Transplantation, (1997) 20, 593–598  1997 Stockton Press All rights reserved 0268–3369/97 $12.00

Expression of CD44 isoforms by highly enriched CD34-positive cells in cord , bone marrow and leukaphereses

S Neu1, A Geiselhart1, M Sproll2, D Hahn1, S Kuc¸i1, D Niethammer1 and R Handgretinger1

1Children’s Hospital, University of Tu¨bingen, Tu¨bingen, Germany; and 2Bender & Co GesmbH, Vienna, Austria

Summary: anical, chemical or immunohistochemical procedures.4 This results in a decrease of malignant cells with values from 3 up to more than 5 log,4–6 but does not eradicate malignant ؍ CD34-positive cells were isolated from cord blood (n -and leukapheresed material cells in any case. A more recently applied method is enrich (4 ؍ bone marrow (n ,(8 using an immunomagnetic isolation technique, ment and reinfusion of hematopoietic stem cells. In humans ,(7 ؍ n) MACS (Miltenyi Biotec, Bergisch Gladbach, Germany). hematopoietic stem cells express CD34. Several approaches In flow cytometric analysis, cell populations after have been developed successfully to isolate CD34-positive enrichment revealed a fraction of 96.1% (cord blood), cells using immunoabsorbent, immunomagnetic or centri- 96.2% (bone marrow) and 98.6% (leukapheresis fugational methods.7 Until now, on the clinical scale it is material) CD34-positive cells. Cells were further stained hardly possible to enrich the desired cell population to hom- with antibodies specific for CD44 isoforms: CD44s ogeneity. This requirement should be met to omit the risk (SFF-2), CD44v5 (VFF-8) and CD44v6 (VFF-18). CD44- of contaminating cells. Therefore, albeit using purging positive cells were detected by directly (FITC, fluor- methods as well as enrichment strategies, the risk escein isothiocyanate) or indirectly (streptavidin-PE, of minimal residual disease in autologous transplantation phycoerythrin)-conjugated fluorochromes in flow cyto- still remains. metric analysis. Analysis was restricted to CD34-posi- A promising new strategy for elimination or minimis- tive cells. A high expression of CD44s was noted in all ation of residual malignant cells may result from the combi- kinds of material under investigation with mean values nation of purging methods and stem cell enrichment. We in the range of 98.6–100%. There was little expression therefore searched for tumor-specific markers, ie antigens, of CD44v6 (mean values in the range of 1.5–3.6%) and which are expressed on tumor cells but not on hematopo- very slight expression of CD44v5 (mean values in the ietic stem cells. range of 0.6–1.4%). The finding that CD34-positive A common antigen expressed by a variety of cells,8 hematopoietic stem cells express CD44v5 and CD44v6 including T cells, granulocytes and , is CD44. to a very small extent offers the possibility of using anti- CD44 is known as the hyaluronic acid receptor.9 A complex bodies specific to CD44v5 and CD44v6 in immunopurg- genomic organisation with at least 20 exons is responsible ing in the course of autologous stem cell transplantation. for the generation of several splice variants.8,10 Until now, Keywords: autologous bone marrow transplantation; 15 different spliced types of CD44 are known. To dis- CD44 standard receptor (CD44s); CD44 isoforms (CD44v); tinguish between the standard receptor phenotype and purging methods; ; stem cell enrich- spliced phenotypes, the commonly expressed hyaluronic ment; magnetic cell separation acid receptor was termed CD44s, all splice variants were termed CD44v.8,11 All splice variants are longer in size and bear additional extracellular domains. The gener- ation of CD44 splice variants seems to be correlated with Autologous bone marrow transplantation is of increasing processes of activation, inflammation and tumor pro- importance not only in the therapy of leukemia, but also in gression.12–15 Possibly by analogy to its function in lympho- the therapy of solid tumors. The benefit of autologous bone cyte migration, CD44 variant molecules have been shown marrow transplantation is hampered by an elevated risk of to confer metastatic capability on rat tumor cell lines.16 relapse in the course of bone marrow reconstitution. This Homologues of these variants, especially CD44v5 und relapse may result from malignant cells, which contaminate CD44v6, have been detected in several human malig- 1–3 the bone marrow to varying extents. nancies including non-Hodgkin’s lymphoma,17 breast,18 Several strategies exist to prevent cotransplantation of stomach,19 and colon20 where they have been reported to malignant tumor cells in autologous transplantation. The be associated with tumor progression and tumor spread as most prominent method of reducing the tumor mass is bone well as with a worse prognosis of patients. marrow purging. Bone marrow cells are subjected to mech- Generation of CD44v5 and CD44v6 is not restricted to metastasizing cells, expression of these splice variants seems to be a general pattern of immune defense.21 The Correspondence: S Neu, Universita¨ts-Kinderklinik Tu¨bingen, Forschungs- abteilung Pa¨diatrische Ha¨matologie und Onkologie, Rumelinstr 23, D- finding that CD44v5 and CD44v6 on the one hand are ¨ 8,22 72070 Tu¨bingen, Germany responsible for the formation of metastases and on the Received 24 February 1997; accepted 12 June 1997 other hand are expressed by immunocompetent cells, led CD44 expression of enriched CD34+ cells S Neu et al 594 us to the question whether expression of these two splice CD34-specific labeling of mononuclear cells variants is restricted to tumor cells and differentiated cells of the immune system. If so, hematopoietic stem cells Isolated mononuclear cells from cord blood as well as should be devoid of the expression of these CD44 variants. enzymatically treated cells (leukapheresis/bone marrow, samples) were resuspended in staining buffer, containing Enriched CD34-positive, CD44v5/6 negative hematopoietic + stem cells in combination with C44v5/6-positive tumor PBS/5 mM EDTA/0.5% BSA (Gibco). Labeling of CD34 cells should be a prerequisite for the above mentioned com- cells with a CD34-specific monoclonal antibody was per- bination of stem cell enrichment and purging methods. formed using the CD34 Progenitor Cell Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer’s instruction.23

Materials and methods Magnetic positive selection of CD34+ cells Collection of blood and isolation of Antibody labeled cells were resuspended in degassed stain- mononuclear cells ing buffer. Separation was performed applying the MACS (magnetic cell separation) technique,23,24 a magnetic separ- Freshly obtained umbilical cord blood (up to 50 ml) was ation system (Miltenyi Biotec). Cells were applied on Mini- mixed with 300 IU of heparin-natrium (Promonta, Ham- MACS columns, type MS, in the presence of a strong mag- burg, Germany). Mononuclear cells were enriched using a netic field. Columns were washed four times with staining modified method of density sedimentation. Blood not older buffer, followed by elution with staining buffer in the than 6 h was diluted 1:2 with PBS (Gibco, Renfrewshire, absence of any magnetic field. For elution a plunger was UK), containing 5 mm EDTA. Lymphoprep (Nycomed applied. Target cells were further purified using a second Pharma, Oslo, Norway), density 1.077, was used as the sep- column with two steps of washing. Purity of the isolated aration medium. Centrifugation was done at rcf 800 cells was assessed by flow cytometry. Fluorochrome-lab- (relative centrifugal force) for 15 min. The fraction of eled antibodies (all from Becton Dickinson) were applied mononuclear cells was washed twice with PBS/EDTA. as follows: nonspecific control antibodies (IgG1-FITC, fluorescein isothiocyanate, and IgG2a-PE, phycoerythrin), CD34-FITC (HPCA-2-FITC) and CD34-PE (HPCA-2-PE). Sample preparation of leukapheresed material and bone marrow Staining of CD34+ cells with CD44-specific antibodies Cryopreserved leukapheresed material was thawed quickly at 37°C and resuspended immediately in RPMI 1640/10% Enriched target cells from bone marrow, cord blood or leu- FCS (Hyclone, Cramlington, UK). RPMI 1640 (Biochrom, kapheresed material were washed with cold PBS (Gibco). Berlin, Germany) was applied without any indicator dye. Cells were resuspended in immunofluorescence buffer, Cells were washed twice with RPMI 1640. Freshly obtained containing 5% Polyglobin N (Troponwerke, Cologne, leukapheresed material and bone marrow samples were Germany) and 0.05% NaN3. CD44-specific antibodies washed once using RPMI 1640. To remove clumps the (murine, IgG1) were provided by Bender & Co GesmbH material was digested with 0.02% collagenase (Sigma, (Vienna, Austria). Antibodies used (CD44s (SFF-2), Deisenhofen, Germany), type V, and 200 U/ml DNAse CD44v5 (VFF-8), CD44v6 (VFF18)) were directly biotin- (Calbiochem, Bad Soden, Germany). Digestion was carried labeled or fluorochrome-conjugated (FITC). The antibody out for 45–90 min at room temperature, gently shaking the concentration applied for CD44 isoforms was in the range cell suspension. Cells were filtered through a 40 ␮m nylon of 1.0–2.5 ␮g/ml. CD34-/CD44-double fluorescence stain- cell strainer (Becton Dickinson, San Jose´, CA, USA). For ing (FITC/PE- or FITC/biotin-labeled antibodies) was per- further information on the leukapheresis material used see formed simultaneously using a CD34-specific antibody and Table 1. Bone marrow samples were obtained from the CD44 antibody of choice. Incubation took place at 4°C healthy donors. for 30 min in the dark. Cells were washed twice with PBS

Table 1 Sample description of leukapheresis material

Sample No. Diagnosis Treatment prior to leukapheresis Date of birth Storage (donor)

LPH001 Neuroblastoma Chemotherapy + G-CSF-stimulation 1991 Freshly obtained LPH002 Unclear Chemotherapy + G-CSF-stimulation 1989 Freshly obtained LPH003 Healthy donor G-CSF-stimulation 1986 Freshly obtained + − ° LPH004 Testicular carcinoma Chemotherapy G-CSF-stimulation 1964 196 C (liquid N2) + − ° LPH005 Testicular carcinoma Chemotherapy G-CSF-stimulation 1976 196 C (liquid N2) + − ° LPH006 Testicular carcinoma Chemotherapy G-CSF-stimulation 1962 196 C (liquid N2) + − ° LPH007 Testicular carcinoma Chemotherapy G-CSF-stimulation 1976 196 C (liquid N2) CD44 expression of enriched CD34+ cells S Neu et al 595 Table 2 Expression of standard and variable CD44 antigens on CD34-positive hematopoietic stem cells as indicated by FACS analysis (summary)

Sample ⌺n CD44s CD44v5 CD44v6 (n) mean range mean range mean range (%) (%) (%) (%) (%) (%)

CD34+ cord blood 8 98.8 95.4–100.0 1.4 0.0–3.2 3.6 1.9–8.1 CD34+ leukapheresis 7 98.6 97.6–100.0 1.3 0.3–2.7 1.8 1.1–2.7 CD34+ bone marrow 4 100.0 — 0.6 0.1–0.9 1.5 0.3–2.4

(4°C). Samples stained with biotin-labeled CD44 antibody analysis. The population of highly enriched CD34-positive were resuspended in immunofluorescence buffer and incu- cells was gained after leukapheresis (healthy volunteer), the bated with streptavidin-PE (Boehringer Mannheim, cell line HCT-116 was generously provided by Bender & Mannheim, Germany) at a concentration of 5 ␮g/ml. Incu- Co GesmbH. bation was carried out at 4°C for 30 min in the dark. Cells were washed twice with PBS (4°C) and were kept on ice until flow cytometric analysis. Flow cytometric analysis To test performance and reliability of the staining sys- tem, the cell lines CCRF CEM (ATCC CLL 119, American Acquisition of data took place using a single argon laser Type Culture Collection, Rockville, MD, USA), Jurkat flow cytometer, FACScan (Becton Dickinson), emitting (ATCC TIB 152) and MOLT-4 (ATCC CRL 1582) as well at 488 nm. Analysis was performed by PC-Lysys as phytohemaglutinin-stimulated lymphocytes from a heal- (Becton Dickinson), Software Version 1.0 and 1.1, on an thy volunteer were used. IBM-compatible personal computer. Different populations In addition, artificial mixing experiments were performed of CD44-positive and CD44-negative cells were dis- (Figures 1 and 2). CD34-positive cells and the carcinoma tinguished using dual parameter statistics in a dot plot of cell line HCT-116, positive for CD44v5 and CD44v6 green fluorescence vs red fluorescence (Figure 3 a–e). expression, were mixed in different ratios and examined for CD34 and CD44v expression by means of flow cytometric

CD44v6 (FL-2) CD44v5 (FL-2) 104 104

HCT-116 3 HCT-116 10 103

102 102

101

101 CD34+ cells CD34+ cells 100

100 0 1 2 3 4 100 101 102 103 104 10 10 10 10 10 CD34 (FL-1) CD34 (FL-1)

Figure 1 Flow cytometric analysis of artificial mixing experiments. Figure 2 Flow cytometric analysis of artificial mixing experiments. Analysis was performed using a FACScan (Becton Dickinson). 10 000 Analysis was performed using a FACScan (Becton Dickinson). 10 000 cells were analyzed. A dot plot of fluorescence intensities is shown cells were analyzed. A dot plot of fluorescence intensities is shown (fluorescence 1 (green) vs fluorescence 2 (red)). CD34+ cells and HCT- (fluorescence 1 (green) vs fluorescence 2 (red)). CD34+ cells and HCT- 116 cells were mixed and stained simultaneously with CD34-specific and 116 cells were mixed and stained simultaneously with CD34-specific and CD44v5-specific antibodies. Signals indicating coexpression of CD34 and CD44v6-specific antibodies. Signals indicating coexpression of CD34 and CD44v5 did not exceed background level. Distinct populations were CD44v6 did not exceed background level. Distinct populations were clearly identifiable according to their specific antibody staining. clearly identifiable according to their specific antibody staining. CD44 expression of enriched CD34+ cells S Neu et al 596 a b 4 FL1-Height (3) vs FL2-Height (4) FL1-Height (3) vs FL2-Height (4) 10 104 Control staining

3 10 103

2 10 102

1 10 101

CD34 0 10 100 0 1 2 3 4 10 10 10 10 10 100 101 102 103 104

c d 4 FL1-Height (3) vs FL2-Height (4) FL1-Height (3) vs FL2-Height (4) 10 4 CD44s 10 CD44v5

3 10 103

2 10 102

1 10 101

CD34 CD34 100 0 0 1 2 3 4 10 10 10 10 10 10 100 101 102 103 104

e FL1-Height (3) vs FL2-Height (4) 104 CD44v6

Figure 3 Flow cytometric analysis of enriched CD34+ cells isolated from 3 cord blood. Analysis was performed using a FACScan (Becton 10 Dickinson). 5000 cells were analyzed. A dot plot of fluorescence inten- sities is shown (fluorescence 1 (green) vs fluorescence 2 (red)). (a) Control staining of enriched cells with nonspecific fluorochrome-labeled antibodies (IgG1-FITC and IgG2a-PE). (b) Staining of enriched cells with CD34- specific antibody HPCA-2-FITC (green). (c) Staining of enriched cells 102 with CD34-specific HPCA-2-FITC (green) and CD44s-specific SFF-2- biotin-streptavidin-PE (red). (d) Staining of enriched cells with CD34- specific HPCA-2-FITC (green) and CD44v5-specific VFF-8-biotin-strep- tavidin-PE (red). (e) Staining of enriched cells with CD34-specific HPCA- 101 2-FITC (green) and CD44v6-specific VFF-18-biotin-streptavidin-PE (red).

CD34 100 100 101 102 103 104 CD44 expression of enriched CD34+ cells S Neu et al 597 Results leukapheresis (mean 1.8%, range 1.1–2.7%), and cord blood (mean 3.6%, range 1.9–8.1%). Efficacy of CD44-specific antibodies In contrast to weak expression of CD44 splice variants, there was strong expression of the standard form CD44s in Performance and reliability of the staining system were all types of material under investigation. With mean values verified by means of CD44-negative and CD44-positive of 98.6% (leukapheresis) and 98.8% (cord blood) almost cell material (cell lines MOLT-4, CCRF CEM, PHA-stimu- all cells showed strong CD44s expression. In bone marrow lated lymphocytes). the mean value reached the maximum of 100%. High Mixing experiments of enriched CD34-positive cells out expression of CD44s and only minor percentages of CD44- of leukapheresis (expression of CD34 Ͼ98%, expression negative cells has already been found by other groups.25 of CD44v5/CD44v6 Ͻ1%) and the cell line HCT-116, posi- In general low expression of CD44v and high expression tive for CD44 isoforms under investigation (no expression of CD44s was confirmed (Table 2). of CD34, expression of CD44v5 60%, expression of CD44v6 66%), gave further evidence for the efficacy of CD44v-specific antibodies used (Figures 1 and 2). Using Discussion flow cytometric fluorescence analysis, it was possible to The fraction of CD34+ cells in bone marrow, leukapheresis obtain distinct populations of CD34+ CD44v− cells + − + material and cord blood is very low. Several investigations (enriched CD34 cells) as well as CD34 CD44v cells report percentages in the range of 0.1–2%. This impedes (HCT-116). With the exception of some minor background proper immunophenotyping of CD34+ cells without further signals, false positive cross-reaction with signals positive enrichment. The method of magnetic cell sorting (MACS) for CD34 as well as CD44v was not observed. represents a powerful tool for the enrichment of CD34+ cells of different origins, ie bone marrow, leukapheresis and Purity of CD34+ cells enriched out of cord blood, cord blood. The method reproducibly results in high yield leukapheresis and bone marrow samples and cells enriched almost to homogeneity. This highly enriched population of CD34+ cells meets the requirement Eight cord blood samples, seven leukapheresis samples and for reliable investigations regarding the immunophenotype four bone marrow samples were processed. In comparison of a small in vivo population. In the recent past profound to nonseparated mononuclear cells, isolated cells showed a efforts have been made to examine the expression of CD44 distinct light scatter signal: cells having undergone mag- isoforms and CD34 by nonmalignant as well as malignant netic cell sorting exhibited higher values of forward scatter transformed cells of the blood system.25,26 Until now, only signals. Purity was assessed by FACS analysis. Percentages + little investigation has been done regarding the CD44 of CD34 cells were evaluated. Mean values (mean ± s.d.) expression by highly enriched CD34+ hematopoietic stem ± were as follows: cord blood 96.1 2.4%, leukapheresis cells. ± ± 98.6 0.9% and bone marrow samples 96.2 0.5%. While a very strong and ubiquitous expression of CD44s was noted, there were only minor traces of expression of Viability of nonseparated and separated cells the tumor-associated CD44 variants CD44v5 and CD44v6. It is widely accepted that CD34+ cells represent, at least to Viability of cells from cord blood, leukapheresis and bone some extent, a pool of hematopoietic stem cells. It must be marrow samples as assessed by means of trypan blue noted that the population of CD34+ cells is heterogeneous exclusion was greater than 98%. with regard to pluripotency and state of differentiation. CD44s resembles a common marker, found on all types + Expression of CD44 antigens by CD34-positive of leukocytes. Expression of CD44s in CD34 cells may hematopoietic stem cells therefore be interpreted as a sign of differentiation towards the various leukocytic lineages. The fact that CD44s has Using logical gates, flow cytometric analysis of CD44 already been found on immature cells such as CD34+ cells expression was restricted to CD34-positive cells. Individual contributes to the assumption that CD44s acts as an early samples of CD34+ cells from cord blood, leukapheresis and marker of differentiation. bone marrow showed weak expression of the CD44 vari- As already mentioned, expression of CD44v5 and ants under investigation, CD44v5 and CD44v6. With mean CD44v6 was restricted to a low percentage of CD34+ cells. values of 0.6% in samples of bone marrow (range 0.1– Lack of variant isoform expression by normal human 0.9%) there was only marginal expression of the splice CD34+ bone marrow cells has already been reported by variant CD44v5. Only slightly elevated values were found Ghaffari and coworkers.25,26 Expression of CD44v5 was for samples of leukapheresis (mean 1.3%, range 0.3–2.7%) even lower than the expression of CD44v6 in all types of and cord blood (mean 1.4%, range up to 3.2%). In compari- material under investigation. In vivo, expression of CD44v5 son to CD44v5 expression, higher expression of CD44v6 and CD44v6 is correlated with efficient trafficking of acti- was noted in all types of samples with a moderate increase vated immunocompetent cells.27 CD34+ cells fulfill the task in samples of leukapheresis (1.4 times higher CD44v6 of maintaining hematopoiesis. There is no evidence that expression), and higher increase in bone marrow and cord CD34+ cells play an active and direct role in immune blood samples (2.5 times higher). Higher mean values for defense. Therefore, it seems appropriate that hematopoietic expression of CD44v6 were accompanied by a broad range CD34+ cells lack any surface marker of immunocompetent in samples of bone marrow (mean 1.5%, range 0.3–2.4%), cells as macrophages or monocytes. CD44 expression of enriched CD34+ cells S Neu et al 598 In contrast, malignant cells mimic immunocompetent CD44 splice variants in human cutaneous melanoma and cells and express variants of CD44.27 Even though the melanoma cell lines is related to tumor progression and meta- physiological role of CD44v expression still remains static potential. Int J Cancer (Pred Oncol) 1995; 64: 182–188. unclear, expression of CD44v5 and CD44v6 may confer 11 Stauder R, Eisterer W, Thaler J, Gu¨nthert U. CD44 variant metastatic potential. isoforms in non-Hodgkin’s lymphoma: a new independent prognostic factor. Blood 1995; 85: 2885–2899. In conclusion, we found that only a small fraction of + 12 Griffioen AW, Horst E, Heider KH et al. Expression of CD44 CD34 cells, highly enriched by immunomagnetic cell sort- splice variants during lymphocyte activation and tumor pro- ing, do express CD44v5 or CD44v6, whereas in malignant gression. Cell Adhes Commun 1994; 2: 195–200. cells expression of CD44v5 and v6 is widely found and 13 Hofmann M, Rudy W, Zo¨ller M et al. CD44 splice variants correlated to an elevated metastatic potential.8,12,27,28 confer metastatic behavior in rats: homologous sequences are Mixing experiments of enriched CD34+ cells out of leu- expressed in human tumor cell lines. Cancer Res 1991; 51: kapheresis (expression of CD34 Ͼ98%, expression of 5292–5297. CD44v5/CD44v6 Ͻ1%) and the cell line HCT-116 (no 14 Matzku S, Wenzel A, Liu S, Zo¨ller M. Antigenic differences expression of CD34, expression of CD44v5 60%, between metastatic and nonmetastatic BSp73 rat tumor vari- expression of CD44v6 66%) have been performed in vary- ants characterized by monoclonal antibodies. Cancer Res ing concentrations. We could show that the antibodies used 1989; 49: 1294–1299. (specific for CD44v5 and CD44v6) were able to detect 15 Pals ST, Koopman G, Heider K-H et al. CD44 splice variants: expression during lymphocyte activation and tumor pro- selectively the population of CD44v5- and CD44v6-posi- gression. Behring Inst Mitt 1993; 92: 273–277. tive population. False-positive cross-reaction with the 16 Gu¨nthert U, Hofmann M, Rudy W et al. A new variant of population of CD34-positive cells did not take place, only CD44 confers metastatic potential to rat carci- some minor background signals were observed (Ͻ3%). noma cells. Cell 1991; 65: 13–24. In vitro purging aims to select cells with hematopoietic 17 Koopman G, Heider K, Horst E et al. Activated human lym- properties but without any malignant capacity. We there- phocytes and aggressive non-Hodgkin’s lymphomas express a fore propose the use of a two-step treatment. Starting with homologue of the rat metastasis-associated variant CD44. J immunomagnetic enrichment of CD34+ cells these cells Exp Med 1993; 199: 897–904. should be subjected to a second step of immunopurging. 18 Kaufmann M, Heider K-H, Peter H-P et al. CD44 variant exon The fact that CD44v5 and v6 are dominantly found on epitopes in primary breast cancer and length of survival. Lan- many malignant cell types but expression is limited on cet 1995; 345: 615–619. CD34+ cells suggests that antibodies highly specific for 19 Han H-J, Ho L-I, Chang J-Y et al. Differential expression of the human metastasis adhesion molecule CD44v in normal and CD44v5 and CD44v6 may provide a powerful tool in 5 carcinomatous stomach mucosa of Chinese subjects. Cancer immunopurging. 1995; 75: 1065–1071. 20 Mulder J-W, Kruyt PM, Sewnath M et al. Colorectal cancer prognosis and expression of exon-v6-containing CD44 pro- References teins. Lancet 1994; 344: 1470–1472. 21 Arch R, Wirth K, Hofmann M et al. Participation of a meta- stasis-inducing splice variant of CD44 in normal immune 1 Shpall EJ, Jones RB. Release of tumor cells from bone mar- response. Science 1992; 257: 682–685. row. Blood 1994; 83: 623–625. 22 Seiter S, Arch R, Reber S et al. Prevention of tumor metastasis 2 Ross AA, Cooper BW, Lazarus HM et al. Detection and formation by anti-variant CD44. J Exp Med 1993; 177: viability of tumor cells in peripheral blood stem cell collec- 443–455. tions from breast cancer patients using immunocytochemical + and clonogenic assay techniques. Blood 1993; 82: 2605–2610. 23 Miltenyi S, Guth S, Radbruch A et al. Isolation of CD34 hematopoietic progenitor cells by high-gradient magnetic cell 3 Brugger W, Bross KJ, Glatt M et al. Mobilization of tumor cells and hematopoietic progenitor cells into peripheral blood sorting (MACS). In: Wunder E, Sovalat H, He´non P, Serke of patients with solid tumors. Blood 1994; 83: 636–640. S (eds.). Hematopoietic Stem Cells. The Mulhouse Manual. 4 Myklebust AT, Godal A, Juell S et al. Comparison of two AlphaMedPress: Dayton, OH, USA, 1994, pp 201–213. antibody-based methods for elimination of breast cancer cells 24 Miltenyi S, Mu¨ller W, Weichel W, Radbruch A. High gradient from human bone marrow. Cancer Res 1994; 54: 209–214. magnetic cell separation with MACS. Cytometry 1990; 11: 5 Grossbard ML, Nadler LM. Immunotoxin therapy of malig- 231–238. nancy. Important Adv Oncol 1992; 111–135. 25 Ghaffari S, Dougherty GJ, Lansdorp PM et al. Altered patterns 6 Myklebust AT, Godal A, Pharo A et al. Eradication of small of CD44 epitope expression in human chronic and acute cell lung cancer cells from human bone marrow with immuno- myeloid leukemia. Leukemia 1996; 10: 1773–1781. toxins. Cancer Res 1993; 53: 3784–3788. 26 Ghaffari S, Dougherty GJ, Lansdorp PM et al. Differentiation- 7 Wunder E, Sovalat H, He´non P, Serke S (eds). Hematopoietic associated changes in CD44 isoform expression during normal Stem Cells. The Mulhouse Manual. AlphaMedPress: Dayton, hematopoiesis and their alteration in chronic myeloid leuke- OH, USA, 1994. mia. Blood 1995; 86: 2976–2985. 8 Rall CJN, Rustgi AK. CD44 isoforms expression in primary 27 Herrlich P, Zo¨ller M, Pals ST, Ponta H. CD44 splice variants: and metastatic pancreatic adenocarcinoma. Cancer Res 1995; metastases meet lymphocytes. Immunol Today 1993; 14: 55: 1831–1835. 395–399. 9 Cannistra SA, Abu-Jawdeh G, Niloff J et al. CD44 variant 28 Hofmann M, Rudy W, Gu¨nthert U et al. A link between ras expression is a common feature of epithelial ovarian cancer: and metastatic behavior of tumor cells: ras induces CD44 pro- lack of association with standard prognostic factors. J Clin motor activity and leads to low-level expression of metastasis- Oncol 1995; 13: 1912–1921. specific variants of CD44 in CREF cells. Cancer Res 1993; 10 Manten-Horst E, Danen EHJ, Smit L et al. Expression of 53: 516–521.