0 6A1, a myeloma cell line expressing a

chimeric antibody (b72.3) against breast tumour

antigen TAG73, was kindly provided by Lonza

Biologics (Slough, UK). The NS0 6A1 cell line

utilises the powerful GS promoter system as described

by Bebbington et al. (1992). The cells were subcul-

tured routinely in T-flasks and maintained at 37 8C in

a humidified atmosphere of 5% CO2 in air. The NS0

6A1 cell line was grown in Dulbecco’s modified

Eagle’s nutrient mixture F-12 Ham without glutamine

supplemented with 10% foetal calf serum (FCS; PAA

Laboratories, Yeovil, UK), 10 units ml�1 penicillin,

10 Ag ml�1 streptomycin and 10 AM MSX (Sigma).

The NS0 wild-type (WT) cell line was grown in

Dulbecco’s modified Eagle’s medium supplemented

with 10% FCS (PAA Laboratories), 10 units ml�1

penicillin and 10 Ag ml�1 streptomycin.

2.2. Cell surface biotinylation studies

NS0 6A1 cells were washed in PBS followed by

washing by PBS adjusted to pH 8 with 1 M NaOH.

The cells were biotinylated by incubation in 500 Al offilter sterilised NHS-LC Biotin (at different concen-

trations in PBS pH 8; Pierce) at room temperature for

30 min. Following incubation, the cells were washed

twice in PBS (pH 7) and resuspended in 1 ml PBS

with 10 Al streptavidin FITC for 10 min in the dark.

The cells were washed twice in PBS (pH 7) and

resuspended in 2 ml PBS with 10 Ag propidium iodide

(PI) to distinguish between viable and non-viable

cells. The control used for this experiment was

unbiotinylated cells stained with streptavidin FITC.

2.3. ACSD labelling

The steps involved during ACSD labelling were

optimised to improve the labelling of producer cells.

A minimum of 1�107 cells was used for the assay.

The cells were washed in PBS, followed by washing

with PBS adjusted to pH 8 with 1 M NaOH; this

removed any primary amines present on the cell

surface. The cells were then biotinylated by incuba-

tion in 1 ml of filter sterilised NHS-LC Biotin (0.05

S. Carroll, M. Al-Rubeai / Journal of Immunological Methods 296 (2005) 171–178 173

mg ml�1 in PBS pH 8; Pierce) at room temperature

for 20 min. Following incubation, the cells were

washed twice in PBS (pH 7) and resuspended in 1 ml

of complete medium containing gelatine (10% w/v;

Sigma), 26 Al biotinylated rabbit antihuman IgG, Fc

specific (Pierce) and 5 Al neutravidin (1 mg ml�1;

Pierce). The cells were allowed to secrete product for

20 min at 37 8C, after which the gelatine was

removed by washing twice in PBS. The bound

product was detected by incubating the cells in ice

for 15 min in the dark with 1 ml of complete media

together with 20 Al FITC conjugated goat anti-human

kappa light chain antibody (Sigma). After washing

twice in PBS, the cells were suspended in complete

medium. Two controls were introduced in the assay:

the NS0 WT (non-secretors) stained with ACSD

labelling and NS0 6A1 stained with ACSD labelling

without the addition of the biotinylated antibody

(negative control).

2.4. Magnetic cell separation

Fig. 1 depicts the combination of ACSD sorting

and magnetic cell separation using the Mini Macs

(kindly provided by Miltenyi Biotech). Once the cells

were labelled using ACSD, the cells were resus-

pended in 90 Al buffer (PBS supplemented with 0.5%

bovine serum albumin and 2 mM EDTA). Following

this, 10 Al of anti-FITC MicroBeads (Miltenyi

Fig. 1. ACSD labelling combined with magnetic cell separation. The cell su

The secreted antibody is captured by biotinylated antibody specific for th

secreted antibody is added and an anti-FITC microbead is also added. The

cells (positive population) will be captured in the column while the non-s

Biotech) per 107 cells were added, mixed well and

incubated for 15 min at 6 8C. The cells were washed

twice and resuspended in 500 Al buffer. Prior to the

separation, the magnetic columns were washed twice

with buffer and placed in the magnetic field of the

separator. Once the suspension had passed through the

column, the magnetic column was rinsed using buffer.

The negative cells were collected and later analysed

using flow cytometry. Following this, the column was

removed from the magnetic field and the positive cells

were flushed out and collected for analysis.

2.5. Flow cytometer and analysis software

Flow cytometric analysis was performed using a

Coulter EPICS Elite Analyser (Coulter Electronics,

Luton, UK) equipped with an argon laser emitting 15

mWat 488 nm. Cells were passed through the focus of

the beam and caused fluorescent emissions. Forward

scattered light (108) was collected by the forward

angle light scatter (FLS) photometer and the 908emissions were routed to photo multiplier tubes

(PMT). The laser light was blocked by a 488-nm

dichroic long-pass filter. FITC emissions (550 nm)

were collected at PMT2 (using a 550-nm dichroic

filter) and PI emissions (550–700 nm) were collected

at PMT4 (635-nm dichroic filter). The data collected

were analyzed using FCS Express 2 (De Novo

Software).

rface of the cells is biotinylated and the cells are allowed to secrete.

e secreted antibody. FITC conjugated antibody also specific for the

cells are then passed through a magnetic column and only secreting

ecreted cells are flushed out (negative population).

S. Carroll, M. Al-Rubeai / Journal of Immunological Methods 296 (2005) 171–178174

2.6. Product analysis

Nunc Maxisrop 96-well plates were coated with

mouse anti-human IgG [Fc Specific (Sigma)] diluted

1/1000 in 0.05 M carbonate buffer pH 9.6 and

incubated overnight at 4 8C. After washing four times

with PBS-Tween (0.05%), unoccupied sites were

blocked with skimmed milk (1% w/v in PBS) for 1

h. The plate was washed four times, and the standards

and supernatant from the cultures were serially diluted

using skimmed milk (1% w/v in PBS) and added to

the wells. Following a 1-h incubation, the plate was

washed four times and 1/20,000 diluted peroxidase

conjugated antihuman kappa light chain (Sigma) was

added to the wells for 1 h. The plate was developed by

adding 100 Al o-phenylenediamine dihydrochloride

[0.4 g l�1 in 0.1 M di-sodium phosphate, 0.004% (v/

v) hydrogen peroxide] per well as a chromogen. The

reaction was stopped after 20 min using 100 Al 2.5 M

sulphuric acid and absorbance values at 495 nm were

determined using a SLT Spectra ELISA plate reader.

Product concentration was determined by linear

regression against external standards.

3. Results and discussion

This study was initiated to separate by ACSD in

combination with magnetic cell separation two pop-

ulations of NS0 cells: recombinant antibody produc-

ing cells (NS0 6A1) and non-producing cells (NS0).

This approach was intended to simulate the hetero-

geneity of cell populations after transfection.

3.1. Cell surface biotinylation of NS0 6A1

When the NS0 6A1 cells were stained using ACSD

labelling, there was a difference between the

unstained and the stained populations of cells, with

an area of overlap between the two populations. Since

surface biotinylation is dependent on the amount of

amines present on the cell surface, and on the viability

of the cells and their growth phase, biotinylation

studies were carried out to determine the optimum

biotin concentration for the NS0 6A1 cell line. Fig. 2

shows the fluorescence intensity after cell surface

biotinylation with different concentrations of biotin of

NS0 6A1 cells. The histograms show that 0.05 mg

ml�1 was the optimum concentration for cell surface

bioitnylation, and that using cells from later stages of

the growth phase would not produce adequate

fluorescence intensity. On day 4 of the growth phase,

there was a bimodial distribution of fluorescence

intensity indicating the presence of two populations of

biotinylated and unbiotinylated cells. During this

experiment, the viability of the cells remained con-

stant throughout the different treatments with different

concentrations of biotin, while specific antibody

secretion rates were found to range between 30 and

45 pg cell�1 day�1 with a maximum value achieved

on day 3 of the growth phase.

The assay used for measuring biotinylation

assumes that one mole of streptavidin FITC molecule

is able to bind to four moles of biotin (Chaiet and

Wolf, 1964) or that one unit will bind to 1.0 Ag biotin

(Green et al., 1971). Since an excess of streptavidin

FITC was added to the cells after biotinylation, it was

assumed that the assay, which monitors cell surface

biotinylation using a biotin concentration of 50 Agml�1, was accurate.

Biotinylation is clearly the most important step in

building the affinity matrices. As mentioned previ-

ously cell surface biotinylation is dependent on the

amount of free amines on the cell surface, but this

reaction differs between cell lines. From our studies

we found that the optimum biotinylation concentra-

tion for the NS0 6A1 cells, was much lower than the

reported values for hybridomas and spleen cells

(Manz et al., 1995), peripheral blood mononuclear

cells (Schuberth et al., 1996) and the CHO DHFR cell

line (Borth et al., 2001). However, besides the amount

of amines present on the cell surface, the growth

conditions and the kinetics of protein secretion during

the labelling procedure are other important factors that

may affect the efficiency of the reaction.

Another concern with cell surface biotinylation is

how long the biotin molecules remain on the cell

surface before a turnover of surface proteins occurs.

Biblia and Flickenger (1991) suggest that, in the case

of hybridoma cells, most antibody is secreted into the

extracellular medium between 30 and 180 min. If the

turnover of proteins is less than 1 h (the duration of

ACSD labelling), there is a possibility that the matrix

built around the cell is lost in the washes during the

labelling procedure. NHS-esters target primary amines

and the biotinylation reagent is not membrane

Fig. 2. Flow cytometric analysis of the NS0 6A1 cell line with different biotin concentrations during exponential growth phase. The cells were

botinylated using different concentrations of biotin during their exponential phase (days 1–4).

S. Carroll, M. Al-Rubeai / Journal of Immunological Methods 296 (2005) 171–178 175

Fig. 3. Cell surface biotinylation of NS0 6A1 cells. The cells were

biotinylated using a concentration of 0.05 mg ml�1, stained with

streptavidin FITC and analysed at 30 and 150 min.

S. Carroll, M. Al-Rubeai / Journal of Immunological Methods 296 (2005) 171–178176

permeable. Therefore, the amines must be present on

the outer surface of the cell in order for biotinylation

to occur. To test this hypothesis, NS0 6A1 cells were

biotinylated using 0.05 mg ml�1 biotin on day 3 of the

growth phase and monitored for 2.5 h. The whole

population of cells remained biotinylated during the

experimental period of 2.5 h (Fig. 3).

3.2. Saturation curve for ACSD labelling

To find the optimum secretion time for ACSD

labelling, NS0 6A1 cells were biotinylated and

allowed to secrete in medium with 10% (v/v) gelatine

for 15, 20, 30, 40, 50 and 60 min with the optimised

concentrations of biotinylated antibody and neutravi-

din. After secretion, the cells were labelled with the

detection antibody and analysed by flow cytometry.

ACSD labelling reached saturation after 30 min

incubation, with fluorescence detected after 15 min

of secretion (Fig. 4). During the incubation period,

cell viability decreased slightly by about 1.7% per 10

min as a consequence of gelatine toxicity.

3.3. ACSD and magnetic cell separation

NS0 6A1 cells and NS0 WT cells were mixed at a

50:50 ratio and stained using ACSD. This was

followed by incubation with anti-FITC magnetic

beads and separation using a magnetic column. The

cells were then analysed using the flow cytometer, in

which the positive population should be the secreting

cells (NS0 6A1) and the negative population should

be the non-secreting cells (NS0 WT). Fig. 5 shows the

distribution of the ungated whole populations of the

secretors and non-secretors labelled with ACSD, and

the positive and negative controls following magnetic

separation. Once the cells were separated, they were

incubated in fresh media. After an 8-h incubation

period, the medium was analysed for antibody

concentration using ELISA (Fig. 6).

Analysis revealed that the positive population of

cells secreted on average 5 times more antibody than

the cells in the negative population, indicating a good

separation of the two populations. The advantage of

using magnetic cell separation instead of cell sorting

to separate the cells was clearly seen since the positive

population of cells retained their high viability

throughout the labelling and separation process (data

Fig. 4. Saturation curve for ACSD labelling on NS0 6A1 cells and the viability of the cells. Following biotinylation of the cells at 0.05 mg ml�1

biotin, the cells were allowed to secrete in medium with 10% (v/v) gelatine with 48 Ag ml�1 biotinylated antibody and 5 Ag ml�1 neutravidin for

10–60 min. The cells were then labelled with the detection antibody and analysed. The error bars indicate the coefficient of variation (CV).

S. Carroll, M. Al-Rubeai / Journal of Immunological Methods 296 (2005) 171–178 177

not shown). It is important to note that the negative

population of cells contained some secreting cells and

it is possible that, likewise, the positive population

contained some non-secreting cells. The productivity

of the selected positive cells was about 64% of those

normally maintained in culture and this reduction may

Fig. 5. NS0 6A1 and NS0 WT separated using Mini Macs magnetic

separation. Legend: black: secretors (NS0 6A1) labelled with

ACSD, red: non-secretors (NS0 WT) labelled with ACSD, blue:

positive population after magnetic separation and green: negative

population after magnetic separation.

have resulted from gelatine toxicity and the relatively

harsh manipulations.

The results reported here demonstrate that an

efficient separation of antibody producing and non-

producing cells can be achieved by combining ACSD

and magnetic cell separation. Magnetic cell separation

is gentle and following separation the cells do not

require the aid of feeder cells for growth, or

supplementation with increased serum concentrations

which can be the case for some cells after sorting

(Rodriguez and Lodish, 1988). The advantages of

magnetic cell separation over flow cell sorting makes

this technique more suitable for the routine selection

of antibody producing cells in mammalian cell

cultures.

Fig. 6. Specific antibody production per cell of the negative and

positive cell populations following magnetic cell separation. The

error bars indicate the standard error of mean (S.E.M.).

S. Carroll, M. Al-Rubeai / Journal of Immunological Methods 296 (2005) 171–178178

Acknowledgements

This work was partially funded by Glaxo-

SmithKline.

References

Bebbington, C.R., Renner, G., Thomson, S., King, D., Abrams, D.,

Yarranton, G.T., 1992. High-level expression of a recombinant

antibody from myeloma cells using a glutamine synthetase gene

as an amplifiable selectable marker. Bio/Technology 10, 169.

Biblia, T.A., Flickenger, M.C., 1991. A model of intraorganelle

monoclonal antibody transport and secretion in mouse hybrid-

oma cells. Biotech. Bioeng. 38, 767.

Borth, N., Zeyda, M., Katinger, H., 2001. Efficient selection of

high-producing subclones during gene amplification of recombi-

nant Chinese hamster ovary cells by flow cytometry and cell

sorting. Biotech. Bioeng. 71 (4), 266.

Brosterhus, H., Brings, S., Leyendeckers, H., Manz, R.A., Miltenyi,

S., Radbruch, A., Assenmacher, M., Schimtz, J., 1999. Enrich-

ment and detection of live antigen-specific CD4+ and CD8+ T

cells based on cytokine secretion. Eur. J. Immunol. 29, 4053.

Carroll, S., Al-Rubeai, M., 2004. The selection of high producing

cell lines using flow cytometry and cell sorting. Expert Opin.

Biol. Ther. 4 (11), 1821.

Chaiet, I., Wolf, F.J., 1964. The properties of streptavidin, a biotin

binding protein produced by Streptomycetes. Arch. Biochem.

Biophys. 106, 1.

Chuck, A.S., Palsson, B.O., 1992. Population balance between

producing and nonproducing hybridoma clones is very sensitive

to serum level, state of innoculum, and medium composition.

Biotech. Bioeng. 39, 354.

Green, N.M, Konieczny, L., Toms, E.J., Valentine, R.C., 1971. The

use of bifunctional biotinyl compounds to determine the

arrangements of subunits in avidin. Biochem. J. 125, 781.

Holmes, P., Al-Rubeai, M., 1999. Improved cell line development

by a high throughput affinity capture surface display technique

to select for high secretors. J. Immunol. Methods 230, 141.

Manz, R., Assenmacher, M., Pfluger, E., Miltenyi, S., Radbruch, A.,

1995. Analysis and sorting of live cells according to secreted

molecules, relocated to a cell-surface affinity matrix. Proc. Natl.

Acad. Sci. U. S. A. 92, 1921.

McKinney, K.L., Dilwith, R., Belfort, G., 1991. Manipulation of

heterogeneous hybridoma cultures for overproduction of mono-

clonal-antibodies. Biotechnol. Prog. 7 (5), 445.

Miltenyi, S., Muller, W., Weichel, W., Radbruch, A., 1990. High

gradient magnetic cell separation with MACS. Cytometry 11

(2), 231.

Ouyang, W., Lohning, M., Gao, Z., Assenmacher, M., Ranganath,

S., Radbruch, A., Murphy, K.M., 2000. Stat6-independent

GATA-3 autoactivation directs IL-4-independent Th2 develop-

ment and commitment. Immunity 12, 27.

Rodriguez, C., Lodish, H.F., 1988. Enhanced efficiency of cloning

FACS-sorted mammalian cells. Biotechniques 24 (5), 750.

Schuberth, H.J., Kroell, A., Leibold, W., 1996. Biotinylation of cell

surface MHC molecules: a complementary tool for the study of

MHC class II polymorphism in cattle. J. Immunol. Methods 189

(1), 89.

Waldrop, S.L., Davis, K.A., Maino, V.C., Picker, L.J., 1988. Normal

human CD4+ memory T cells display broad heterogeneity in

their activation threshold for cytokine synthesis. J. Immunol.

161, 5284.

Waldrop, S.L., Pitcher, C., Peterson, D.M., Maino, V.C., Picker,

L.J., 1997. Determination of antigen-specific memory/effector

CD4+ T cell frequencies by flow cytometry. J. Clin. Invest. 99

(7), 1739.