Stem Cells Concise Review New Strategies in the Treatment of Acute Myelogenous Leukemia (AML): In Vitro Culture of AML Cells—The Present Use in Experimental Studies and the Possible Importance for Future Therapeutic Approaches ØYSTEIN BRUSERUD,a BJØRN TORE GJERTSEN,a BRYNJAR FOSS,a TIEN-SHENG HUANGb a
Division for Hematology, Department of Medicine, Haukeland University Hospital, Bergen, Norway; b Institute of Molecular Biology, The University of Bergen, Bergen, Norway Key Words. Acute myelogenous leukemia · Experimental models · In vitro culture · Chemotherapy · Apoptosis
A BSTRACT In vitro studies of cultured native acute myelogenous leukemia (AML) blasts and cell lines have contributed significantly to our present knowledge about the pathogenesis of AML. In the present article we review different techniques for preparation and in vitro culture of AML blasts. Well-characterized serum-free in vitro conditions can now be used in experimental studies of AML, and this makes comparisons between different studies easier. We also describe assays for characterization of AML progenitor subsets (i.e., suspension cultures, colony assays, long-term in vitro culture, xenotransplantation in immunocompromised mice), and we discuss the possible use of AML cell lines as experimental models in AML. Furthermore, clinical studies suggest that the in
vitro growth characteristics of AML blasts assayed by short-term culture of the total native populations can be used as a predictor of prognosis after intensive chemotherapy. These in vitro assays may therefore be used for more accurate identification of prognostic parameters and thereby form a basis for the development of simplified laboratory techniques suitable for routine evaluation of patients undergoing risk-adapted therapy. However, it will be equally important to further evaluate the clinical relevance of assays for primitive AML progenitors, and to develop simplified methods that can be used to characterize these progenitor subsets in the routine clinical evaluation. Stem Cells 2001;19:1-11
INTRODUCTION Studies of in vitro cultured acute myelogenous leukemia (AML) cell lines and native blasts have been important for the characterization of proliferation, differentiation, and apoptosis in leukemic hematopoiesis, and for our understanding of chemotherapy effects in AML [1-6]. One would also expect
experimental studies to become important in the future characterization of genetic abnormalities as pathogenetic factors in AML [3, 4], and based on the in vitro results several new therapeutic strategies have already been suggested [1, 5, 6]. Furthermore, in vitro growth characteristics of native AML blasts may be useful as a prognostic parameter in AML
Correspondence: Øystein Bruserud, M.D., Medical Department, Haukeland University Hospital, N-5021 Bergen, Norway. Telephone: 47-55-29-80-60; Fax: 47-55-97-29-50; e-mail:
[email protected] Received October 3, 2000; accepted for publication October 3, 2000. ©AlphaMed Press 1066-5099/2001/$5.00/0
STEM CELLS 2001;19:1-11
www.StemCells.com
In Vitro Culture of AML Cells
2
[7-12]. An understanding of these AML models is therefore essential for the interpretation of experimental data and for the understanding of possibilities and limitations of the new therapeutic approaches. WHAT KIND OF AML CELL POPULATIONS SHOULD BE STUDIED? Bone Marrow Versus Peripheral Blood-Derived Native AML Blasts The bone marrow of AML patients includes by definition at least 20% or 30% leukemia blasts [1, 13], whereas only a subset of the patients has high blast counts in peripheral blood (Table 1). A high blast count in the blood seems to be an adverse prognostic factor [14-16], but it is not known whether this prognostic impact reflects a more advanced disease status (simply a quantitative difference) or a qualitative difference between the malignant cells. Thus, if AML blasts are derived from peripheral blood, the experimental results may be representative only for this particular subset of patients. Furthermore, even AML cells derived from marrow and blood in the same patients may have different phenotypes [17]. Taken together these observations emphasize the importance of standardized cell sampling from the same in vivo compartment in comparative AML studies.
Gradient Separation and Positive/Negative Selection for Preparation of Native AML Blasts Gradient separation (specific density 1.077 g/ml) is widely used for preparation of AML cells [7-12], and this method results in populations of >95% purity when samples are derived from selected patients with a high number of AML blasts relative to other low density cells [18]. Additional positive and/or negative selection is commonly used when gradient-separated blasts contain a relatively high percentage of contaminating cells. Enriched AML blast population can then be prepared by depletion of lymphocytes and monocytes, for example, by using CD2- (T lymphocyte) and CD19-specific (B lymphocyte) immunomagnetic negative selection in combination with removal of adherent cells [18]. Various methods for positive selection may also become useful for preparation of AML cell subsets, for example, CD34+ AML cells [19, 20]. However, extensive ex vivo manipulation, including cell separation procedures, can induce gene expression and thereby alter the release of soluble mediators as well as the expression of membrane molecules by native AML blasts [21-24]. The Platelet Contamination of Native AML Blast Populations Gradient-separated AML blasts are contaminated with platelets, but in our experience the platelet numbers are
Table 1. Preparation of enriched AML blast populations and culture of human AML cells; the importance of separation procedures and culture conditions for interpretation of experimental results Procedure
Major advantage
Major disadvantage
Density gradient separation used alone [7-12, 18]
The procedure is simple, cheap, fast, and suitable for handling large sample volumes, and it has only a minor influence on AML blast characteristics.
Patient selection is required because a high purity is achieved only when the sample contains a low number of contaminating low-density cells (monocytes, lymphocytes, early granulocyte stages).
Additional positive/negative selection [19-24]
Can be used for all patients independent of the number of contaminating low-density cells in the original sample.
Functional alterations can be induced in the AML blasts.
AML cell lines [28, 29]
Homogeneous and well-characterized populations; allows comparison between different studies.
Often multiple chromosomal abnormalities that are aquired during in vitro expansion; additional alterations may occur during further in vitro expansion.
Serum-containing medium [7-12, 18, 32, 33]
A medium commonly used and rich in essential nutritients.
Nonstandardized culture conditions including a wide range of unidentified soluble serum components.
Serum-free media [19, 31-33]
Standardized media allowing comparison between different studies.
The requirement for defined nutritients differs between patients, and functional characteristics are thereby influenced by the culture conditions.
Preparation of enriched AML cell populations
Culture medium
Endotoxin levels [38]
Alters AML blast functions and activates contaminating normal macrophages.
Antibiotics [39]
Reduced risk of infections.
Aminoglycosides may interfere with receptor binding of certain growth factors.
Exogenous growth factors [18, 19, 32, 33, 40]
In vitro proliferation can be induced in the presence of exogenous factors for a majority of patients; the extent of in vitro apoptosis is reduced.
The optimal growth factor or growth factor combinations differ between individual patients and between FAB types (Table 3).
Bruserud, Gjertsen, Foss et al.
3
usually too low to have any major influence on the functional characteristics of native AML blasts. First, we determined the number of platelets in gradient-separated mononuclear cells (>95% AML blasts after separation; peripheral blood samples with blood leukocyte counts >30 × 109/l and >80% AML blasts) derived from 18 patients, and the AML blast:platelet ratio was always ≥1.0 even for patients with normal platelet counts. At this ratio, normal platelets do not alter AML blast proliferation and cytokine secretion [25, 26]. Second, although activated platelets can adhere to native AML blasts [27], detectable platelet adhesion is not observed after incubation of gradient-separated AML populations under in vitro conditions that would be expected to activate the platelets (Foss et al., unpublished data). Third, we have also examined the release of platelet-derived growth factor (PDGF) isoform AB and vascular endothelial growth factor (VEGF) by gradient-separated AML populations. Although both these cytokines can be released by activated platelets, PDGF and VEGF levels showed no significant correlation, and only a minority of patients showed detectable levels of PDGF (22 of 51 patients) or VEGF (16 of 40 patients) (Foss, unpublished data). This observation indicates that both these mediators are released independently by AML blasts without detectable contribution from a low number of contaminating platelets. Taken together, the observations suggest that the platelet contamination of gradient-separated AML blasts is of minor functional importance.
by rendering the cells responsive to differentiation-inducing cytokines in the serum [34]. Salem et al. [35] concluded that bovine serum albumin (15 mg/ml), cholesterol (7.8 µg/ml), transferrin (7.7 × 10–6 M), and insulin (1 µg/ml) were important for proliferation of native AML blasts under serum-free conditions. The relative importance of each factor differs between patients [35], and the experiences of Blair et al. [19] and Bruserud et al. [32] suggest that the presence of all four factors is not essential for most patients. First, although both cholesterol and fatty acids seem to be important for regulation of proliferation and apoptosis of human AML cells [35-37], serum-free Iscove’s modified Dulbecco’s medium (IMDM) supplemented with only bovine serum albumin (2%), transferrin (200 µg/ml), and insulin (10 µg/ml) can be used for in vitro culture of native AML blasts [19]. Second, a medium developed for culture of normal stem cells and contaning IMDM supplemented with bovine serum albumin (20 ng/ml), low-density lipoprotein (0.05 mg/ml), human iron-saturated holotransferrin (0.2 mg/ml), and 2-mercaptoethanol 5 × 10–5 M (no insulin) can be used for culture of native AML blasts [32]. Several standardized serum-free media that are commercially available, can also be used for in vitro culture of native AML blasts [32, 33]. However, even serum-free culture conditions can modulate the functional characteristics of native AML blasts, particularly the constitutive cytokine secretion and the ability to function as accessory cells during T cell activation [32, 33].
AML Cell Lines: An Alternative to Native AML Blasts? Human AML cell lines are often used as in vitro models for AML disease (Table 1). Several cell lines have been extensively characterized and are regarded as typical for different AML subtypes [28, 29]. However, AML cell lines often show extensive chromosomal abnormalities, and this is a fundamental difference from native AML blasts that usually show normal karyotypes or only a limited number of abnormalities [4, 30]. Thus, results from studies of cell lines have to be interpreted with great care and need to be verified by studies of native AML blasts.
The Effects of Endotoxin on In Vitro Cultured AML Blasts Endotoxin can induce cytokine secretion by normal monocytes, and it can also alter the functional characteristics of native AML blasts [38]. It is therefore important to culture AML blasts in media with a minimal endotoxin level for two reasons: A) to minimize the influence of a minor population of contaminating, endotoxin-activated normal monocytes, and B) to avoid induction of functional alterations in AML cells.
IN VITRO CULTURE OF NATIVE AML BLASTS The Culture Medium Both native AML blasts and AML cell lines can be cultured in medium with inactivated fetal calf or human serum [7-12, 31-33], and the serum then represents a nonstandardized parameter with unidentified mediators that may have effects on AML blasts [34]. For example, cytarabine cannot induce differentiation of the AML cell line ML-1 when cultured in serum-free medium, but the drug induces differentiation of ML-1 cells cultured in serum-containing medium
Addition of Antibiotics to the Medium Antibiotics are usually added to culture media, and an aminoglycoside (e.g., streptomycin, gentamicin) alone or in combination with penicillin is often used. An effect of penicillin on in vitro-cultured AML blasts has not been described. On the other hand, the aminoglycoside neomycine has a PDGF-isoform- and receptor-specific antagonistic effect, whereas gentamicin and streptomycin have weaker effects [39]. PDGF receptors can also be expressed by native AML blasts (Foss, manuscript in preparation), and for these patients aminoglycosides may interfere with the functional characteristics of AML cells. In fact, we found that neomycin (3.3 mM) had a minor effect on the constitutive cytokine secretion
In Vitro Culture of AML Cells
4
Table 2. Cytokine-dependent proliferation of native AML blasts, a summary of the results for 50 consecutively examined AML patients1 FAB-classification of AML patients Cytokine
AML-M0/M1 (undifferentiated)
No exogenous growth factor
5/192
Lineage nonspecific factors Flt3L SCF IL-1β IL-3 GM-CSF
10/18 (1/11)2 8/18 (1/11) 8/18 (0/11) 9/18 (1/11) 11/18 (3/11)
Lineage-associated factors G-CSF M-CSF Thrombopoietin
10/18 (5/11) 4/18 (0/11) 4/18 (0/11)
2,043 (1,968-33,966)3 5,471 (1,228-23,664)3 9,894 (2,921-55,239) 3,496 (1,508-125,026) 5,850 (1,409-94,564) 9,119 (1,027-69,865)5 9,558 (1,779-66,167) 10,560 (6,550-27,289) 6,939 (4,949-45,585)
AML-M2 (granulocytic differentiation)
AML-M4/M5 (monocytic differentiation)
2,222 (1.084-28,192)4
5/18
9/12 (1/11) 11/12 (1/11) 11/12 (0/11) 10/12 (3/11) 10/12 (0/11)
5,992 (1,145-15,583)4 3,941 (1,243-20,142)4, 5 3,456 (1,283-31,165)4, 5 3,456 (1,283-31.165)4 9,973 (1,895-67,721)4
10/18 10/18 10/18 11/18 13/18
(0/13) (0/13) (0/13) (5/13) (3/13)
5,713 (1,091-31,554) 11,741 (2,240-64,064) 14,481 (2,216-47,005) 19,590 (1,106-74,072) 12,821 (1,349-43,384)5
11/12 (6/11) 7/12 (0/11) 8/12 (0/11)
6,685 (1,056-77,244)4, 5 5,340 (1,179-30,016)4 3,912 (1,705-21,960)4
10/18 7/18 8/18
(5/13) (0/13) (0/13)
20,284 (1,246-62,072) 5,304 (1,198-23,491) 7,072 (1,031-25,602)
4/13
3,502 (1,577-8,119)
1
Cultures were performed as described in detail previously [18, 31-33]. Briefly, native AML blasts were derived by density gradient separation from the peripheral blood of patients with a high number and percentage of AML blasts in the blood. The cells were cultured in 150 µl medium per well (RPMI 1640 with 10% inactivated fetal calf serum) for six days before 3H-thymidine was added and nuclear radioactivity determined 18 hours later. All cytokines were used at 50 ng/ml except for G- and GM-CSF that were used at 100 ng/ml. A significant effect of the indicated cytokine was defined as: A) for patients not showing spontaneous proliferation: alteration from undetectable to detectable proliferation with an incremental response exceeding 1,000 counts per minute (cpm), or B) for patients showing spontaneous proliferation: a difference in 3H-thymidine incorporation corresponding to at least 20% and an absolute value exceeding 2,000 cpm compared with the response for cells cultured in medium alone. 2 The results are presented as the number of patients with a detectable effect of the indicated cytokine relative to the total number of investigated patients. In parentheses is shown the number of patients that showed the highest proliferative response to the indicated cytokine relative to the total number of patients showing detectable proliferation in that group. For each patient group the cytokine(s) resulting in the highest fraction of responding patients is underlined, and for each cytokine the patient group having the highest fraction of responders is set in bold. The difference between the first and second nominator of each data pair thus represents the number of patients showing neither detectable spontaneous nor cytokine-dependent proliferation, and this was seven patients for AML-M0/M1, one patient for AML-M2, and five patients for AML-M4/M5. 3
The results are presented as the median response (cpm) of the responder patients (response >1,000 cpm) with the variation range in parentheses.
4
The AML subset with the highest response to a cytokine is set in bold.
5
The cytokine giving the highest median response for an AML subset is underlined.
(interleukin 1β [IL-1β], IL-6, tumor necrosis factor α [TNF-α]) by native AML blasts for a subset of PDGF-receptor positive patients (Foss et al., unpublished data). However, one would expect other aminoglycosides to have no or only minimal effects on AML blast functions. Growth Factor-Dependent Proliferation of Native AML Blasts For a subset of patients the AML blasts show autocrine (spontaneous) in vitro proliferation when cultured in medium alone, whereas for other patients the leukemia cells can only proliferate in the presence of exogenous growth factors [7-12, 18, 31-33]. These growth factors may, in addition, induce AML blast differentiation [1]. Several cytokines can function as growth factors for AML blasts as exemplified by the results summarized in Table 2. It can be seen that the optimal growth factor for expansion of AML cells differs between patients/FAB subclasses. AML-M2 cells show generally higher proliferative responses than AML-M0/M1 and AML-M4/M5 cells. For AML-M2 cells the highest number of responders was observed in the presence of IL-1β, stem cell factor (SCF) and G-CSF, whereas the highest number of responders for AML-M0 and AML-M4/5 cells was detected with GM-CSF. IL-9 is another cytokine that also functions as
a growth factor for a majority of AML patients [40]. However, when standardized in vitro conditions have to be used throughout an experimental study, we have often used the combination IL-3 + SCF + GM-CSF that initiates a strong proliferative response for a majority of patients [33, 41]. Combinations that include additional factors (e.g., erythropoietin/IL-6/G-CSF) have also been used [19, 42]. Apoptotic Cell Death During In Vitro Culture An effective treatment of AML seems to depend on induction of apoptosis, and resistance against chemotherapy can often be explained by altered regulation of apoptosis [43]. Apoptosis is detected for an increasing fraction of cells during in vitro culture of native AML blasts [18], and the proliferative in vitro response of the leukemic cells is then caused by another subset of cells [7-12]. The number of apoptotic cells is determined by several factors, for example, a decreased fraction is observed in AML populations that show autocrine proliferation, in the presence of exogenous growth factors, and when an optimal culture medium is used [33, 42, 44-45]. The occurrence of spontaneous in vitro apoptosis may thus serve as an experimental model of cell death regulation in AML.
Bruserud, Gjertsen, Foss et al.
5
Table 3. Methods used for detection and analysis of apoptotic cell death (for detailed discussion and additional references [47]) Technique
Major advantage
Major disadvantage
Morphology Transmission electron microscopy of cellular morphology
The gold standard, investigates global cellular morphology
Expensive and work-consuming technique
Light microscopy (May-Grünewald-Giemsa staining)
Low cost for preparation and analysis
Reflects late events in apoptosis. Biased sample collection when using cytospin preparation?
Epifluorescens (Hoechst, DAPI staining)1
Detects nuclear morphology of single cells in culture; used as a vital staining
Nuclear morpholgy is a late marker for apoptosis.
A vital assay detecting early and specific events; can be combined with immunolabeling and flow cytometry
Not present in all cell types
Specific for apoptosis in thymocytes
Cell type and apoptogen dependent, requires 1-3 × 106 cells
Large DNA fragments
Specific for apoptosis
—
Sub-G1 DNA peak determined by flow cytometry
—
Not present in all cell types during apoptosis
Cell surface molecule exposure Annexin V binding2
Macromolecular fragmentation Internucleosomal DNA fragmentation
RNA fragmentation
Specific for apoptosis?
A cell type and apoptogen dependent event?
Protein fragmentation, e.g., PARP, HDM2, ATM kinase3
Information about accessory pathways involved in the regulation of apoptosis
Cell type and apoptogen dependent events
Correction for cell growth is required, ideal for high throughput of samples
Measures secondary necrosis as the apoptosis endpoint
Mitochondrial function MTT3 (determination of mitochondrial dehydrogenase activity in living cells) Mitochondrial depolarization
Direct information about a regulatory pathway
Cytochrome C leakage from mitochodrium
Detects a specific and early event involving cytochrome C-dependent caspase activation
Not present in all cells
Reflects downstream caspase cascade activation Triggered by death receptor activation (e.g., CD95) Triggered by damage of endoplasmic reticulum
Caspase-3 is deleted in certain malignant cell lines, and caspase-independent apoptosis may occur by certain apoptogens.
Enzyme activation in apoptosis signaling pathways Caspase-3 activation Caspase-8 activation Caspase-12 activation 1
Hoechst, bisbenzimide 33342 or 33258; DAPI, 4,6-diamidino-2-phenylindole. Hoechst and DAPI are both DNA specific dyes exiticed by UV epifluorescence.
2
Annexin V binds to phosphatidylserine residues which are exposed on the cell surface early in apoptosis.
3
MTT, 3-[4,5dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide; PARP, polyadenosine 5′-diphosphate-ribose polymerase; HDM2, human homologue of murine double minute 2 gene, an inhibitor of p53.
Since the first definition of apoptosis in 1972 [46], a wide array of methods have been developed for detection of apoptosis (Table 3) [47]. Apoptosis was initially a morphological diagnosis [46], and transmission electron microscopy still appears to be the gold standard [47]. Our suggestion is that experimental studies of apoptosis should include initial examination by two or three independent methods, and thereafter the most convenient assay should be chosen for further studies. The results always have to be interpreted with great care because diverging apoptotic phenotypes are frequently detected both by morphological and biochemical examination, and the apoptotic phenotype may also differ between cell types. For example, in AML cells, different types of apoptosis result in differences in cell surface and nuclear
morphology, as well as differences in the levels of internucleasomal fragmentation and TUNEL positivity [48]. Simple morphological examination of the nucleus (e.g., bisbenzamide or May-Grünewald-Giemsa staining) is suitable for the initial examination of apoptosis in experimental studies. If there is doubt about the apoptotic morphology, for example, by atypical or lacking nuclear condensation or fragmentation, annexin V binding may be the method of choice. Flow cytometric analysis of annexin V can also be combined with immunostaining to allow examination of apoptosis in smaller subsets in mixed populations, for example, gradient-separated AML populations with relatively high contamination of other low-density cells [47]. However, there are certain cell types that do not expose
6
enough phosphatidylserine on the cell surface during apoptosis to be labeled with annexin V [49]. The initial studies of nuclear morphology or annexin labeling can be supplemented with characterization of molecular fragmentation or studies of selected intracellular pathways that are involved in the regulation of apoptosis, for example, assays detecting mitochondrial alterations, caspase activation, or cleavage of caspase substrates like polyadenosine 5′-diphosphate-ribose polymerase. These techniques have in common that they often describe population (and not single cell) characteristics, and mixed cell populations are thus difficult to characterize. With the growing knowledge of signaling pathways directing apoptosis [50], techniques involving detection of the caspase activation or detection of caspase substrates may be used in immunoblotting or immunostaining protocols. Again, careful interpretation of the results is necessary because not all cases of apoptosis include caspase activity, and certain cells may even lack caspase-3 [51]. These observations thus support our suggestion that an established method based on nuclear morphology in AML cells should be chosen as an initial screening assay before further characterization of the molecular mechanisms. Gene Induction During In Vitro Culture As described above, extensive cell separation procedures seem to induce gene expression in AML blasts. Several studies have also demonstrated that gene expression can be induced or modulated by in vitro culture conditions, either by the culture medium itself or by agents added to the medium (e.g., by the presence of agents interacting with histone acetylation [52]). First, altered gene expression possibly contributes to the increased expression of certain membrane molecules after in vitro culture [23, 24]. Second, the ability of native AML blasts to respond to differentiation induction depends on the culture conditions [34], and experimental evidence suggests that altered gene expression is involved in medium-induced modulation of blast differentiation [53]. Third, alterations in iron metabolism of cultured cells will also result in complex network events acting at the transcriptional and translation levels to change the expression of proteins involved in transport/uptake/utilization/storage of iron [54]. The importance of iron-saturated transferrin for optimal in vitro proliferation of native AML blasts indicates that similar alterations may occur in cultured AML cells. The occurence of spontaneous in vitro apoptosis in native AML blasts is modulated by serum-free culture conditions [33]. Other studies have described that induction of spontaneous apoptosis in AML cells cultured under serumfree conditions can be suppressed by inhibition of c-jun expression [55]. Taken together, these observations thus suggest that altered gene expression is involved in the regulation
In Vitro Culture of AML Cells of spontaneous in vitro AML blast apoptosis and possibly also in the suppression of apoptosis when using optimal serum-free conditions. THE HIERARCHICAL ORGANIZATION OF THE AML CLONE Experimental Studies Experimental data suggest that AML populations have a hierarchical organization from relatively rare primitive progenitors down to more mature cells that dominate in native AML blast populations derived from blood or bone marrow [56] (Table 4). The cells at different steps in this hierarchy differ in their expression of cell surface molecules and their self-renewal (proliferative) capacity. Short-term suspension cultures of native AML populations have been used for characterization of both proliferative responses (seven days of culture) and constitutive cytokine secretion (two days of culture) by AML cells [18]. An increasing fraction of apoptotic cells is then detected during culture [18], and assays based on 3H-thymidine incorporation of the total AML population after seven days of culture seem to reflect enrichment of more primitive colony-forming (clonogenic) progenitors [57]. This 3H-thymidine incorporation assay seems to be more sensitive than the colony-formation assay for detection of cytokine-induced proliferation [58], and proliferative characteristics reflected in this assay have also been used as a determinant of prognosis in AML [11]. Autocrine or spontaneous AML blast proliferation can be assayed by the ability to form colonies during culture in a semisolid medium (e.g., containing methylcellulose) without exogenous growth factors [7-11]. Although colony-forming (clonogenic) cells constitute a minority of the total AML cell population [7-11], their frequency is increased when cells are cultured at high densities [57]. This observation suggests that spontaneous colony formation is determined both by the growth characteristics of the clonogenic subset, and by the ability of nonproliferating neighboring AML cells to release soluble mediators that facilitate the proliferation of clonogenic cells. This dependency of clonogenic cells on local cytokine release by nonproliferating blasts is reduced when colony formation is assayed in the presence of exogenous growth factors [57], but even then the colony-forming cells constitute a minority of the total AML cell population. The frequency of clonogenic cells shows a wide variation and depends both on the culture conditions and on differences between patients. As an example, in the study of Sutherland et al. [59] the colony-forming AML cells that could respond to exogenous growth factors had a frequency between 3 and 2,600 per 105 cells and were usually derived from the CD34– subset of native AML cells.
Bruserud, Gjertsen, Foss et al.
7
Table 4. Experimental models for the study of AML cell proliferation; a summary of studies indicating a hierarchy of the human AML clone Experimental model
Description of the experimental procedure
Dominating phenotype of the proliferating cells
Comments
Short-time suspension culture with 3H-thymidine incorporation
The total population of native AML cells cultured for six to seven days before nuclear radioactivity is determined [18]
Probably clonogenic cells of the phenotype CD34– [42, 57, 59]
Reflects an enrichment of colony-forming cells, and assays the response of a subset of AML cells able to proliferate after one week of in vitro culture [57]. Regarded as more sensitive than the colony assay [58].
Primary colony formation
Native AML cells seeded directly in a colony-forming assay [18, 19, 42, 57]
Fluorouracil-sensitive CD34– cells [60]
Depending on the culture conditions (medium alone or addition of exogenous cytokines), colonies can be differentiated into small abnormal clusters of uniform morphology, blast-like/monocytic and erythroid colonies [33, 42, 59]. Colonyforming cells are a minority among native AML cells (usually
