JOURNAL OF HEMATOTHERAPY 3:37-46 (1994)Mary Ann Liebert, Inc., Publishers

Separation of Lectin-Binding Cells Using Polystyrene CultureDevices with Covalently Immobilized Soybean Agglutinin

LISA R. SCHAIN, DAVID OKRONGLY, THOMAS B. OKARMA, andJANE S. LEBKOWSKI

ABSTRACT

The plant lectin, soybean agglutinin (SBA), has been widely used to separate heterogeneouspopulations of cells. In the field of bone marrow transplantation, SBA has been used for partialdepletion of T cells from bone marrow allografts to reduce graft-vs.-host disease. SBA's high affinityfor many different tumor cells has also indicated its use as a tumor purging agent for autologous bonemarrow transplants. We have compared two methods of cell separation using either soluble SBAagglutination, or SBA covalently attached to an activated polystyrene surface. The nonbinding SBA-cell populations generated by these two procedures were very similar in terms of cell recovery, lightscatter properties, and phenotypic profile. Notably, both SBA- fractions were enriched in cells withthe known progenitor markers, CD34, CD33, and HLA-DR, and were relatively depleted of SBAbinding cells. In addition, the activity of each SBA- cell population was measured in vitro inshort-term progenitor assays. Here, both SBA- populations were significantly enriched for CFU-GM. When device-separated SBA- cell populations were seeded into long-term bone marrow culture,they produced both increased progenitor activity and cell proliferation compared to unseparatedBMMCs. The polystyrene technology described here could reduce or eliminate many of thedrawbacks of soluble SBA agglutination, making SBA cell separation a viable and convenienttechnique for clinical application.

INTRODUCTION

THE PLANT LECTIN, SOYBEAN AGGLUTININ (SBA) is a

tetrameric glycoprotein that, in its native form, existsprimarily as ß-pleated sheets ( 1 ). Each of its four subunitsis 30 kDa, with an abundance of acidic and hydroxylicamino acid residues. One SBA molecule contains fourcarbohydrate binding sites, which have the highest affin-ity for N-acetylgalactosamine and its derivatives (1).

Historically SBA has been used for a variety of pur-poses, for example, to purify glycoproteins and to sepa-rate heterogeneous cell populations. In mice, SBA can be

used to fractionate splenocytes into B cells (SBA+) and Tcells (SBA-) (2). Moreover, murine hematopoietic stemcells bind both SBA and peanut agglutinin (PNA), and,when isolated, will engraft allogeneic recipients withoutoccurrence of graft-vs.-host disease (3,4).

In humans, SBA has been used to fractionate a varietyof cell types. SBA agglutinates both B and T cells and hasbeen used to crudely separate T helper (SBA+) and Tsuppressor (SBA-) cells (5). In addition, SBA binds tocells transformed by viral or chemical agents, and, as a

result, has been suggested as a tumor-purging agent inautologous bone marrow transplantation (6-8).

Applied Immune Sciences, Inc., Santa Clara, CA 95054.

37

SCHAIN ET AL.

Lastly, soybean agglutinin has been used extensively to

process allogeneic bone marrow grafts (9,10). SBA bindsapproximately 60-90% of all bone marrow mononuclearcells, including fibroblasts, red blood cells, stromal cells,and mature cells of both the myeloid and lymphoidlineages, yet depletes only 10-25% of hematopoieticprogenitor activity (11,12). Over 400 bone marrow trans-

plants that have been T cell depleted using SBA have beenused to engraft patients (9,10) clearly demonstrating thatSBA has minimal toxic effects on hematopoietic stemcells.

Soluble soybean agglutination is a process commonlyused in the laboratory to separate heterogeneous mixturesof cells, and, though it is effective in removing the desiredSBA+ population of cells, nonspecific trapping of SBA-cells is problematic. Soluble agglutination has otherdrawbacks, especially in the clinical setting. A standardbone marrow graft contains approximately 2-10 x 109bone marrow mononuclear cells. Using the standardagglutination process with a sample of this size is bothtime-consuming and cumbersome. The potential expo-sure to outside contaminants during this procedure isappreciable.

In order to take advantage of the binding properties ofSBA, yet minimize the problems associated with theagglutination process, we covalently attached SBA toderivitized polystyrene tissue culture flasks and usedthese devices to separate cells. Theoretically, the SBAdevices should be able to bind the same SBA+ cells whileleaving the SBA- cells in the nonadherent cell fraction.

In this report, we compare SBA- cell populationsgenerated by either a standard agglutination procedure, or

by flasks with covalently immobilized SBA. To assess theequivalence of these two populations, bone marrow

mononuclear cells were isolated from normal donors andsubjected to either soluble soybean agglutination or to

capture on the SBA devices. The two SBA- populationsproved to be similar in size, phenotype, and hematopoie-tic function.

MATERIALS AND METHODS

Cell preparationBone marrow was collected into heparin from the

posterior iliac crest of normal adult volunteers by standardprocedures. The marrow was then diluted 1:16 withDulbecco's phosphate-buffered saline—Ca2+-Mg2+ free(DPBS-CMF; Life Technologies, Grand Island, NY)—1mM ethylenediaminetetraacetic acid (EDTA; SigmaChemical Co., St. Louis, MO), and low-density bonemarrow mononuclear cells (BMMC) were isolated on

Ficoll-Hypaque density gradients after centrifugation atlOOOg for 20 min at room temperature.

Preparation offlasks with covalently immobilizedsoybean agglutinin

The devices containing covalently immobilized SBA(AIS MicroCELLector® SBA and AIS CELLector® SBA)were prepared by first chemically modifying the surfaceof untreated polystyrene T-25 flasks (Corning, Coming,NY), and then covalently attaching the soybean lectin. Toderivitize the surface polystyrene, the surface Styrolgroups were substituted with bromoacetamide groups bythe amidoalkylation reaction of /V-(hydroxymethyl)-2-bromoacetamide (0.1 M) and trifluoromethanesulfonicacid (1.0 M) (Aldrich Chemical, Milwaukee, WI) intetramethylene sulfone (Phillips Petroleum, Bartlesville,OK) for 2 hr at room temperature. The flasks were

repeatedly washed with large volumes of water, thenrinsed several times with ethanol, and air dried. Becausethe bromoacetamide group is stable at room temperaturefor several months, 25-50 "activated" flasks were pre-pared at a time and stored for coupling until needed.

Soybean agglutinin was covalently attached to thebromoacetamide activated T-25 flasks by coating eachflask with 5 ml of a solution containing 50 pg/ml SBA(Vector Labs, Burlingame, CA) in DPBS for 2 hr at room

temperature. After coupling, the solution was removedfrom each device and the flasks were rinsed 10 times withDPBS. The remaining activated sites were blocked withhuman serum albumin. The excess blocking solution was

poured off and the flasks were dried under vacuum.

Sterilization of the devices was achieved by electron beamirradiation totaling 2.7 Mrad (IRT, San Diego, CA).Flasks were then stored at 4°C.

Just prior to use, each SBA device was rehydrated withfour 10 ml rinses with DPBS. The last wash was left on thebinding surface until the flask was ready to be used.

SBA+ cell capture

BMMC were suspended in a solution of DPBS contain-ing 0.5% Gamimune (Cutter Biological, West Haven,CT) at a concentration of 5 x 106 cells/ml and incubatedfor 30 min at room temperature.

Immediately following removal of the last rehydrationwash from the SBA devices, 4 ml of cells (2 x 107 cells)was loaded into each SBA flask and the devices were

rocked to coat the binding surface. To allow cell capture,the flasks were then incubated for 1 hr at room tempera-ture on a level, vibration-free surface.

Collection of nonadherent SBA—

cells

After the 1-hr cell capture, the SBA flasks were gentlyrocked, allowing the buffer to flow across the bindingsurface and the nonadherent cells were collected. Each

38

SBA SEPARATION OF BONE MARROW

flask was then washed gently two times with 4 ml DPBS.For each wash, the DPBS was pipetted down a nonbind-ing surface in order to avoid disruption of the adherent celllayer. The flask was then capped and the fluid was gentlyrocked over the binding surface as before. Each wash was

pooled with the nonadherent cells. The adherent SBA+cells can be recovered after incubation of these cells withRPMI-1640 containing 200 mM Af-acetylgalactosaminefor 5-30 min at 37°C. These SBA+ cells can be used forfurther experimentation.

Soluble soybean agglutinationBMMC were suspended in DPBS at a concentration of

3.0 x 108 cells/ml. An equal volume of SBA (2 mg/ml)was added to the cells and the suspension was agitated byhand for 2 min at room temperature. The cell suspensionwas then overlaid onto 8 ml of DPBS containing 5% BSA(Sigma). The agglutinated cells were allowed to settle bygravity for 5-10 min at room temperature on a vibration-free surface. The SBA- cells were then collected from thetop of the tube, leaving behind the interface and theagglutinated cells penetrating the BSA layer. The SBA-cells were then washed once with 10 mM D-(+)-galactose(Sigma) in DPBS and resuspended in DPBS.

Cell phenotypingCell populations (5 x 104-1 x 106 cells/sample) were

stained with various fluorochrome-labeled antibodies bystandard methods. Samples were analyzed on an OrthoCytofluorograf Ils optical bench with a 2151 computer(Ortho Diagnostic Systems, Westwood MA). A Lexel 75argon laser (Cooper LaserSonics, Santa Clara, CA),emitting in light mode 50 mW of power at 488 nm was an

illumination source.The phenotyping panel consisted of the following

phycoerythrin and fluorescein isothiocyanate-conjugatedantibodies: mouse immunoglobulin control, anti-CD3clone #SK7, anti-CDIO clone #W8E7, anti-CD14 clone#M/-P9, anti-CD 15 clone #MMA, anti-CD 16 clone#NKP15, anti-CD19 clone #467, anti-CD20 clone#L27, anti-CD33 clone #P676, anti-HLA-DR clone#L243, anti-CD34 clone #MY10 (Becton Dickinson,San Jose, CA), sheep F(ab)2-anti-mouse IgG (New En-gland Nuclear, Boston, MA).

Colony assays

Cells (5 x 105) were suspended in 6 ml of Iscovesmodified Dulbecco's medium (IMDM) (Life Technolo-gies) containing 20% fetal bovine serum, penicillin/streptomycin (100 U/100 g/ml; Life Technologies), 5ng/ml interleukin-1, 10 ng/ml interleukin-3, 10 ng/ml

interleukin-6 (Amgen Biologies, Thousand Oaks, CA),plus 0.3% Bacto Agar (Difco Labs, Detroit, MI). Thissuspension was then plated into 2 wells of a 6-well plate (3ml/well) and incubated at 37°C in humidified air contain-ing 5% or 7% C02. Fourteen days later the plates werecounted for CFU-GM by visual inspection with an in-verted microscope.

More recently, CFUs were assayed using media con-

taining methylcellulose plus 5% PHA-LCM (Terry FoxLaboratory, Vancouver). These cultures were incubatedas above and CFU-GM, BFU-E, and CFU-GEMM were

counted 14 days later.

Long-term bone marrow culture assays

Bone marrow stromal layers were established by seed-ing 2-4x 106 BMMC into T-25 tissue culture flasks(Corning) containing McCoy's complete growth media(13). Cultures were incubated at 37°C in a humidifiedchamber containing 5% or 7% C02 for 2-4 weeks.Freshly prepared medium was used for complete mediachanges as needed.

Prior to seeding, well-established stromal cultures were

irradiated with 25 Gy and their media was changed.Within 48 hr after irradiation, 2-10 x 105 cells were

seeded into each irradiated stromal culture flask andincubated at 37°C in 5% or 7% C02. Equal numbers ofeach cell population were used to seed the stromal layers.Each cell population was seeded in triplicate. Once a

week for the following 6 weeks, nonadherent cells were

collected; cells from similar populations were then pooledand counted. Cells (4 x 105) from each population were

plated for progenitor colony assays. The remaining cellsfrom each population were distributed to the original threeflasks and returned to the incubator until the followingweek.

RESULTS

Bone marrow mononuclear cell samples from 5 healthydonors were split and simultaneously processed usingeither SBA polystyrene devices or soluble SBA aggluti-nation to directly compare these cell separation proce-dures. The number of cells recovered after either SBAtreatment was determined (Table 1). After soluble SBAagglutination, 25% (range 10-33%) of BMMC wererecovered. Likewise, after use of the SBA devices, an

average of 27% (range 12-40%) of the input cells were

collected as the nonadherent, SBA- cells. These directstudies suggest that these two procedures produced simi-lar SBA- cell recoveries (p = not significant).

Further support for this conclusion comes from 90additional experiments where either agglutination or SBAflask capture was independently performed. In these tests,

39

SCHAIN ET AL.

Table 1. Number of Cells Recovered afterSBA Treatment"

Cell Recovery

Experiment Agglutination SBA device SABCDE

Mean

2410333129

25

4012153139

27

"The percent of cells recovered for each SBA- cell fraction isshown for both agglutination and SBA device separation. Thesefigures were calculated for five separate experiments and theiraverages are presented at the bottom.

32.2% of the BMMC (n = 22 experiments, SD = 13.2)were recovered after SBA agglutination. Similarly 26.2%(n = 90 experiments, SD = 15.3) of the BMMC were

present in the SBA- fraction after flask capture (data notshown).

The binding of cells to the SBA flask was specific forthe immobilized SBA lectin. Devices coated with an

irrelevant protein, human serum albumin, failed to bindcells. Moreover, those cells bound to the SBA flask couldbe completely removed using 200 mM N-acetylgalac-tosamine in RPMI-1640 medium for 10-30 min at 37°C.

The cell size and granularity of each cell populationwere studied by light scatter flow cytometry (Fig. 1). Theinput BMMC had a large number of small, agranularcells, consisting mostly of mature lymphocytes (Fig. 1).These small cells made up 44% (range 35-55%) of allBMMC, but comprised only 27% (range 14-36%) and24% (range 14-31%) of the SBA- cells produced aftersoluble agglutination or device separation, respectively.Moreover, both treatments produced SBA- populationsenriched in large granular cells, which rose from 30%(range 20-40%) of the total BMMC to 46% (range40-52%) of SBA- agglutinated cells and 52% (range43-67%) of SBA device nonadherent cells. Neither treat-ment produced a significant change in the percent ofmedium to large agranular cells. These results indicatethat the morphological characteristics of these two SBApopulations are very similar.

To determine more conclusively that both procedureslead to the binding of SBA+ cells, the BMMC and finaltwo SBA" fractions were phenotyped using an SBA-FITC conjugate. This analysis was performed to quantifythe specific depletion of SBA+ cells. Figure 2 shows that30% of the input population stain brightly with SBA+ as

measured by SBA-FITC flow cytometry. After agglutina-tion or flask binding, the frequency of SBA+ cells in theSBA- fractions was reduced to 10 and 17, respectively.

p5

InputBMMC

i i—i—i—i—i—i—i i—i—t20 40 60 80 100

90° Light Scatter

SBA~CellsAgglutination

60

90° Light Scatter

SBA~CellsFlask

40 60 80 100

90° Light Scatter

FIG. 1. Light scatter properties of each cell population. Theforward (X axis) vs. 90° (Y axis) light scatter properties were

compared for input (BMMC), agglutination (SBA~ cells fromagglutination), and flask (SBA~ cells generated by the SBAflasks). An Ortho Cytofluorograph Ils was used as described inMaterials and Methods.

Likewise, the mean fluorescence channel was reducedfrom 53.5 to 31 and 22 in flask treated and agglutinatedpopulations, respectively.

40

SBA SEPARATION OF BONE MARROW

.oE3z

"3ü

B

Hi ,L_

Log 10 Fluorescence

FIG. 2. SBA binding of BMMC and SBA- cell populationsThe input BMMC and both SBA- populations were labeled with20 jAg/ml FITC-conjugated SBA (Vector Labs, Burlingame,CA) and analyzed by flow cytometry. (A) Input BMMCs; (B)SBA- population after flask depletion; (C) SBA- populationafter agglutination. The arrows indicate the position in thecoltrols where 1.0% of the cells display positive fluorescence.

To characterize the two SBA- cell populations further,phenotypic analyses were performed on all cell popula-tions using flow cytometry. In these studies, two cell-gating protocols were used. The first was a total cell gatethat included all cells and excluded only debris. Toexclude the highly autofluorescent granular cells, whichcomprise nearly half of the SBA- cells from analysis, asecond lymphoblastoid cell-gate containing agranularcells of all sizes was used.

Table 2A shows that the proportions of lymphoblastoidcells expressing the CD3, CD15, CD 16, CD33, andHLA-DR surface antigen are all enriched after removal ofthe SBA+ cells regardless of the procedure used. Espe-cially noteworthy are the 2.3 to 5.8-fold enrichments inCD16+ NK cells seen in these experiments. Also evident

are the 1.9 to 3.3-fold increases in CD33+ cells and 1.5 to3.3-fold enrichments in HLA-DR+ cells after SBA treat-ment. These last two markers are present on progenitorcells and their enrichment is suggestive of stem cellenrichment in the SBA- fractions (14,15). Other cellsurface antigens investigated in these experiments were

CD10, CD14, CD19, and CD20. Cells expressing thesemarkers were minimally changed in proportion upon SBAtreatment.

In most cases examined, the SBA devices and aggluti-nation performed similarly to enrich or deplete certain cellsubsets. However, in three of four experiments (Table2A, Experiments A, C, and F), there was a greaterenrichment of CD33+ cells after use of the SBA polysty-rene device. In addition, in the two experiments whereHLA-DR+ cells were analyzed, again the SBA- cellsfrom the polystyrene devices showed greater enrichmentfor this marker. These results may be due to less nonspe-cific trapping of SBA- cells on the flask surface, and are

consistent with greater progenitor retention and enrich-ment after use of the SBA devices.

Table 2B shows the phenotypic analyses on the same

samples using total cell gating. This gating procedure isespecially suitable for tracking CD15+ and CD33+ cellsthat exhibit various degrees of granularity. Exact quanti-tation of lymphoid cells using this gating procedure can beproblematic. Using these parameters, CD15+ and CD33+cells show a measurable enrichment in both SBA- frac-tions, consistent with the overall enrichment of granularcells after SBA treatment. Changes in the levels of theremaining markers by SBA treatment were undetectable.

CD34+ cells were also enriched in the SBA- cellfraction after SBA polystyrene device capture. Althoughwe have not directly compared the CD34+ cell enrich-ment generated using both SBA treatments, we haverepeatedly observed on average 2.5-fold enrichments(n = 28) of CD34+ cells after removal of SBA+ cellsusing the SBA flasks (Table 3). The rise in the proportionof CD34-positive cells in the SBA- fraction is againconsistent with the reported progenitor enrichment ob-served after SBA agglutination (12). The percent ofCD34+ cells in the input BMMC is within the noise levelsof flow cytometry and therefore accurate estimation of thetrue number of CD34+ cells in the population is extremelydifficult. However, these rough values together with thecell number in each fraction can be used to calculate therecovery of CD34+ cells in the SBA- fraction. In theexperiments shown in Table 3, the mean CD34+ cellrecovery was 65.8% (SD 47.7).

SBA binds to many mature, differentiated cells in bonemarrow, resulting in an enrichment of progenitors in theSBA- cell fraction (12). To assess the equivalence ofstandard agglutination and the SBA device, the progenitoractivity within the input BMMC and the two differentSBA- populations was measured in semisolid agar as-

41

SCHAIN ET AL.

Table 2. Phenotypic Analysis of Input and SBA Negative Cell Populations3

A. Lymphoblast cell gate used to analyze phenotypeExperiment Population CD3 CD10 CD14 CD15 CD16 CD19 CD20 CD33 HLA-DR

D

InputAgglutDeviceInputAgglutDeviceInputAgglutDeviceInputAgglutDeviceInputAgglutDevice

16.919.323.218.432.527.821.038.035.0

1.81.10.00.00.01.4

2.62.06.14.50.26.2

5.211.98.22.6

17.914.9

5.522.212.5

1.79.86.58.0

19.019.0

4.99.78.40.00.02.3

10.09.0

15.0

3.47.31.40.71.33.6

0.82.27.82.74.98.25.5

18.114.47.4

14.123.9

6.510.018.67.86.5

26.0

Experiment

B. Total cell gate used to analyze phenotypePopulation CD3 CD10 CD14 CD15 CD16 CD19 CD20 CD33 HLA-DR

InputAgglutDeviceInputAgglutDeviceInputAgglutDeviceInputAgglutDevice

9.43.89.7

13.42.1

11.7

0.00.20.00.00.01.0

4.50.04.02.10.02.7

20.848.434.718.454.358.1

4.83.54.22.10.02.7

6.510.36.30.00.00.8

3.40.00.10.00.02.1

24.540.245.6

5.810.015.832.342.042.918.016.949.9

7.83.2

10.63.62.9

11.1

"The input BMMC population and each SBA cell population were analyzed by flow cytometry for various phenotypic markers.

says. On day 14, frequency of CFU-GM activity wasdetermined for each population. The results are summa-rized in Table 4.

Both SBA- populations were enriched in progenitoractivity compared to that of the BMMC population. Onaverage, the agglutinated SBA- cells were 2.5-fold en-

riched in CFU-GM activity (range 1.3-3.4) compared tothe input BMMC. Likewise, the nonadherent cells fromthe SBA devices possessed an average 3.5-fold (range1.7-6.2x) higher CFU-GM activity than the BMMC.

Similar results were observed in an additional 64experiments where either SBA agglutination or devicecapture were independently evaluated. In these experi-ments, agglutination yielded 2.6-fold enrichment ofCFU-GM activity (n = 24 experiments, SD = 1.8),whereas flask capture produced an average 4.2-fold en-

richment (n = 66 experiments, SD = 4.9, data notshown). In these experiments, higher enrichments ofCFU-GM were routinely observed (p = 0.05) after cellpreparation with the SBA devices. These results are

consistent with the phenotypic analysis and may be due toreduced nonspecific loss of progenitor cells on the SBAflasks. These collective results demonstrate that bothprocedures produce SBA- cell populations that are notonly morphologically similar, but also share enhancedhematopoietic activity.

In these experiments, the percent of CFU-GMs recov-ered in the SBA- population after either soluble aggluti-nation or SBA device separation was compared. In 14experiments utilizing agglutination only, a mean of72.4% of the CFU-GM's were retained in the SBA-population (SD = 19.5, data not shown). Similarly, in 66

42

SBA SEPARATION OF BONE MARROW

Table 3. CD34 Phenotype Analysis of Input andSBA- Cells"

CD34^Experimentnumber

123456789

10111213141516171819202122232425.262728

x

SD

Input7.62.74.52.71.02.95.97.25.32.84.78.39.0

12.02.4

18.31.73.52.82.94.74.11.65.53.41.6214.0

4.83.7

SBA~

18.311.915.014.212.22.9

31.97.1

13.76.1

14.027.113.09.03.3

30.06.7

11.24.46.6

10.17.4

12.717.66.05.44.78.1

11.87.5

X-Enriched

2.44.43.35.3

12.21.05.41.02.62.23.03.31.40.81.41.63.93.21.62.32.11.87.93.21.83.42.22.1

2.5

"Input BMMC and SBA device generated SBA"analyzed by flow cytometry for CD34+ cells.

cells were

experiments using only the SBA devices, a mean of72.2% recovery of CFU-GM activity was observed(SD = 33.6, data not shown). The results again indicatethe functional equivalence of the two SBA cell separationprocesses, and that the SBA devices can be used to enrichhematopoietic progenitors with a high yield of these cellsin the SBA- cell fraction.

The hematopoietic activity of the SBA- cell populationproduced from the polystyrene devices was also analyzedin long-term bone marrow culture. Cultures seeded withthe SBA- cells showed a greater proliferation at eachweekly time point (Fig. 3A and B). In addition, theproduction of CFU-GM progenitors in the nonadherentfraction of these cultures was also greater in those culturesseeded with SBA- cells even after several weeks inculture (Fig. 3C and D). The vast majority of progenitorexpansion occurred early during the culture, however,CFU-GM production was noted even at later time points,suggesting the presence of more primitive stem cells in theSBA- fraction. As a result, greatly enhanced cumulativeprogenitor yield was observed in those cultured seededwith SBA- cells (Fig. 4A and B). These results supportthe conclusion that SBA- cells are enriched in hematopoi-etic progenitors.

DISCUSSION

Soybean agglutinin is a well characterized lectin cur-

rently in use to deplete T cells from bone marrow al-lografts (9,10). This study compared the standard proce-dure of agglutination using soluble SBA to a uniquemethod whereby SBA is covalently attached to an acti-vated polystyrene surface. In these experiments, bonemarrow mononuclear cells were fractionated by bothprocedures to yield SBA- cell populations.

When evaluating these two SBA- populations, wefound that the number of cells recovered in the SBA-

Table 4. Enrichment of CFU-GM after SBA Treatment"

Experiment

Number of CFU-GM/250,000 cells plated

InputCFU-GM

Agglutination SBA devices

CFU-GM Enrichment CFU-GM Enrichment

ACDE

1237974

113

156266229252

1.33.43.12.2

213490176410

1.76.22.43.6

Mean 2.5 3.5

"In semisolid agar, 2.5 x 105 cells from each population were plated and incubated at 37°C in5% C02 as described in Materials and Methods. After 14 days, CFU-GM were counted for theinput (BMMC), agglutination (SBA- cells generated by agglutination), and device (SBA- cellsproduced by SBA flask separation). Enrichment of colonies within each SBA- fraction wascalculated relative to the number of CFU-GM present in the input population.

43

SCHAIN ET AL.

% of original call« »••dad % of original calla aaadad

2 3 4

weeks in culture

2 3 4

weeks in culture

«CFU-GM par cullur. #CFU-GM par cultura

2 3 4

weeks in culture2 3 4

weeks in culture

FIG. 3. Long-term bone marrow culture of SBA- populations. BMMC and SBA- cell populations were cultured on irradiatedstromal layers as described in Materials and Methods. On a weekly basis, the nonadherent cells were collected, counted, and afraction of the population was plated for progenitor assays. The percent of the original number of cells seeded onto the irradiatedstromal layers is shown (A and B). The CFU-GM content in the nonadherent cells of each stromal is determined as a function of timein culture (C and D). (•) Cultures seeded with BMMC; (+) cultures seeded with SBA- cells from the SBA devices.

fraction was similar using either procedure. Moreover,the light scattergrams and phenotypic profiles were alsosimilar for both SBA- populations showing equivalentcell distribution with regard to size, granularity, andsurface antigens. This characteristic SBA- pattern isdistinct, however, from that of the input BMMC pattern

being enriched in granular cells expressing CD15, CD33,and HLA-DR surface antigens.

The two SBA- populations showed similar function inprogenitor assays with both populations exhibiting en-

hanced progenitor activity compared to that of theBMMC. These collective data strongly suggest that SBA

2 3 4

weeks in culture

-BMMC +SBA-

2 3 4

weeks in culture

BMMC—

SBA-

FIG. 4. Cumulative production of CFU-GM from SBA cells in long-term bone marrow culture. The cumulative production ofCFU-GM from week 1 to the end of the cultures is plotted as a function of time.

44

SBA SEPARATION OF BONE MARROW

immobilized on the surface of the flask can recognize thesame population of cells that is agglutinated by solubleSBA.

Pretreatment of a bone marrow sample with SBA doeshave advantages in a number of cell processing schemes.SBA removes a variety of committed cells that are not

required for successful engraftment in BMT (9,10). Thustreatment with SBA is an efficient means to reduce thenumber of cells to be handled in a transplant without lossof those cells needed for long-term survival. Such pre-treatment would facilitate any subsequent processing ofthe marrow such as T cell depletion or stem cell purifica-tion.

The potential utility of SBA in the selection of humanstem cells is evidenced by the fact that SBA- cells are

enriched in CD34 cells. Because CD34 cells only com-

prise 0.5-3% of BMMC (15-17), the preenrichment ofthese cells using the SBA flasks can greatly facilitate theisolation and collection of a pure CD34 cells population.Additional evidence for progenitor enrichment by SBAcell separation is demonstrated by the enhanced long-termbone marrow culture activity seen in SBA- cells from thelectin device.

SBA also has potential for widespread application inthe area of bone marrow tumor purging. SBA has beenshown to bind to a variety of tumor lines including manybreast cell carcinomas, T cell leukemias, small cell lungcarcinomas, neuroblastomas, and Burkitts lymphomas(6-8). We have tested the SBA devices for their ability to

capture tumor cells (18). In these experiments, BMMCwere spiked with 7-10% radiolabeled neuroblastomas(SK-N-SH), breast cell carcinomas (BT-20), small celllung carcinomas (NCI-H69), acute bone marrow leuke-mia (RS4-11), and a human primary breast carcinoma(UACC-893). These cell mixtures were then passed overSBA devices, and the nonadherent, SBA- cell fractionscollected. These SBA- cell populations on average were60-80% depleted of all tumor cells.

Widespread use of SBA in allogeneic and autologousBMT will require a more expedient technique. The advan-tage of attaching SBA to a solid substrate becomesapparent when dealing with a full transplant size bonemarrow sample. The technology described here is easilyscaled up to a compact, sterile, large surface area device,providing convenient handling, while reducing exposureto possible outside contaminants. Such SBA devicescould significantly improve cell processing for bonemarrow transplantation.

REFERENCES

1. Goldstein, I.J., & Poretz, R.D. Isolation, PhysicochemicalCharacterization, and Carbohydrate-Binding Specificity ofLectins. In The Lectins: Properties, Functions and Appli-cations in Biology and Medicine, 1st ed., eds. Liener, I.E.,

Sharon, N., Goldstein, I.J. Academic Press, Orlando,1986, p. 33.

2. Reisner, Y., Ravid, A., & Sharon, N. Use of soybeanagglutinin for the separation of mouse B and T lympho-cytes. Biochem. Biophys. Res. Commun. 72: 1585, 1976.

3. Reisner, Y., Itzicovitch, L., Meshorer, A., & Sharon, N.Hematopoietic stem cell transplantation using mouse bone-marrow and spleen cells fractionated by lectins. Proc. Nati.Acad. Sei. U.S.A. 75: 2933, 1978.

4. Reisner, Y., Ikehara, S., Hodes, U.Z., & Good, R.A.Expression of peanut agglutinin receptors on vims-inducedpreleukemic cells in mice. Proc. Nati. Acad. Sei. U.S.A.77: 1164, 1980.

5. Reisner, Y., Pahwa, S., Chiao, J.W., Sharon, N, Evans,R.L., & Good, R.A. Separation of antibody helper andantibody suppressor human T cells by using soybeanagglutinin. Proc. Nati. Acad. Sei. U.S.A. 77: 6778, 1980.

6. Reisner, Y., & Gan, J. Differential binding of soybeanagglutinin to human neuroblastoma cell lines: Potentialapplication to autologous bone marrow transplantation.Cancer Res. 45: 4026, 1985.

7. Morecki, S., Margel, S., & Slavin, S. Removal of breastcancer cells by soybean agglutination in an experimentalmodel for purging human marrow. Cancer Res. 48: 4573,1988.

8. Mumcuoglu, M., Favrot, M., & Slavin, S. Lectin-bindingproperties of Burkitt's lymphoma cell lines: Application tobone marrow purging. Exp. Hematol. 18: 55, 1990.

9. Reisner, Y., Kapoor, N., Kirkpatrick, D., Pollack, M.S.,Cunningham-Rundles, S., Dupont, B., Hodes, M.Z.,Good, R.A., & O'Reilly, R.J. Transplantation for severe

combined immunodeficiency with HLA A, B, Dr incom-patable parent marrow cells fractionated by soybean agglu-tinin and sheep red blood cells. Blood 61: 341, 1983.

10. Reisner, Y., Kapoor, N., Kirkpatrick, D., Pollack, M.S.,Dupont, B., Good, R. A., & O'Reilly, R.J. Transplantationfor acute leukemia with HLA A and B nonidentical parentmarrow cells fractionated with soybean agglutinin andsheep red blood cells. Lancet 2: 327, 1981.

11. Ebell, W., Castro-Malaspina, H., Moore, M.A.S., &O'Reilly, R.J. Depletion of stromal cell elements in humanmarrow grafts separated by soybean agglutinin. Blood 65:1105, 1985.

12. Georgiou, G.M., Roberton, D.M., Ellis, W.M., Shen,B.J., Ekert, H., & Hosking, C.S. CFU-C enrichment fromhuman bone marrow using a discontinuous percoll gradientand soybean agglutinin in comparison with ficoll-hypaque.Clin. Exp. Immunol. 53: 491, 1983.

13. Gartner, S., & Kaplan, H.S. Long-term culture of humanbone marrow cells. Proc. Nati. Acad. Sei. U.S.A. 77:4756, 1980.

14. Veraillie, C, Blaklmer, K., & McGlave, P. PrimitiveHuman Hematopoietic cells with long-term repopulating

45

SCHAIN ET AL.

capacity adhere selectively to irradiated bone marrow

stroma. J. Exp. Med. 172: 509, 1990.

15. Andrews, R.G., Singer, J.W., & Bernstein, I.D. Mono-clonal antibody 12-8 recognized a 115 Kd molecule presenton both unipotent and multipotent hematopoietic differen-tiation antigens. Blood 67: 842, 1986.

16. Civin, C.I., Strauss, L.C., Brovall, C, Fackler, M.J.,Schwartz, J.F., & Shaper, J.H. A hematopoietic progenitorcell surface antigen defined by a monoclonal antibodyraised against KG-la cells. J. Immunol. 133: 157, 1984.

17. Watt, S., Karhi, K., Gatter, K., Furley, A.J.W., Katz,F.E., Healy, L.E., Altass, L.J., Bradley, N.J., Sutherland,D.R., Levinsky, R., & Greaves, M.F. Distribution andepitope analysis of the cell membrane glyco-protein

(HPCA-1 ) associated with human hematopoietic progenitorcells. Leukemia 1: 417, 1987.

Address reprint requests to:Lisa R. Schain

Applied Immune Sciences, Inc.5301 Patrick Henry Drive

Santa Clara, CA 95054

18. Lebkowski, J.L., Schain, L.R., Okrongly, D., Levinsky,R., Harvey, M.J., & Okarma, T.B. Rapid isolation ofhuman CD34 hematopoietic stem cells-purging of humantumor cells. Transplantation 53: 1011, 1992.

46