The toxicity of bovine α-lactalbumin made lethal to tumor cells is highly dependent on oleic acid and induces killing in cancer cell lines and noncancer-derived primary cells


C. R. Brinkmann, Department of Medical Microbiology and Immunology, Aarhus University, Wilhelm Meyers Allé 4, DK-8000 Aarhus C, Denmark
Fax: +45 8619 6128
Tel: +45 8942 1778


A complex between α-lactalbumin and oleic acid (C18:1, 9 cis) has been reported to be cytotoxic to cancer cells. We have prepared such complexes and tested their activity against both cancer cell lines and noncancer-derived primary cells. Unexpectedly, some primary cell types were more sensitive to treatment than cancer cell lines. We found the complex to be cytotoxic to all of the tested cells, with a 46-fold difference between the most sensitive and the least sensitive cell type. Oleic acid by itself exhibited a remarkably similar activity. The cell-killing mechanisms of the complex and of oleic acid alone were examined by flow cytometry, testing for apoptosis- and necrosis- inducing activity. The T-cell leukemia-derived Jurkat cells primarily underwent cell death resembling apoptosis, whereas the monocytic leukemia-derived THP1 cells adopted a more necrotic-like cell death. Erythrocytes were sensitive to lysis by the complex and oleic acid. We conclude that oleic acid is cytotoxic by itself and that, in contrast to the literature, a complex of α-lactalbumin and oleic acid has cytotoxic activity against primary cells, as well as cancer cells.


7-amino-actinomycin D




American Type Culture Collection


bovine α-lactalbumin made lethal to tumor cells


Deutsche Sammlung von Mikroorganismen und Zellkulturen


human α-lactalbumin made lethal to tumor cells


human umbilical vein endothelial cells


human vascular smooth muscle cells


oleic acid


peripheral blood mononuclear cell






A complex between the milk protein α-lactalbumin (α-LA) and oleic acid (OA; C18:1, 9 cis) reportedly kills tumor cells but not healthy differentiated cells [1]. The complex, termed human α-LA made lethal to tumor cells (HAMLET), was originally made from α-LA purified from human milk [2], although complexes between OA and α-LA from other species, such as cow, goat and horse, have been shown to be equally cytotoxic to tumor cells [3]. HAMLET has a beneficial effect in vivo on human glioblastoma xenografts in the brain of nude rats [4], on bladder cancer development in mice [5], as well as in clinical trials on skin papillomas [6] and bladder cancer [7]. Accordingly, HAMLET appears to be a promising anti-cancer candidate.

To obtain the cytotoxic complex, α-LA has to be in the partially unfolded apo-state (i.e. devoid of calcium) before complex formation with OA [2]. In addition, OA was found to be a necessary cofactor for the cytotoxicity [2]. Recently, it was reported that the conformation of α-LA is not crucial because α-LA denatured by heat treatment, and a recombinant variant of α-LA, where all four cysteines have been exchanged with alanine, can be made into cytotoxic complexes using the OA-conditioned column. The resulting complexes differ structurally from HAMLET, although their cytotoxic effect is similar to that of HAMLET [8,9]. Furthermore, fragments of α-LA produced by limited proteolysis have been reported to form cytotoxic complexes with OA [10], and even equine lysozyme, a structural homolog to α-LA, can form cytotoxic complexes with OA [11,12], supporting the view that the cytotoxic effect does not require a specific interaction between OA and α-LA. A study found that mixing of α-LA and OA results in the formation of complexes consisting of four to five protein molecules that bind 68–85 OA molecules [13], whereas other studies have shown that HAMLET consists mainly of monomers and only 2–5% dimers and higher oligomers of α-LA [14,15]. In addition, the ratio between α-LA and OA differs between preparations: α-LA : OA is 1 : 4 to 1 : 8 for HAMLET [14] and 1 : 10 to 1 : 13 for bovine α-lactalbumin made lethal to tumor cells (BAMLET) [16], whereas the originally suggested 1 : 1 complex [17] appears to be an exception. It is not known how OA binds to α-LA and whether equilibrium exists between OA bound to α-LA and OA bound to other OA. There are reports on the cytotoxic activity of α-LA, either alone or in combination with OA [13,18–24]. The various investigations differ greatly with respect to the source and purity of the α-LA used, the time of exposure or concentration required, and the type of cells targeted. Thus, there is no consensus as to whether the cytotoxicity can be ascribed to a synergistic effect of α-LA and OA or exerted per se by either of the two components.

The present study investigates the proposed tumor selectivity of the α-LA:OA complex by testing the cytotoxicity of the bovine variant of the α-LA:OA complex (i.e. BAMLET) on different cell types, including cancer cells and noncancer-derived primary cells. In addition, the cytotoxicity of BAMLET is compared with the cytotoxicity of α-LA, OA and different mixtures of α-LA and OA.


Cytotoxicity of BAMLET against different cell types

The cytotoxicity of BAMLET was tested on a range of cell types. The ATP content was used as an estimate of the number of living cells because ATP is rapidly hydrolyzed and not regenerated in damaged or metabolically inactive cells. Loss of viability was quantified as decrease in ATP levels compared to NaCl/Pi-treated controls.

Figure 1 shows representative dose–response curves for six of the cell types tested. All tested cell types were sensitive to BAMLET treatment, whereas α-LA alone showed no toxicity. However, the concentration of BAMLET required to kill the cells differed between the cell types. Table 1 shows the sensitivity to BAMLET treatment of seven cancer cell lines and five primary or untransformed cell types as measured by the ATP assay. The effect of BAMLET treatment on erythrocytes is also indicated in Table 1 (see below). Both cancer cells and primary cells were sensitive to BAMLET. Peripheral blood mononuclear cells (PBMCs) were the most sensitive cells, with a 50% loss of viability (LC50) at a BAMLET concentration of 20 μg·mL−1, whereas the endothelial cells were the least sensitive with a LC50 of 1.37 mg·mL−1. A general observation was that adherent cells were more resistant to BAMLET-induced viability loss compared to cells growing in suspension. α-LA alone was not toxic in any of the concentrations or cell types tested.

Figure 1.

 BAMLET cytotoxicity estimated by assay of ATP. The effect of BAMLET and α-LA was measured by the ATP assay on six different cells types. The cells were incubated with BAMLET or α-LA for 4.5 h. (A) HL-60 (human promyelocytic leukemia-derived cell line); (B) Skov-3 (human ovarian adenocarcinoma-derived cell line); (C) B16-F0 (murine melanoma-derived cell line); (D) PBMCs; (E) keratinocytes (normal adult human keratinocytes); and (F) HUVEC. Viability was calculated from the ATP content (measured as luminescence) of test sample-treated cells expressed as percentage of controls: [test sample-treated cells − detergent-lysed cells)/(NaCl/Pi-treated cells − detergent-lysed cells)] × 100%. The results shown for each cell type are representative of at least three independent experiments. Each experiment was performed in triplicate and the range and mean is shown for each point.

Table 1.   Cytotoxicity of BAMLET on different cell types. The BAMLET concentration needed to kill 50% of the cells (LC50) is estimated using the ATP assay after 4.5 h of incubation. The LC50 is shown as the mean ± SD calculated from at least three independent experiments. Cell types (a–d) and (g–i) are cancer cell lines, whereas cell types (e) and (k–m) are primary cells cultured for a maximum of four passages before the experiment (marked with an asterisk). The fibroblasts (j) are untransformed cells from adult foreskin (marked with a double asterisk). Erythrocytes have been isolated from human blood (marked with a triple asterisk). The LC50 for erythrocytes is estimated from the hemolysis assay.
Cell typeLC50 (mg·mL−1)Relative to HL-60
Suspension cells
 (a) HL-600.03 ± 0.0141
 (b) U9370.06 ± 0.0062
 (c) Jurkat0.09 ± 0.023
 (d) THP10.1 ± 0.0363.3
 (e) PBMC*0.02 ± 0.0030.67
 (f) Erythrocytes***0.06 ± 0.032
Adherent cells
 (g) Skov-30.2 ± 0.046.7
 (h) B16-F00.29 ± 0.0610.3
 (i) CaCo20.55 ± 0.1918.3
 (j) Fibroblasts**0.16 ± 0.0375.3
 (k) Keratinocytes*0.16 ± 0.085.3
 (l) HVSMC*0.75 ± 0.0725
 (m) HUVEC*1.37 ± 0.2646

BAMLET-induced cell death resembles apoptosis or necrosis dependent on the target cells

The characteristics of BAMLET-induced cell death were investigated by flow cytometry using phycoerythrin (PE)-conjugated Annexin V and 7-amino-actinomycin D (7AAD). Annexin V binds to exposed phosphatidylserine (PS), whereas 7AAD is a DNA intercalating dye excluded by viable cells. When investigating cell death with this set-up, viable cells are negative for both Annexin V and 7AAD because the membrane is not permeable to 7AAD, and PS is present only on the inner leaflet of the plasma membrane and thereby unable to bind Annexin V. Cells positive for Annexin V and negative for 7AAD show exposure of PS as a result of loss of membrane asymmetry, although they have intact plasma membranes (a feature seen in apoptotic cells). Cells positive for both Annexin V and 7AAD are indicative of cells with permeable cell membrane (e.g. necrotic cells) [25,26].

The effect of BAMLET treatment on Jurkat and THP1 cells was investigated by flow cytometry and some of the results obtained are illustrated in Fig. 2. For both Jurkat and THP1 cells, the majority of untreated cells were viable and negative for both Annexin V and 7AAD (Jurkat, 88.6%; THP1, 97.9%). The controls included in the experiment comprised treatment with camptothecin, an inducer of apoptosis, or heat treatment to induce necrosis. Camptothecin treatment (Fig. 2B,F) resulted in single-positive cells (Annexin V+, 7AAD; Jurkat, 23.3%; THP1, 70.6%) and heat treatment (Fig. 2C,G) resulted in double-positive cells (Annexin V+, 7AAD+; Jurkat, 99.6%; THP1, 98.6%).

Figure 2.

 BAMLET-induced cell death analyzed by flow cytometry. Flow cytometry analysis for apoptosis and necrosis using Annexin V-PE and 7AAD. (A–D) Jurkat cells. (E–H) THP1 cells. The cells were incubated with BAMLET, camptothecin or NaCl/Pi for 3 h before flow cytometry analysis. (A, E) Untreated cells (NaCl/Pi). (B, F) Positive control for apoptosis obtained by treatment with 10 μm Camptothecin. (C, G) Positive control for necrosis obtained through heat treatment of the cells (56 °C for 30 min). (D, H) BAMLET treated cells (80 μg·mL−1).

As shown in Fig. 2D, 3 h of incubation of Jurkat cells with 80 μg·mL−1 BAMLET caused PS exposure without membrane permeability (Annexin V+, 7AAD) for 28.8% of the cells, indicative of apoptosis. Only 3.25% of the Jurkat cells were double positive (Annexin V+, 7AAD+), indicative of necrosis or late apoptosis. By contrast, under the same experimental conditions, the membranes were rendered permeable (Annexin V+, 7AAD+) for 50% of THP1 cells (Fig. 2H), indicative of necrosis or late apoptosis, with only 5.34% of the cells being single positive (Annexin V+, 7AAD). Cells treated with α-LA resembled NaCl/Pi treated controls (data not shown).

Figure 3.

 BAMLET cytotoxicity towards peripheral blood mononuclear cells. Flow cytometry analysis for apoptosis and necrosis on PBMCs using Annexin V-PE and 7AAD. PBMCs isolated from human blood were incubated with BAMLET, staurosporine or NaCl/Pi for 3 h before flow cytometry analysis. (A) Untreated cells (NaCl/Pi). (B) Positive control for apoptosis obtained by treatment with 1 μm staurosporine. (C) Positive control for necrosis obtained through heat treatment of the cells (56 °C for 30 min). (D) BAMLET treated cells (6 μg·mL−1). (E) BAMLET treated cells (12 μg·mL−1). (F) BAMLET treated cells (24 μg·mL−1).

The effect of BAMLET on PBMCs was also investigated by flow cytometry. The PBMCs analyzed by flow cytometry were a mixture of several cell types: T-cells, B-cells, NK-cells and monocytes. The cells were incubated with different concentrations of BAMLET for 3 h before measurement by flow cytometry. As a result of the incubation on a polystyrene plastic surface, a substantial part of the monocytes was lost during the incubation because of adherence (data not shown). At a BAMLET concentration of 24 μg·mL−1, there was an increase in both single-positive (Annexin V+, 7AAD) and double-positive cells (Annexin V+, 7AAD+) compared to lower concentrations of BAMLET (Fig. 3D,F).

Cytotoxicity of BAMLET and OA

Tolin et al. [10] suggested that a simple mixing of α-LA and OA forms complexes with cytotoxic activity similar to that of BAMLET [10].

We compared the cytotoxicity of α-LA and OA in different combinations: BAMLET, OA and α-LA alone, and by mixing OA and α-LA. For investigation of the cytotoxicity of OA alone and OA mixed with α-LA, the OA dilutions were prepared using two different methods; OA was dissolved in ethanol, diluted in NaCl/Pi and was then either sonicated (OA1) or adjusted to pH 8.0 (OA2) [10] before mixing with α-LA.

As shown in Fig. 4A, based on the calculated initial OA concentrations, BAMLET and the mix of α-LA and pH-adjusted OA (OA2 + α-LA) had very similar cytotoxicities, whereas the mix of α-LA and sonicated OA (OA1 + α-LA) had a lower cytotoxicity. Interestingly, the cytotoxicity of OA prepared with sonication was low compared to pH-adjusted OA. Importantly, OA alone was as active as the mixture of OA and α-LA, regardless of the method used for preparation of the OA dilutions (Fig. 4A).

Figure 4.

 Cytotoxicity of OA and mixtures of α-LA and OA. The cytotoxicity of OA and OA + α-LA was measured on HL-60 cells with the ATP assay after incubation for 4.5 h and compared with the cytotoxicity of BAMLET. OA was dissolved in ethanol and diluted in NaCl/Pi followed by either sonication (OA1) and mix with α-LA (OA1 + α-LA) or by adjustment to pH 8.0 (OA2) and mix with α-LA (OA2 + α-LA). (A) The theoretically calculated concentration of OA is plotted on the x-axis. (B) The actual concentration of OA in the samples was measured using the fatty acid quantification kit and plotted on the x-axis. Viability was calculated from ATP levels as in Fig. 1. Each experiment of three was performed in triplicate, and the range and mean is shown for each point.

To measure the actual amount of OA incubated with the target cells, the fatty acid content of the different samples was measured using a fatty acid quantification kit. The theoretically calculated concentration of OA is shown on the x-axis in Fig. 4A and the experimentally determined OA concentration is shown on the x-axis in Fig. 4B. The actual amount of fatty acid supplied in the cell assays at incubation with OA1 and with the mixture of OA1 and α-LA was almost ten-fold lower than the calculated value. However, for OA2 and OA2 + α-LA, the actual value was much closer to the calculated value. As a result, the dose–response curves for the different combinations of α-LA and OA coincide when the measured OA values are employed (Fig. 4B).

Effect of BAMLET and OA on erythrocytes

To test the activity of BAMLET on erythrocytes, these were isolated from human blood. After 3 h of incubation with BAMLET at 37 °C in the absence of serum, the degree of hemolysis was estimated by measuring the release of hemoglobin (Fig. 5). Almost full lysis was observed at a BAMLET concentration of 175 μg·mL−1, with no lysis at 88 μg·mL−1. For comparison, the activity of OA was also tested on the erythrocytes, and the actual concentration of OA in the samples (OA and BAMLET) was measured (Fig. 5B). OA and BAMLET resulted in almost full lysis at a measured OA concentration of 62 and 51 μm, respectively. Noticeably, α-LA did not lyze the erythrocytes at any of the concentrations tested (Fig. 5A).

Figure 5.

 Hemolysis of BAMLET- and OA-treated human erythrocytes. Erythrocytes isolated from human blood treated with BAMLET, α-LA and OA for 3 h were investigated for hemolysis. (A) Cytotoxicity of α-LA and BAMLET. (B) Cytotoxicity of BAMLET and OA. The content of OA in the BAMLET and OA samples was measured and plotted on the x-axis in (B), whereas the protein content is plotted on the x-axis in (A). The values on the y-axis are given as a percentage of the maximal signal at 405 nm obtained with total lysis of the cells by treatment with Triton X-100. The experiment was performed in duplicate and the range and mean is shown for each point.

Cytotoxicities of OA and BAMLET investigated by flow cytometry

Flow cytometry experiments were carried out to investigate the mode of cell death induced by OA. As shown in Fig. 6, OA alone induced cell death closely resembling that induced by equivalent amounts of OA in the BAMLET preparation. Treatment of Jurkat cells with BAMLET resulted in 26.2% of cells being single positive (Annexin V+, 7AAD) and 4.15% of cells being double positive (Annexin V+, 7AAD+) (Fig. 6A) and OA treatment resulted in 11.5% of cells being single positive (Annexin V+, 7AAD) and 2.92% of cells being double positive (Annexin V+, 7AAD+) (Fig. 6B). For THP1 cells, BAMLET treatment resulted in 4.77% of cells being single positive (Annexin V+, 7AAD) and 13.2% of cells being double positive (Annexin V+, 7AAD+) (Fig. 6C), whereas OA treatment resulted in 4.03% of cells being single positive (Annexin V+, 7AAD) and 16.5% of cells being double positive (Annexin V+, 7AAD+) (Fig. 6D). In summary, Jurkat cells went into a mode of cell death that resembled apoptosis, whereas THP1 cells underwent cell death resembling necrosis, regardless of whether the cells were treated with BAMLET or OA alone.

Figure 6.

 Comparison of the cytotoxicity of OA and BAMLET. Flow cytometry analysis for apoptosis and necrosis using Annexin V-PE and 7AAD. (A, B) Jurkat cells. (C, D) THP1 cells. The cells were incubated with BAMLET or OA for 3 h before flow cytometry analysis. (A) Jurkat cells treated with BAMLET (protein = 57 μg·mL−1, measured OA = 28 μm). (B) Jurkat cells treated with OA (measured OA = 19 μm). (C) THP1 cells treated with BAMLET (protein = 87 μg·mL−1, measured OA = 43 μm). (D) THP1 cells treated with OA (measured OA = 38 μm). Cells treated with NaCl/Pi, Camptothecin and heat-treated cells are not shown but closely resembled the corresponding controls shown in Fig. 2.


According to several studies, HAMLET kills tumor cells and undifferentiated cells but not healthy differentiated cells [1,3,4,6,7,14,15,17,27–32]. The BAMLET used in the present study was prepared by a method that differs slightly from the one used by Svensson et al. [2] for the preparation of HAMLET, and the cytotoxicity of this BAMLET preparation is very similar to that of HAMLET, as well as BAMLET prepared according to the method described by Svensson et al. [2], as examined in Brinkmann et al. [16].

The results obtained in the present study show that both cancer cells and normal primary cells are sensitive to BAMLET-induced killing. Even though the most resistant cells tested were indeed normal primary cells (i.e. endothelial cells), the most sensitive cells were also normal primary cells, namely PBMCs freshly isolated from human blood (Fig. 1 and Table 1). Significantly, BAMLET also caused lysis of erythrocytes (Fig. 5 and Table 1). Of the cell types presented in Table 1, Jurkat, HL-60, U937 and CaCo2 have been investigated in previous studies [1,33] and treatment with HAMLET or MAL (an early, crude preparation, directly from breast milk, with activity similar to that of HAMLET) showed results similar to those obtained in the present study with BAMLET. Håkansson et al. [33] report that MAL is cytotoxic to lymphocytes, granulocytes and foreskin fibroblasts, whereas Svanborg et al. [1] state that MAL treatment does not result in apoptosis in either of these cell types. We find that both fibroblasts and PBMCs are highly sensitive to BAMLET treatment (Table 1).

The steepness of the dose–response curves for BAMLET seen in Fig. 1 emphasizes the compulsory requirement for testing a range of concentrations of BAMLET: as little as a four-fold dilution can result in the change of the observed killing of the target cells from 100% to negligible. In addition, both the incubation time and the number of cells tested influences the observed cytototoxicity of BAMLET [16]. Some of the discrepancies obtained regarding the cytotoxicity of HAMLET and BAMLET may thus be a result of differences in the experimental set-up.

The number of cell types tested in the present study is limited to some degree, and several of the normal cell types reported to be resistant to HAMLET have not been tested. Nevertheless, the results obtained in the present study are at odds with the purposed selectivity for cancer cells over healthy cells because we find that some noncancerous primary cells are even more sensitive than cancer cells.

The cytotoxicity of different α-LA:OA complexes and their constituents has been investigated previously [10,13,18–24,34]. However, it is not clear whether the cytotoxicity of α-LA:OA complexes such as HAMLET and BAMLET resides with the protein or the lipid moiety, or whether a synergistic effect of α-LA and OA is needed.

We did not find α-LA to be cytotoxic in any of the concentrations tested (Fig. 1). When studying the cytotoxicity of OA versus α-LA:OA complexes, it was crucial not to rely on calculated final concentrations of OA but, instead, to directly measure the OA content in the incubation mixture. OA is not soluble in aqueous solutions such as NaCl/Pi or in cell culture medium and, as a result, the amount of OA added to the cells is likely to be lower than the calculated value as a result of a loss of OA to plastic surfaces of eppendorf tubes, pipette tips, etc. Measurements of the OA content in the incubation mixture demonstrated that the actual concentration of OA added to the cells was as much as ten-fold lower than expected. This may explain why earlier tests of OA have shown that OA alone was not cytotoxic compared to the concentrations present in cytotoxic concentrations of HAMLET [2,35] and BAMLET [10].

The solubility of OA is increased by the presence of α-LA [10,13]. The increased solubility of OA when α-LA is present could possibly prevent the loss of OA to plastic surfaces, etc. However, as shown in Fig. 4, according to our measurements, the actual experimental OA content was the same for OA alone compared to the mix of OA and α-LA. This suggests that the presence of α-LA does not entirely prevent the loss of OA to the plastic surfaces. Nevertheless, pH-adjustment of the OA solution to pH 8.0 appears to help prevent the loss of OA compared to sonication of OA, possibly as a result of the formation of oleate during the pH increase, which is more soluble in aqueous solutions than OA.

Other proteins are indeed also present during the experimental procedures. Some of these could potentially adsorb OA and influence the solubility and toxicity. In this context, it is important to note that the first hour of incubation of cells with the various combinations of BAMLET, OA, etc., is performed in the absence of fetal bovine serum. Only then is fetal bovine serum added to all samples, including positive and negative controls.

The very similar cytotoxicity of OA alone compared to the mix of OA and α-LA (Fig. 4A) could indicate that the cytotoxicity observed for the mixture of α-LA and OA was a result of only OA. This is supported by the dose–response curves shown in Fig. 4B, where the activity per OA molecule is indistinguishable, regardless of whether OA is present in a complex with α-LA made on the OA pretreated column (BAMLET), by mixing and incubation for 1 h, or even OA alone without the presence of α-LA.

Non-esterified fatty acids are reported to be cytotoxic [36–38]. However, the sensitivity of different cell lines to treatment with free fatty acids varies with both concentration and the type of fatty acid. For example, 200 μm OA has been shown to induce a loss of membrane integrity and DNA fragmentation in the macrophage cell line J774 [36], the murine melanoma cell line S91 [37] and freshly-isolated human lymphocytes [38]. By contrast, this treatment did not affect murine melanoma cell line B16-F0 or the human melanoma cell lines SK-Mel 23 and SK-Mel 28 [37], indicating that the cytotoxicity of OA is highly dependent on the specific cell type. We found that OA was cytotoxic to HL-60, THP1 and Jurkat cells in concentrations of 40, 38 and 19 μm, respectively, when the amount of OA actually added to the cells was measured (Figs 4 and 6).

The mechanism behind the cytotoxicity of HAMLET and BAMLET is a matter of debate. It has been shown that HAMLET is taken up by cancer cells and is translocated to the nucleus where it colocalizes with histones [29]. Other mechanisms for HAMLET-induced cell death have been suggested, including autophagy [39], depolarization of the mitochondria [40], binding to histones and pertubation of chromatin structure [29], as well as changes in proteasome structure and function [41]. A recent study by Rammer et al. [42] describes how BAMLET induces a lysosomal cell death program involving lysosomal membrane permeabilization and not autophagy, and that BAMLET colocalizes with lysosomes but not with the nucleus or mitochondria [42]. However, apart from these quite specific effects of HAMLET, there is also evidence of a more crude effect connected to membrane interactions and alterations. Both α-LA and HAMLET bind to the membrane, but only HAMLET causes loss of membrane integrity and leakage of lipid membranes [35].

The results shown in Fig. 2 suggest that the mode of cell death induced by BAMLET is dependent on the target cell. BAMLET treatment of Jurkat cells resulted in cell death resembling apoptosis with PS exposure, without the integrity of the plasma membrane being compromised. By contrast, THP1 cells responded to BAMLET treatment with permeabilization of the plasma membrane indicative of necrosis. When we subjected PBMCs, consisting of a mixture of T-cells, B-cells and natural killer cells, to increasing BAMLET concentrations, we found both cells with increased PS exposure without a loss of membrane integrity and cells with a loss of membrane integrity. It should be noted that exposure of PS can be seen in non-apoptotic cells and that double-positive staining (Annexin V+, 7AAD+) can be seen in both necrotic cells and late-stage apoptotic cells [43]. A range of different BAMLET concentrations and incubation times were tested for both Jurkat and THP1 cells (data not shown). Although THP1 cells did not become Annexin V+, 7AAD in either of the set-ups tested, Jurkat cells did become 7AAD+ and Annexin V+ (necrosis or late apoptosis) upon exposure to high concentrations and long incubation times. Another observation was that increasing concentrations of BAMLET resulted in cell lysis, which was manifested as a decreased cell count by flow cytometry analysis.

The induction of cell death resembling apoptosis in Jurkat cells (Fig. 2) and the lysis of erythrocytes (Fig. 5) suggest that the cytotoxicity of BAMLET could consist of two parts (i.e. high concentrations results in lysis of the cells, whereas, at lower concentrations, a more subtle response is seen, resulting in cell death with features resembling apoptosis and/or necrosis).

It is tempting to propose that lysis is caused by OA alone, whereas the complex between α-LA and OA induces a more ordered form of cell death, as seen for the Jurkat cells shown in Fig. 2D. We suggest that OA is responsible for more than the simple lysis effect. The experiments measuring ATP content (Fig. 1 and Table 1) and the hemolysis assay (Fig. 5) cannot distinguish between lysis and a more subtle form of cell death. However, the flow cytometry experiments presented in Fig. 6B show that incubation with OA results in cells with exposure of PS but negative membrane permeability, which could be indicative of apoptosis or other forms of programmed cell death.

In conclusion, the results obtained in the present study suggest that, when testing BAMLET and OA in a setting where the actual concentration of OA is identical, BAMLET and OA induce the same degree and type of cell death (Figs 4–6). In addition, BAMLET is able to induce killing not only in cancer cells, but also in several normal primary cells, including PBMCs, and can cause the lysis of erythrocytes. However, in vivo studies with HAMLET on glioblastoma [4] and bladder cancer [5,7] have been reported to cause cell death in tumor cells but not in healthy surrounding tissue. This indicates that the difference in sensitivity of different cell types may be beneficial in certain settings.

Materials and methods

All buffers were prepared from analytical grade chemicals purchased from Sigma-Aldrich (St Louis, MO, USA) or Merck (Darmstadt, Germany). The cell lines used were: HL-60 [human promyelocytic leukemia-derived cell line; code ACC3, Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ)], Jurkat (human T cell leukemia-derived cell line; code ACC 282, DSMZ), U-937 [human monocytic cell line derived from a histiocytic lymphoma; code CRL-1593.2, American Type Culture Collection (ATCC)], THP1 (acute monocytic leukemia-derived cell line; code TIB-202, ATCC), CaCo2 (colorectal adenocarcinoma-derived cell line; code ACC169, DSMZ), B16-F0 (murine melanoma-derived cell line; code CRL-6322, ATCC) and SKOV-3 (human ovarian adenocarcinoma-derived cell line; code HTB-77, ATCC). In addition, nontransformed fibroblasts (code GM08680; Coriell Cell Repository) and primary cells (PBMCs), keratinocytes, smooth muscle cells and endothelial cells (see below) were used in the present study.

Cell culture

All cells were grown in 75 cm2 culture flasks (90075; Techno Plastic Products, Trasadingen, Switzerland) in a humidified atmosphere with 5% CO2 at 37 °C.

Human vascular smooth muscle cells (HVSMC) and human umbilical vein endothelial cells (HUVEC) were kind gifts from Professor Thomas Ledet, Research Laboratory for Biochemical Pathology, Aarhus University Hospital, Denmark. HVSMCs were obtained from explants of normal aortic tissue from excess donor vasculature at kidney transplantations [44], whereas HUVECs were obtained from collagenase-digested umbilical veins [45].

HUVECs were grown in gelatine coated culture flasks in DMEM growth medium (Invitrogen, Carlsbad, CA, USA) containing 10% normal human serum, NaHCO3 (44 mm), endothelial cell growth factor (40 μg·mL−1, E2759; Sigma-Aldrich) and heparin (0.1 mg·mL−1, H9399; Sigma-Aldrich), whereas HVSMCs were grown in culture flasks with DMEM growth medium with NaHCO3 (28 mm), 5% normal human serum and 5% newborn bovine serum (Sigma-Aldrich). The growth medium for both HUVECs and HVSMCs contained gentamycin sulfate (50 μg·mL−1, 733-1807; Bie & Berntsen, Herlev, Denmark), l-glutamine (2 mm; Sigma-Aldrich), Fungizone (5 μg·mL−1, amphotericin B; Bristol-Myers Squibb, New York, NY, USA) and freshly-added pentrexyl (100 μg·mL−1, ampicillin; Bristol-Myers Squibb). The cells were subcultured after detachment with 0.25% trypsin (w/v) in NaCl/Pi at 37 °C (5 min for HUVECs and 1 min for HVSMCs) and washing in growth media, and were used for experiments between the second and fourth subculturing.

Normal adult human keratinocytes, cultured from excess skin from elective surgery [46], were a kind gift from Professor Knud Kragballe, Department of Dermatology, Aarhus University Hospital, Denmark. The keratinocytes were grown in Keratinocyte SFM Serum Free Medium with supplement [human recombinant epidermal growth factor (5 ng·mL−1) and bovine pituitary extract (50 μg·mL−1)] (17005-075; Invitrogen) and gentamicin (5 μg·mL−1; 15710049; Invitrogen). At 70% confluency, the keratinocytes were detached using 0.025% trypsin (w/v) in NaCl/Pi with 0.53 mm EDTA for 5 min at 37 °C. The trypsin was inactivated with culture medium containing 2% serum and the cells were spun down at 200 g for 7 min. The cells were subcultured at a ratio of 1 : 3. The keratinocytes were used for experiments before the fifth subculturing.

For PBMCs, blood from a healthy donor was drawn into 5 mL vacuum tubes containing sodium citrate (VF054SBCS71, Venosafe; Seelen-Care, Holstebro, Denmark). After diluting the blood 1 : 1 with 0.9% NaCl, 5 mL of the diluted blood was centrifuged on 3 mL Ficoll-Paque PLUS (17-1440-02; GE Healthcare, Milwaukee, WI, USA) at 900 g without brakes for 25 min at room temperature. The interphase containing the PBMCs was harvested and diluted in NaCl/Pi with 0.5% human serum albumin. The cells were pelleted at 300 g for 10 min at room temperature and resuspended in NaCl/Pi with 0.5% human serum albumin. The cells were counted, centrifuged at 200 g for 10 min at room temperature and resuspended in culture medium (RPMI-1640 with glutamine (13018-031; Gibco, Gaithersburg, MD, USA) supplemented with 10 mm Hepes, 23.8 mm NaHCO3, 10% heat-treated (56 °C for 30 min) fetal bovine serum (10270-106; Gibco), Glutamax (2 mm; 35050; Gibco), penicillin (60 μg·mL−1; Benzylpenicillin; Panpharma, Fougères, France) and streptomycin (0.1 mg·mL−1, S6501; Sigma) or frozen in freeze medium consisting of culture medium with 20% fetal bovine serum and 10% dimethylsulfoxide and stored at −135 °C until use.

HL-60, Jurkat, THP1, fibroblasts, CaCo2, B16-F0, U937 and SKOV-3 cells were grown in culture medium composed of RPMI-1640 with glutamine supplemented with 10 mm Hepes, 23.8 mm NaHCO3, 10% heat-treated fetal bovine serum, penicillin (60 μg·mL−1) and streptomycin (0.1 mg·mL−1). Cells growing in suspension (HL-60, Jurkat, THP1, U937) were kept at between 1 × 105 to 1.5 × 106 cells·mL−1. Adherent cells (fibroblasts, CaCo2, B16-F0, SKOV-3) were grown to ∼ 90% confluence and split at subculture ratios between 1 : 2 and 1 : 6 using NaCl/Pi containing 0.53 mm EDTA (B16-F0) or 0.063% trypsin (w/v) and 0.53 mm EDTA (fibroblasts, CaCo2, SKOV-3) to detach the cells (for details, see below).

Preparation of α-LA:OA complexes

α-LA was purified from bovine milk. In brief, α-LA was purified by subjecting the whey fraction of bovine milk to two consecutive DEAE-Sepharose fast flow columns in the presence or absence of Ca2+ on the first and second column, respectively. The α-LA containing fractions were identified by SDS/PAGE followed by western blotting using affinity purified anti-α-LA serum (A10-128P; Bethyl Laboratories, Montgomery, TX, USA). The identification of α-LA was verified by N-terminal amino acid sequencing of the α-LA-containing fractions, carried out by automated Edman degradation (Procise Protein Sequence Model 491; Applied Biosystems, Foster City, CA, USA). The α-LA-containing fractions were pooled, lyophilized and stored at −20 °C until use.

BAMLET was prepared from the purified α-LA. In brief, a DEAE-Sepharose fast flow conditioned with OA was prepared and unbound OA was removed from the column by washing with a high salt buffer (1 m NaCl, 10 mm Tris, pH 8). The purified bovine α-LA dissolved in an EDTA containing buffer (10 mm Tris, 0.08 mm EDTA, pH 8.5) was added to the OA-conditioned column. After removal of excess α-LA with a low salt buffer (0.2 m NaCl, 10 mm Tris, pH 8.5), the α-LA:OA complex was eluted using the high salt buffer (1 m NaCl, 10 mm Tris, pH 8). The eluate was dialyzed against 10 L of NaCl/Pi (138 mm NaCl, 2.7 mm KCl, 8 mm Na2HPO4, 1.5 mm KH2PO4, pH 7.4) and concentrated to ∼ 2 mg·mL−1 using an Amicon stirred cell with an Ultracell Amicon Ultrafiltration Disc membrane (molecular weight cut-off of 10 000 Da) (Millipore, Billerica, MA, USA). The samples were frozen and kept at −20 °C until use. The protein content of the samples was determined by acid hydrolysis followed by quantitative amino acid analysis, or measurement of A280 using a NanoDrop 1000 Spectrophotometer (Thermo Scientific, Waltham, MA, USA) and E1%, 280 nm = 20.1 for bovine α-LA [47]. The concentrations of BAMLET were thus based on the α-LA content.

Measurement of cytotoxic activity

To analyze the cytotoxic activity of the various compounds, cell viability was estimated by quantifying the ATP content of the cells through detection of luminescence (counts·s−1) resulting from an ATP-driven conversion of luciferin to oxiluciferin by luciferase (CellTiter-Glo Luminescent cell viability assay; Promega, Madison, WI, USA).

Adherent cells were detached from the culture flasks. The trypsin activity was stopped by adding 10 mL of culture medium containing 10% heat-treated fetal bovine serum. The cells were washed by centrifugation at 150 g and resuspended in culture medium to 1 × 105 cells·mL−1. One hundred microliters of cell suspension was added to each well of 96-well plates (136101; Nunc, Rochester, NY, USA) and cultured for 24 h. Immediately before the addition of the test samples, the cells were washed twice with 100 μL of serum-free culture medium and 50 μL of serum-free culture medium was added.

The cells in suspension were harvested by centrifugation at 150 g for 5 min, resuspended at 2 × 105 cells·mL−1 in serum-free culture medium and seeded into 96-well plates with 50 μL of cell suspension per well.

Lyophilized α-LA or BAMLET to be used in the assays was dissolved in NaCl/Pi. When investigating the cytotoxicity of OA alone, OA was dissolved in ethanol and diluted in NaCl/Pi. To try to keep the OA in solution, the OA dilutions were then either mixed by vortexing and sonication (called OA1) or by pH adjustement to pH 8.0 (called OA2). For OA1, 4 μL of OA was dissolved in 40 μL of 96% ethanol and transferred to 1 mL of NaCl/Pi (∼ 4 mg·mL−1 of OA) followed by vortexing and sonication (Branson 200; Branson Ultrasonics, Danbury, CT, USA) for 3 × 5 min. Two-fold dilutions in the range 4 mg·mL−1 to 7.8 μg·mL−1 were made in NaCl/Pi. Immediately before addition to the cells, the OA1 dilutions were sonicated for 5 min. For OA2, the samples were prepared as described previously [10]. In brief, OA was dissolved in ethanol to 70 mm followed by dilution to 1 mm in NaCl/Pi and adjustment to pH 8.0 by adding 0.1 m NaOH by drops to the OA/ethanol layer in NaCl/Pi followed by gentle mixing without vortexing.

To investigate the effect of mixing of α-LA and OA, α-LA were mixed with either sonicated OA (OA1) or pH adjusted OA (OA2) followed by incubation in the dark for 1 h.

Fifty microliters of each dilution of α-LA, BAMLET, OA, α-LA+OA or NaCl/Pi was added to the cell-containing wells (final volume, 100 μL·well−1). As a control, 50 μL of NaCl/Pi and 5 μL of lysis solution [Cytotoxicity Detection KitPLUS (LDH), Roche Diagnostics, Basel, Switzerland] were added to cells in 50 μL of culture medium. A control with 50 μL of NaCl/Pi and 50 μL of culture medium (no cells) was also included. In all experiments, the cells and proteins or OA were incubated for 1 h in cell culture medium without fetal bovine serum in a humidified 5% CO2 atmosphere at 37 °C (referred to below as ‘37 °C’). After 1 h, 5 μL of heat inactivated fetal bovine serum was added to all the wells followed by incubation for 3.5 h at 37 °C. The experiments were thus ended after a total incubation time of 4.5 h for the ATP assay in accordance with the manufacturer’s recommendations. For ATP assays, the luminescence was detected on a VICTOR3 1420 Multilabel Counter (Perkin Elmer, Boston, MA, USA).

For the hemolysis assay, human erythrocytes were isolated after centrifugation with Ficoll-Paque PLUS as described for isolation of mononuclear cells above. The isolated erythrocyte population also contained granulocytes. However, the granulocytes did not interfere with the hemolysis assay. The erythrocytes were washed three times in NaCl/Pi by centrifugation for 15 min at 228 g at room temperature. NaCl/Pi was added to the pellet to yield a 20% (v/v) erythrocytes/NaCl/Pi suspension. This suspension was either used directly or stored for 1 day at 4 °C.

Dilution series of BAMLET, α-LA, or OA were made as described above and 50 μL was added in duplicate to a V-bottomed 96-well microtiter plate (249570; Nunc). The 20% erythrocyte suspension was diluted 1 : 20 in NaCl/Pi and, from this suspension, 50 μL was added to each well. Tween-20 was added to 1% (v/v) for total hemolysis. No fetal bovine serum was added in this experiment. The plates were incubated for 3 h at 37 °C and then centrifuged for 5 min at 1500 g at 20 °C. From the supernatant, 50 μL was transferred to a flat-bottom 96-well microtiter plate (269620; Nunc) and A405 was measured. The percentage hemolysis was calculated as: [(A405 of the test sample treated erthrocytes − A405 of buffer treated erythrocytes)/(A405 of Tween-20 treated eryhtrocytes − A405 of buffer treated erythrocytes)] × 100%, and the HC50 values, which represent the concentrations of protein at which 50% hemolysis was observed, were determined.

Flow cytometry

BAMLET-induced cell death was also investigated by flow cytometry using Annexin V and 7AAD modified from the Annexin V/phosphatidylinositol assay [25]. Cells (THP1, Jurkat or PBMC) were harvested by centrifugation at 150 g for 5 min followed by resuspension in serum-free cell culture medium. One milliliter of cell suspension (3.75 × 105 cells·mL−1) was added to the wells of six-well plates (92006; Techno Plastic Products). Five hundred microliters NaCl/Pi, α-LA, OA or BAMLET was added to the appropriate wells followed by incubation at 37 °C. After 1 h, 75 μL of heat-inactivated fetal bovine serum was added. Heat treatment for 30 min at 56 °C was used as a positive control for necrosis. Treatment with 10 μm camptothecin for 4.5 h (Jurkat, THP1) or 1 μm staurosporin treatment for 3 h (PBMC) was used as a positive control for apoptosis. After a total incubation time of 4.5 h (Jurkat, THP1) or 3 h (PBMC), the cells were washed twice in NaCl/Pi and resuspended in Hank’s balanced salt solution. The cells were stained with PE-conjugated Annexin V and 7-AAD (Annexin V-PE Apoptosis Detection kit I, 55976; Becton-Dickinson Biosciences, Franklin Lakes, NJ, USA) using 5 μL of Annexin V or 7AAD per 1 × 105 cells in 100 μL of buffer followed by 15 min of incubation in the dark at room temperature immediately before analysis on a Cytomics FC 500 flow cytometer (FC 500 MCL System with cxp software; Beckman Coulter, Fullerton, CA, USA) with a 488 nm argon laser for excitation. Annexin V-PE was detected in FL-2 by a 575 nm band pass filter, whereas 7AAD was detected in FL4 by a 675 nm band pass filter. Standard compensation using unstained and single-stained cells was carried out using flowjo Flow Cytometry Analysis Software (Tree Star Inc., Ashland, OR, USA). For each sample 7500 cells were analyzed, and the double-negative (Annexin V, 7AAD), single-positive cells (Annexin V+, 7AAD) and the double-positive cells (Annexin V+, 7AAD+) were expressed as a percentage of the total cell number.

Fatty acid measurement

The amount of fatty acid in the solutions added to the cells was quantified using the Free Fatty Acid Quantification kit (K612-100; BioVision, Mountain View, CA, USA) in accordance with the manufacturer’s instructions. In brief, the samples were added to the wells of a 96-well plate in several dilutions. In parallel, a standard curve was prepared with known amounts of palmitic acid. Fatty acid assay buffer was added to all wells to reach a final volume of 50 μL. Acyl CoA synthetase (2 μL) was added to each well followed by incubation at 37 °C for 30 min to allow conversion of the fatty acids to their CoA derivatives. A mix of enzymes, enhancer and a fatty acid probe was added to each well, resulting in the oxidation of the CoA derivatives and the generation of hydrogen peroxide and red color during 30 min of incubation at 37 °C in the dark. The color was detected at 570 nm using an ELISA.


C.R.B. was supported by The Faculty of Science, Aarhus University, through a PhD stipend.