Characterization of Cells with a High Aldehyde Dehydrogenase Activity from Cord Blood and Acute Myeloid Leukemia Samples



Aldehyde dehydrogenase (ALDH) is a cytosolic enzyme that is responsible for the oxidation of intracellular aldehydes. Elevated levels of ALDH have been demonstrated in murine and human progenitor cells compared with other hematopoietic cells, and this is thought to be important in chemoresistance. A method for the assessment of ALDH activity in viable cells recently has been developed and made commercially available in a kit format. In this study, we confirmed the use of the ALDH substrate kit to identify cord blood stem/progenitor cells. Via multicolor flow cytometry of cord blood ALDH+ cells, we have expanded on their phenotypic analysis. We then assessed the incidence, morphology, phenotype, and nonobese diabetic/ severe combined immunodeficiency engraftment ability of ALDH+ cells from acute myeloid leukemia (AML) samples. AML samples had no ALDH+ cells at all, an extremely rare nonmalignant stem/progenitor cell population, or a less rare, leukemic stem cell population. Hence, in addition to identifying nonmalignant stem cells within some AML samples, a high ALDH activity also identifies some patients' CD34+/ CD38 leukemic stem cells. The incidence of normal or leukemic stem cells with an extremely high ALDH activity may have important implications for resistance to chemotherapy. Identification and isolation of leukemic cells on the basis of ALDH activity provides a tool for their isolation and further analysis.


Aldehyde dehydrogenase (ALDH) is a cytosolic enzyme that is responsible for the oxidation of intracellular aldehydes. More than 17 human ALDH genes have been identified, and the ALDH superfamily is highly conserved across a variety of species [1, 2]. This enzyme is thought to have an important role in oxidation of alcohol and vitamin A and in cyclophosphamide chemoresistance [3, 4].

Elevated levels of ALDH have been demonstrated in murine and human progenitor cells compared with other hematopoietic cells. Because these early studies used Western blotting and intra-cellular antibody staining, they were limited to the assessment of nonviable cells [5]. More recently, a method has been developed for the assessment of ALDH activity in viable cells and has been made commercially available in a kit format. This noncytotoxic method uses a cell-permeable fluorescent substrate to identify cells with high ALDH activity [6]. Substrate converted by ALDH is a charged molecule and is unable to leave the cell as freely as the unconverted substrate. In this way, converted ALDH substrate accumulates in cells with a high ALDH activity. This approach has allowed the analysis of viable murine and human ALDH+ progenitors by flow cytometry. Human cord blood hematopoietic cells with high ALDH activity are highly enriched for primitive CD34+ cells and depleted for lineage-positive (Lin+) cells (CD3, CD14, CD20, and CD56), indicating that they do indeed represent a primitive hematopoietic cell population [7].

Acute myeloid leukemia (AML) is characterized by a relentless accumulation of immature, abnormal hematopoietic cells in the bone marrow and peripheral blood. It has been postulated that AML is a disease maintained by leukemic stem cells and, hence, may be organized in a similar way to normal hematopoiesis. Indeed, only a subset of AML cells is capable of forming colonies in vitro, and an even smaller fraction can maintain colony production for 6 weeks while on feeder layers [8]. Definitive proof that a small population of putative leukemic stem cells produces the AML blasts comes from 6-week primary and secondary engraftment experiments in nonobese diabetic/severe combined immunodeficiency (NOD/SCID) mice [9]. It has been postulated that the hierarchical organization of AML explains the pattern of remission and subsequent relapse that is typical of the response to cytotoxic chemotherapy. This has led to the suggestion that although most AML blasts are killed by cytotoxic therapy, leukemic stem cells may be spared and might be able to propagate the disease at a later time.

Two leukemia cell lines exist that have cyclophosphamide-sensitive and -resistant clones. In both cell lines reported in the literature (L1210 and BNML), the cyclophosphamide-resistant clone exhibits a higher ALDH activity as detected by Western blot and antibody labeling [10, 11]. Furthermore, constitutive expression of human ALDH1 or ALDH3 in human hematopoietic cells increases their resistance to cytotoxic agents [3, 12]. Accordingly, when expression of ALDH1 is blocked by the expression of antisense ALDH mRNA, cell sensitivity to 4-hydroperoxycyclophosphamide is greatly increased in vitro [13]. In this study, we have investigated the incidence and significance of cell subsets with high ALDH activity from different patients with AML. First, we confirmed the use of the ALDH substrate kit to identify cord blood CD34+ stem/progenitors cells and expanded on their phenotypic analysis. Via multicolor flow cytometry, we then assessed the incidence, morphology, phenotype, and NOD/SCID engraftment ability of ALDH+ cells in AML samples. AML samples had either no ALDH+ cells at all, an extremely rare nonmalignant stem/progenitor cell population, or a less rare, leukemic stem cell population. Accordingly, when injected into mice, ALDH+ cells demonstrated either normal, multilineage engraftment or malignant, unilineage AML growth. Hence, in addition to identifying nonmalignant stem cells within some AML samples, a high ALDH activity also identifies some patients' leukemic stem cells. The incidence of normal or leukemic stem cells with an extremely high ALDH activity may have important implications for resistance to chemotherapy. Furthermore, the identification of leukemic stem cells on the basis of ALDH activity offers a new technique for their isolation that relies on stem cell function rather than surface phenotype.

Materials and Methods

Primary Human Cells

Peripheral blood cells were obtained from newly diagnosed and relapsed patients with AML at St. Bartholomew's Hospital, London, after informed consent. All AML samples were frozen on the day of collection by hospital staff and were thawed on the day of ALDH analysis. We obtained cord blood from mothers attending University College Hospital, London, after informed consent. All cord blood samples were stored overnight at room temperature before ALDH analysis. The protocol was approved by the hospital research ethics committees. Mononuclear cells (MNCs) were obtained by Ficoll-Paque density centrifugation and ammonium chloride red cell lysis.


All animal experiments were performed in compliance with Home Office and institutional guidelines. NOD/SCID mice were originally obtained from Dr. Leonard Schultz (Jackson Laboratory, Bar Harbor, ME) and bred at Charles Rivers Laboratories, London. They were kept in microisolators and fed sterile food and acidified water. Mice aged 8–12 weeks were irradiated at 375 rads (137Cesium source) up to 24 hours before intravenous injection of cells.

Cell Labeling

Cells were labeled with Aldefluor reagent (Becton, Dickinson [BD] Biosciences, Oxford, U.K., as described by the manufacturer. Cells were then stained with phycoerythrin (PE)–conjugated, PE-cyanin-5 (PE-CY5)–conjugated, PE-CY7–conjugated, or allophycocyanin-conjugated anti-CD34, anti-CD7, anti-CD38 (all BD Biosciences), and anti-AC133 (CD133) (Miltenyi Biotec, Cologne, Germany, antibodies for 30 minutes at 4°C. Cells were washed and resuspended in phosphate-buffered saline (PBS) with 2% fetal calf serum and 4,6-diamidino-2-phenylindoiole (DAPI). Cells were then analyzed on a BD LSR flow cytometer. Aldefluor reagent was excited at 488 nm. Gates were set up to exclude nonviable cells and debris. The negative fraction was determined using appropriate isotype controls.

Assessment of Engraftment Potential

Samples were injected into the tail vein of sublethally irradiated 8- to 12-week-old mice. Six weeks after transplantation, mice were euthanized by cervical dislocation. The femurs, tibias, and pelvis were dissected and flushed with PBS. Red blood cells were lysed via ammonium chloride. Cells were stained with human-specific fluorescein isothiocyanate–conjugated anti-CD19, PE-conjugated anti-CD33, and PE-Cy5–conjugated anti-CD45 antibodies (all from BD Biosciences). Dead cells and debris were excluded via DAPI staining. A BD LSR flow cytometer was used for analysis. More than 100,000 DAPI-negative events were collected for each sample. Engraftment of AML was said to be present if a single population of CD45+ CD33+CD19 cells was present without accompanying CD45+ CD33CD19+ cells.

Fluorescent In Situ Hybridization

Briefly, FACSorted human cells were swollen in hypotonic (0.075 M) KCl solution and fixed in Carnoy's fixative (3:1 methanol: acetic acid) before dropping onto clean glass slides. Nuclei were aged overnight before pepsin digestion, dehydration, and application of fluorescent probes. Nuclei were incubated with probes overnight at 37°C before analysis at ×1,000 magnification on an Axioplan-2 microscope (Carl Ziess, Jena, Germany, equipped with Axiovision software.


The Student's paired t-test for significance of no difference was used throughout this report.


For consistent results, Aldefluor-stained cells must be analyzed within 2 hours of labeling. However, cells retain their ability to convert the ALDH substrate for at least 24 hours after collection. We typically stored our cord blood samples overnight before Aldefluor labeling and analysis without any detectable effect on the ALDH profile (data not shown). For consistency, all AML samples compared in this study (see below) were prepared in the same fashion and for the same time before Aldefluor labeling and analysis.

High ALDH Activity Identifies Cord Blood Cells with Immature Cell Morphology

During preliminary experiments, it was noted that the signal-to-noise ratio of ALDH staining could be improved considerably if, after labeling, cells were washed twice in a large volume of buffer to reduce residual background labeling. This refinement gave us more clearly ALDH-positive and -negative populations. Consequently, we were able to visualize a subpopulation of cells that were very bright for the ALDH substrate (Fig. 1A). These cells had the medium side and forward scatter that is characteristic of stem/progenitor cells (superimposed in Fig. 1B). These low-side-scatter, highly ALDH-positive cells accounted for 0.82% ± 0.39% of mononucleated cells in cord blood (range, 0.35%–1.29%; n = 17).

Figure Figure 1..

High aldehyde dehydrogenase (ALDH) activity identifies cells with stem/progenitor cell morphology. (A): Example given of cord blood mononuclear cell ALDH staining exhibits a bright, low/ medium side-scatter population that is extremely bright for ALDH substrate when compared with monocytes (M), lymphocytes (L), or debris. Dead cells were excluded via 4, 6-diamidino-2-phenylindoiole negativity. (B): When superimposed over the whole mononuclear cell population (gray), ALDH+ cells (black) seem to have the scatter characteristics of stem/progenitor cells.

Most Lin Cells Are CD34+, ALDH+ Cells

Once we had investigated the presence of cells with a high ALDH activity in mononuclear cells, we progressed to examine Lincells (n = 9). Cord blood mononuclear cells were StemSep depleted for cells positive for lineage antigens (CD2, CD3, CD14, CD16, CD19, CD24, CD56, CD66b, and CD235a). When representative negative fractions (n = 4) were analyzed, the mean (± SD) purity of Lin cells was 99.3% ± 0.35%. Of these Lin cells, 71.1% ± 9.1% possessed a high ALDH activity (Fig. 2A). These ALDH+ cells were separate and distinct from the ALDH cells and therefore easily identified.

Figure Figure 2..

Phenotyping of cord blood lineage-negative (Lin) cells. (A): CD34 versus aldehyde dehydrogenase (ALDH) pseudo-color dotplot of a representative cord blood, Lin ALDH/CD34/CD38/ CD133 stain. Density of events is represented by color in 25% intervals from red to yellow to green to blue. Dead cells were excluded via 4,6-diamidino-2-phenylindoiole staining. (B): CD34/CD38 dot-plot generated from the same data file with CD34+/CD38, CD34+/ CD38+, CD34/CD38+, and CD34/CD38 populations gated for further analysis. (C): A CD34 versus ALDH dotplot that displays the populations defined by CD34 and CD38 in (B). Provided in the background for comparison is the whole Lin population (gray). In addition, CD34+/CD38 cells (blue), CD34+/CD38+ cells (green), Lin/CD34/ALDH+/CD38+ cells (turquoise, upper left quadrant), and Lin/CD34/ALDH+/CD38 stem cells (red dots, upper left quadrant) are displayed. (D): CD34 versus CD133 dotplot of a representative cord blood, Lin ALDH/CD34/CD38/CD133/CD7 stain with ALDH+ (red) and ALDH (blue) populations indicated. (E): CD133/ CD7 dotplot of Lin cells with CD133+, CD133, and CD7+ cell populations identified. (F): CD34 versus ALDH dotplot that displays the populations defined by CD133 and CD7 in (E). Provided are CD133+/ CD34+ cells (green), CD133/CD34+ cells (red), and Lin/CD34/ ALDH/CD7+ cells (blue dots, lower left quadrant).

Simultaneous analysis of ALDH, AC133 (CD133), CD34, and CD38 (n = 8) revealed some interesting staining patterns. Although the Lin/CD34+ cell population virtually overlaps with the Lin/ALDH+ cell population, this is not a complete correlation. Indeed, 93.3% ± 3.4% of Lin/CD34+ cells are ALDH+, and 94.3% ± 2.5% of Lin/ALDH+ cells are positive for the CD34 antigen. Interestingly, this leaves a small proportion of Lin cells that were positive for ALDH but negative for CD34 (3.4% ± 1.9%), and a slightly larger proportion (5.1% ± 2.6%) were Lin/CD34+/ ALDH cells (Fig. 2A; Table 1). Although CD38 coexpression analysis revealed that most of the rare Lin/CD34/ALDH+ cells coexpressed CD38 in large amounts (86.3% ± 4.1%), indicating that most of this subset are probably committed cells [14], rare Lin/CD34/ALDH+/CD38 cells do exist. These Lin/CD34/ ALDH+/CD38 cells represent an extremely rare cell population, which accounts for 0.19% of Lin cells and approximately one mononuclear cell in six million. Lin/CD34/ALDH+/CD38 cells coexpressed low levels of CD133, but the level was the same as the whole Lin/CD34/ALDH+ cell fraction (p = .73).

Table Table 1.. Summary of lineage-negative cord blood analysis
  • a

    Data are from seven experiments. Provided are the mean and SD for each fraction. Data are expressed as a percentage of DAPI-negative (live), debris-free, lineage-negative cells in the upper row and as percentage of DAPI-negative (live), debris-free, stem/progenitor cell subsets in the lower row. Most ALDH+ cells within the lineage-negative population coexpress CD34. A high ALDH activity is more associated with a primitive cell phenotype (CD34+/CD38/CD133+) than with a mature CD34+ cell phenotype (CD38+/CD133).

  • a

    aSignificantly different ALDH expression to corresponding CD38- or CD133-defined subset.

  • c

    Abbreviations: ALDH, aldehyde dehydrogenase; DAPI, 4,6-diamidino-2-phenylindoiole.

Subsets defined by CD34 and ALDH expressed as a percentage of lineage-negative cells
Mean ± SD3.4% ± 1.9%63.0% ± 5.7%5.1% ± 2.6%28.4% ± 7.7%
Percentage of phenotypically defined stem/progenitor cell subsets that are ALDH+
Mean ± SD91.3% ± 2.5%a78.3% ± 6.3%91.5% ± 2.7%a48.7% ± 26.0%

Previous work has indicated that the Lin/CD34 fraction of cord blood cells contains committed CD7+ lymphoid progenitors, and we wondered if these were ALDH+. Our experiments reveal that most CD7 expression is restricted to Lin/CD34/ALDH cells and, accordingly, we could not detect any CD7 expression in Lin/CD34/ALDH+/CD38 cells.

Interestingly, there seemed to be an association between the primitive CD34+/CD38 cell phenotype and ALDH+ cells (91.3% ± 2.5% ALDH+). Relatively more mature cells with a CD34+/ CD38+ cell phenotype seemed to possess a slightly lower proportion of ALDH+ cells (78.3% ± 6.3%; example in Fig. 2C and summary of data in Table 1). This difference in percentage of ALDH+ cells between the CD38+ and CD38 subsets of cord blood CD34+ cells was statistically significant (p = .001; n = 7). Accordingly, the primitive CD133+ subset of CD34+ cells was more enriched for ALDH+ cells than the relatively mature CD133 fraction of CD34+ cells (p = .001, n = 7; data in Table 1).

Because ALDH+/Lin cells have already been assessed in NOD/SCID mice [15], we merely confirmed the NOD/SCID engraftment of ALDH+ cells in our laboratory (nine of nine mice injected gave 6-week, multilineage engraftment).

Three Different Patterns of ALDH Activity Are Detectable in AML Samples

Having confirmed the ability of the kit to identify CD34+stemcells, we then progressed to the analysis of malignant hematopoietic cells with a high ALDH activity. In contrast to the remarkable consistency of cord blood ALDH labeling, the staining of AML samples gave more varied profiles. When cells with a high ALDH activity were detected in AML samples, two highly different patterns of high ALDH activity were observed. In one pattern, ALDH+ cells were similar to cord blood ALDH+ cells; they were extremely rare and possessed typical stem/progenitor cell scatter characteristics (rare pattern). In the other pattern, ALDH+ cells were more frequent and often possessed a higher side scatter than normal stem/ progenitor cells (numerous pattern). In approximately one fourth of AML samples examined (5 of 19), no subset of cells with a high ALDH activity was observed (defined as negative pattern).

The rare pattern occurred in 37% (7 of 19) of samples tested (Fig. 3). To confirm the specificity of our ALDH labeling, we incubated an aliquot of the stain with an inhibitor specific to ALDH (diethylaminobenzaldehyde [DEAB], supplied with kit). The mean (± SD) percentage frequency of ALDH+ cells in these six AML samples was 0.14% ± 0.14% (range, 0.02%–0.35%). The frequency of cells in the same gate, but applied to the inhibitor control, was 0.01% ± 0.03%. This difference in ALDH+ cell frequency in the presence of the inhibitor was statistically significant (p = .03). Similar to cord blood, most of these rare ALDH+ cells were CD34+ (88.7% ± 9.2%), confirming their primitive nature. This was even true when the AML itself was CD34. On a CD34 versus ALDH dotplot, these cells appeared to be completely distinct and separate to the main AML cell population (Fig. 3, first row).

Figure Figure 3..

Three patterns of aldehyde dehydrogenase (ALDH) activity are observed in acute myeloid leukemia (AML) patient samples. Representative analysis from each of the three patterns of ALDH activity observed in AML samples. In all analyses, dead cells were excluded via 4,6-diamidino-2-phenylindoiole staining. Provided for comparison is the appropriate inhibitor control for each ALDH stain diethylaminobenzaldehyde (DEAB). Each row was generated from one data file. In the first row, ALDH+ cells appear as a small but distinct subset of the main AML cell population (rare pattern; patient 2, 91% leukemic blasts). Regardless of the CD34 status of the AML, these rare ALDH+ cells were almost exclusively CD34+, confirming their primitive nature. The second row is an example from the more frequent ALDH pattern (numerous pattern; patient 15, 55% leukemic blasts). In this pattern, the CD34 status of ALDH+ cells was more varied than the rare pattern. ALDH+ cells were not distinct to the main AML population; rather they appeared as continuation of the main body of cells. The AML analysis presented in the third row did not exhibit any detectable ALDH activity and is defined as the negative pattern (example given is patient 16, 50% blasts).

The numerous pattern of ALDH labeling was also observed in 37% (7 of 19) of AML samples tested (Fig. 3, second row). In this pattern, ALDH+ cells were often of a higher side scatter than normal stem/progenitor cells. The mean percentage frequency of ALDH+ cells in this pattern was 19.6% ± 18.4% of total live cells. The frequency of cells in the DEAB inhibitor control was 0.03% ± 0.03%. This difference in ALDH+ cell frequency in the presence of the inhibitor was also statistically significant (p = .03). In contrast to the coexpression profile of the rare pattern, in the numerous pattern, a lower proportion of ALDH+ cells coexpressed the CD34 antigen (38.0% ± 33.7% CD34+). This difference in CD34 expression between the two ALDH staining patterns was statistically significant (p = .003).

NOD/SCID Repopulating Activity of Patterns 1 and 2 Reveals a Normal Versus Leukemic Stem Cell Potential, Respectively

To investigate the leukemic or nonleukemic nature of ALDH+ cells in AML samples, we injected cells sorted on the basis of ALDH into sublethally irradiated NOD/SCID mice. Murine marrows were analyzed 6 weeks after transplant for the presence of human myeloid (CD45+/CD33+) and lymphoid (CD45+/CD19+) cells (Table 2).

Table Table 2.. Summary of NOD/SCID engraftment data
  • a

    ALDH+ or ALDH cells were sorted and injected into sublethally irradiated NOD/SCID mice. Marrows were analyzed for the presence of human myeloid and lymphoid engraftment. Provided here is the number of mice engrafted per number of mice injected for each fraction, as well as the type of cells produced. ALDH+ cells from the rare-pattern AML samples produced normal, multilineage engraftment, whereas ALDH+ cells from the numerous-pattern AML samples produced myeloid cells only without any accompanying B cells.

  • a

    aThe only mouse that did engraft received five times as many ALDH cells as any other mouce in this experiment.

  • c

    Abbreviations: ALDH, aldehyde dehydrogenase; AML, acute myeloid leukemia; NOD/SCID, nonobese diabetic/severe combined immunodeficiency.

Sample typeALDH+ALDH
Cord blood9 of 9; multilineage0 of 3
Rare pattern2 of 4; multilineage3 of 3; myeloid only
Numerous pattern4 of 4; myeloid only1of 5a; myeloid only

When injected into NOD/SCID mice, ALDH+ cells from patient 15 (Fig. 3, second row) and patient 4 (data not shown) produced only CD33+ myeloid cells (four of four). This generation of myeloid cells without any accompanying CD19+ lymphoid cells suggests to us that in the numerous pattern, ALDH+ cells are enriched for leukemic stem cells.

Despite the low numbers of cells injected (up to 40,000), normal multilineage engraftment was observed in the two representative rare-pattern samples analyzed (patients 2 and 8; example of patient 2 in Fig. 3, first row). Interestingly, when 107 total cells (equivalent to 13,000 ALDH+ cells) from patient 2 were injected into NOD/SCID mice, leukemia was propagated (three of three). However, when the rare ALDH+ cells present in this sample were isolated and injected into NOD/SCID mice (21,000 cells per mouse), normal multilineage engraftment was observed, suggesting a nonleukemic nature. Hence, in this particular leukemia, it seems that the AML cell population inhibits normal hematopoietic development, and ALDH activity provides a tool to examine this phenomenon.

Evaluation of the Leukemic Status of ALDH+ Cells via Fluorescence In Situ Hybridization Analysis

Wherever possible, ALDH+ cells were FACSorted from the main AML population and examined for genetic abnormalities that were characteristic of the particular AML. ALDH+ cells were sorted from patient 6 (numerous pattern) and were almost exclusively (95%) +21 (Fig. 4A), confirming their leukemic nature. Although ALDH cells from the same patient (6) contained a significant proportion of +21 leukemic cells (65%), the remainder were normal hematopoietic cells that possessed the usual two copies of chromosome 21 (Fig. 4B). Most (91%) ALDH+ cells from a numerous-pattern sample (patient 15; Fig. 3, example plot in middle row) were lacking one copy of the q arm of chromosome 5, indicating a leukemic origin (Fig. 4F). ALDH cells from patient 15 also featured many leukemic cells (58%), but the remaining cells were nonleukemic (data not shown).

Figure Figure 4..

Aldehyde dehydrogenase–positive (ALDH+) cells present at a high frequency possess characteristic acute myeloid leukemia (AML) genetic abnormalities. Cells were FACSorted before fluorescence in situ hybridization (FISH) labeling of cellular DNA. (A): Single-color labeling of chromosome 21 (green). Most ALDH+ cells possessed three copies of chromosome 21, confirming the leukemic nature of this numerous-pattern population (patient 6). (B): Analysis of ALDH cells from the same patient as (A). Although leukemic cells were present, this was at a lower frequency than ALDH+ cells. (C): Montage of images from the analysis of rare-pattern patient 12 ALDH+ cells, which possess the normal two copies of chromosome 3. (D): ALDH from the same patient as in (C). Most cells possess the abnormality that is characteristic of this AML: three copies of chromosome 3. (E): Example of dual-color, dual-fusion labeling of the t [15:17] translocation. ALDH+ cells from rare-pattern patient 8 possessed two red and two green spots, indicating a lack of the AML translocation. (F): Labeling of chromosome 5p (green) and 5q (red). Most ALDH+/CD34+/− cells exhibited one red and two green FISH spots, indicating a deletion of the q arm of chromosome 5 (numerous pattern, patient 15).

Because of the rarity of ALDH+ cells in the rare pattern, enough cells for fluorescence in situ hybridization (FISH) analysis could be prepared from only two patients. ALDH+ cells from rare-pattern patient 12 were almost exclusively normal (88% of cells possessed two copies of chromosome 3; Fig. 4C), whereas ALDH cells contained a higher proportion of leukemic cells (81% of cells had three copies of chromosome 3; Fig. 4D). Almost all (96%) ALDH+ cells from patient 8 lacked the characteristic translocation of this patient (t [15:17], Fig. 4E).

Numerous-Pattern ALDH+ Cells Are Enriched for Cells with a Primitive Phenotype

To further define the significance of leukemic cells with a high ALDH activity, we analyzed ALDH in conjunction with CD34, CD133, CD38, and CD7. This simultaneous assessment of five parameters revealed a relationship between CD38 and ALDH within CD34+ cells that was similar to our cord blood profile. Although in all experiments analyzed (n = 5), primitive CD34+/ CD38 cells displayed a higher ALDH activity than their more differentiated CD34+/CD38+ counterparts, because of the wide variation between AML samples, this association did not reach statistical significance (45.1% ± 40.7% ALDH+ within CD34+/ CD38 cells versus 25.1% ± 34.8% of CD34+/CD38+ cells; p = .18) (Fig. 5) [9].

Figure Figure 5..

Phenotyping of leukemic (numerous-pattern) aldehyde dehydrogenase–positive (ALDH+) cells. (A): Psuedo-color dotplot of a representative ALDH/CD34/CD133/CD7/CD38 stain on a leukemic sample (patient 6) that exhibited a numerous-pattern ALDH staining profile. Density of events is represented by color in 25% intervals from red to yellow to green to blue. Dead cells were excluded via 4,6-diamidino-2-phenylindoiole staining. Cells were either CD34+/ALDH+, CD34/ALDH+, CD34/ALDH, or CD34+/ALDH. (B): CD34 versus CD38 dotplot with CD34+/CD38, CD34+/CD38+, CD34/CD38+, and CD34/CD38 populations indicated. (C): A CD34 versus ALDH dotplot that displays the populations defined by CD34 and CD38 in (B). Provided in the background for comparison is the whole lineage-negative (Lin) population (gray). Displayed are CD34+/CD38 cells (blue), CD34+/CD38+ cells (green), CD34/ CD38+ cells (turquoise), and CD34/CD38 cells (red). (D): CD133/ CD7 dotplot of leukemic cells with CD133+, CD133, and CD7+ cell populations identified. (E): CD34 versus ALDH dotplot that displays the populations defined by CD133 and CD7 in (D). CD133+/CD34+ cells (green) and CD133/CD34+ cells (red) are indicated. CD7 expression was mostly restricted to the Lin/CD34/ALDH cell population (blue).

In every experiment we analyzed, the ALDH activity was higher in CD34+/CD133+ cells than CD34+/CD133 cells. However, this difference in ALDH activity was not as pronounced as in cord blood and did not reach statistical significance (55.4% ± 47.1% ALDH+ cells within CD34+/CD133+ cells compared with 43.4% ± 44.7% of CD34+/CD133 cells). These abnormal staining profiles within numerous-pattern AML samples support our suggestion that numerous-pattern ALDH+ cells are leukemic in origin.

Also in contrast to the cord blood situation, CD7 expression was not completely restricted to CD34/ALDH cells (70.3% ± 33.3%; Fig. 5). AML samples were not depleted for Lin+ cells, and hence these CD7+/CD34/ALDH cells possessed scatter characteristics typical of mature lymphocytes. However, the remaining 30% of CD7+ cells that coexpressed CD34 and ALDH were probably part of the AML clone, because they often had abnormal scatter characteristics (data not shown).


Overall, the ALDH kit is quick (1 hour in total), easy to use, and does not significantly affect cell viability or repopulation ability. The fluorescent substrate may be analyzed in conjunction with other common fluorochromes on a standard benchtop flow cytometer equipped with a 488-nm laser line. These properties suggest that this is a technique more suitable for the clinic than alternative techniques that are toxic and require expensive analytical equipment (e.g., a UV laser) [16].

Combining our refined Aldefluor protocol with our use of >99% pure Lin fractions and a live/dead assay (DAPI staining) allowed an accurate analysis of Lin cells for ALDH activity. We confirm that a large proportion of Lin cells is ALDH+ but can now report that this represents almost three times the proportion of Lin cells that was previously quoted [15]. Furthermore, the extent to which the Lin/ALDH+ and Lin/CD34+ populations overlapped has not been fully appreciated. Although it has been reported that a high proportion of Lin/ALDH+ cells are CD34+, the proportion of Lin/CD34+ cells that are ALDH+ was previously underestimated [17]. Although we report that most Lin/ CD34+ cells are also positive for ALDH and, hence, cord blood Lin/ALDH+ and Lin/CD34+ are almost overlapping populations, a high ALDH activity did identify other candidate stem cell populations (further discussed below).

Multiparameter analysis of Lin populations has allowed us to confirm that most cells within phenotypically defined stem cell subsets are enriched for ALDH+ cells. Conversely, cell subsets known to contain a higher proportion of maturing progenitors contained a lower proportion of ALDH+ cells. Interestingly, there was a significant proportion of CD34+ cells with a very high ALDH activity that did not possess the CD34+/CD38 phenotype. This suggests that analysis of ALDH activity may provide an opportunity to isolate previously unidentified CD34+ stem cells.

A high ALDH activity also identifies a small population of Lin/CD34 cord blood cells. It has been reported that the NOD/SCID repopulating subset within Lin/CD34 cells is negative for CD38, and although most of our Lin/CD34/ ALDH+ cells are CD38+, we can observe a small population of Lin/CD34/ALDH+/CD38 cells [14]. In addition, CD7 is reportedly expressed on a population of Lin/CD34 natural killer cell progenitors that do not have a multilineage NOD/ SCID repopulating ability [18, 19]. Interestingly, our Lin/ CD34/ALDH+/CD38 cells are CD7. We believe this to be a candidate stem cell population and are currently investigating their NOD/SCID engraftment.

Once we were familiar with the Aldefluor kit's ability to identify human hematopoietic stem/progenitor cells, we progressed to examine samples from patients with AML. Phenotyping and functional analysis of ALDH+ cells in samples from patients with AML has allowed us to suggest that in roughly one third of cases, cells with a high ALDH activity actually represent nonleukemic stem cells. Although the presence of nonleukemic stem/progenitors in AML samples has been previously demonstrated [20], to our knowledge, this is the first description of a nontoxic, functional method for their discrimination and isolation from the main AML cell population [21, 22].

It could be argued that the proportion of blast cells in each AML peripheral blood sample would affect the ALDH pattern observed. If all rare-pattern and negative-pattern AML samples possessed a low leukemic blast percentage, then this could explain the apparent lack of leukemic ALDH+ cells. However, comparison of the ALDH pattern we observed with each patient's leukemic blast percentage (Table 3) has revealed that many of the rare-pattern and negative-pattern AML samples had a high leukemic blast cell percentage. Because we analyzed a million live cells during each acquisition, the lowest number of leukemic blasts we analyzed was approximately 80,000 in one sample (patient 9). Consequently, we have assessed the ability of the leukemic blasts in these samples to convert the Aldefluor reagent and can conclude that the AML cells are nonreactive.

Table Table 3.. Summary of patients' clinical characteristics
  1. a

    Displayed here are the FAB type, karyotype, and resulting prognosis for the patients in this study. Provided for comparison is the type of aldehyde dehydrogenase staining patterns observed for each patient: either the rare nonleukemic population (rare pattern), the leukemic pattern (numerous pattern), or no staining at all (negative pattern).

  2. b

    Abbreviations: FAB, French-American-British morphological classification; KF, karyotyping was attempted but failed at diagnosis; ND, not done at diagnosis; NK, normal karyotype; tAML, therapy-related acute myeloid leukemia.

PatientFABKaryotypePrognosisDiagnostic blast (%)Pattern
10−9q +19Intermediate90Rare

The results of our phenotypic, functional, and FISH analysis suggest that in approximately one third of patients, cells with a high ALDH activity represent the CD34+/CD38 leukemic stem cells. This observation provides a tool to separate putative leukemic stem cells from their progeny via a functional ability rather than surface markers. This approach may reveal previously undetected leukemic stem cell populations. For instance, we often observed highly significant numbers of CD34/ALDH+ cells in AML samples and, indeed, have observed engraftment from this population (patient 4).

The isolation of leukemic and nonleukemic stem cells from the same patient should allow further investigations into the interaction between the two populations. As mentioned previously, even in this preliminary study we have been able to observe an apparent inhibition of normal stem cell development by the leukemic cell population (NOD/SCID engraftment patterns of patient 2).

Our data demonstrate that CD7+ T cells do not react with the Aldefluor reagent. Therefore, T-cell contamination of leukemic stem cells isolated via ALDH activity is likely to be minimal and restricted to CD34+ progenitors rather than mature T cells.

The coincidence of a high ALDH activity with the leukemic stem cell has important implications for resistance to chemotherapy. Successful chemotherapy is dependent on the relative advantage of the nonleukemic stem cell population over the leukemic stem cells. If the ALDH activity we detected is a major chemoresistance factor, one would expect the leukemia cases with a detectable leukemic ALDH+ population to have a worse prognosis than cases without any detectable ALDH+ cells. During this study we did not detect any association between the ALDH pattern detected and the patient's karyotype-defined prognosis or French-American-British morphological classification type. Furthermore, we cannot find any prognostic link in the literature; however, analysis of more patients and other isoforms of ALDH may well reveal an association between prognosis and ALDH activity.

According to the manufacturers, the Aldefluor kit is active against the ALDH-1 isoform but not the ALDH-3 isoform. Both ALDH1 and ALDH3 are reportedly involved in chemoresistance [3, 11, 13]. Although we have investigated ALDH-1 expression, we have not assessed the levels of ALDH-3. It is possible that certain leukemias may express detectable levels of cellular ALDH-3. It will be interesting to develop reagents that are specific to particular forms of ALDH and assess the incidence of ALDH-3+ cells in leukemia. Various other tumor cell types possess elevated levels of ALDH, and it should be possible to investigate the incidence/ significance of ALDH+ cells in other malignant cell types.

It has been reported that certain ALDH isoforms are involved in the metabolism of retinoic acid from retinal [4]. It is known that promyelocytic leukemia (M3) responds to retinoic acid–mediated differentiation, and this is an effective therapy for this disease [23, 24]. During our analysis of ALDH staining in AML, we investigated three patients with the promyelocytic form of the disease. Each of the three possible situations was observed in these patients. One patient possessed an extremely rare, multilineage-engrafting population, another seemed to have a large leukemic ALDH+ cell population, and the other did not contain any detectable ALDH+ cells at all. Hence, we could not find an association between promyelocytic leukemia and ALDH activity. This is not surprising considering the observation that although inhibition of ALDH affects the oxidation of retinal to retinoic acid, this inhibition does not affect differentiation of HL-60 cells in the presence of retinal [25]. Our results support the notion that other mechanisms are also responsible for the conversion of retinol to retinoic acid in AML-M3.

In summary, a high ALDH-1 activity identifies primitive subsets of Lin/CD34+ cells as well as candidate Lin/CD34 stem cells in cord blood. In addition, in approximately one third of patients with AML, a high ALDH activity identifies the leukemic stem cell, and in a similar proportion of patients, nonleukemic hematopoietic stem cells are highlighted.


D.P. and D.T. contributed equally to this work. This work was supported by Cancer Research UK and Association for International Cancer Research grant No. GA3160 to D.B. The authors thank Derek Davies, Gary Warnes, and Ayad Eddaoudi of the FACS Lab at Cancer Research UK for their invaluable technical expertise. Furthermore, this study would not have been possible without Julie Bee, Clare Millum, and Ella Smallcombe of our Biological Research Unit. We are also grateful to Ian Dimmick of Becton Dickinson for our first Aldefluor kit.