Induction of erythroid differentiation and increased globin mRNA production with furocoumarins and their photoproducts.

Differentiation-therapy is an important approach in the treatment of cancer, as in the case of erythroid induction in chronic myelogenous leukemia. Moreover, an important therapeutic strategy for treating beta-thalassemia and sickle-cell anemia could be the use of drugs able to induce erythroid differentiation and fetal hemoglobin (HbF) accumulation: in fact, the increased production of this type of hemoglobin can reduce the clinical symptoms and the frequency of transfusions. An important class of erythroid differentiating compounds and HbF inducers is composed by DNA-binding chemotherapeutics: however, they are not used in most instances considering their possible devastating side effects. In this contest, we approached the study of erythrodifferentiating properties of furocoumarins. In fact, upon UV-A irradiation, they are able to covalently bind DNA. Thus, the erythrodifferentiation activity of some linear and angular furocoumarins was evaluated in the experimental K562 cellular model system. Quantitative real-time reverse transcription polymerase-chain reaction assay was employed to evaluate the accumulation of different globin mRNAs. The results demonstrated that both linear and angular furocoumarins are strong inducers of erythroid differentiation of K562 cells. From a preliminary screening, we selected the most active compounds and investigated the role of DNA photodamage in their erythroid inducing activity and mechanism of action. Moreover, some cytofluorimetric experiments were carried out to better study cell cycle modifications and the mitochondrial involvement. A further development of the work was carried out studying the erythroid differentiation of photolysis products of these molecules. 5,5'-Dimethylpsoralen photoproducts induced an important increase in γ-globin gene transcription in K562 cells.

Introduction

Furocoumarins are well known natural or synthetic compounds, which derive from a linear (psoralens) or angular (angelicins) condensation of a coumarin with a furan ring. Some of them are employed in PUVA (Psoralen + UVA) therapy for the treatment of autoimmune or hyper-proliferative skin diseases, including psoriasis and vitiligo. PUVA therapy efficacy is due to a combination of psoralen administration and UV-A irradiation. In fact, when activated by UV-A light, furocoumarins induce many biological effects, such as photocycloadditions to DNA, immune system modulation, reactions with proteins, RNA and lipids [1]. Thanks to the development of the photopheresis, the PUVA therapy has amplified its application to some specific tumor forms such as cutaneous T-cell lymphoma [2].

Although the first furocoumarin was introduced in clinical practice as early as 1974 [3], these molecules still draw the attention of the scientific community. In fact, many new potential therapeutic applications for furocoumarins are found. For instance, some psoralen derivatives, such as 8-methoxypsoralen, showed anticonvulsant properties [4]; 4,6,4′-trimethylangelicin demonstrated to be potentially useful in the treatment of cystic fibrosis thanks to its anti-inflammatory activity and its potentiating action on the CFTR membrane channel whose dysfunction causes that disease [5]. Moreover, furocoumarins were found to induce various processes of differentiation. Psoralen is able to stimulate osteoblast differentiation without irradiation as demonstrated by Tang et al. [6], while with or without light activation, many furocoumarins induce erythroid differentiation in different cellular models [7-9]. This latter property can be useful for the treatment of hematologic diseases, such as β-thalassemia: at present, an important therapeutic strategy is the administration of fetal hemoglobin (Hb) inducers to reduce clinical symptoms and blood transfusion requirement [10].

The aim of our study was to evaluate the activity of six linear and five angular furocoumarins on the induction of erythroid differentiation expression of globin genes in the human leukemia cell line K562. These molecules were not fully checked for their potential erythro-differentiation so far. The K562 cell line, isolated from a patient with chronic myelogenous leukemia in blast crisis, is often used as in vitro experimental system for the first screening of new fetal Hb inducers [11]. The K562 cell line presents a low amount of Hb-synthesizing cells under standard cell-growth conditions. After the treatment with suitable inducing compounds, massive erythroid induction occurs, with a clear increase in the expression of human α and γ globin genes and a cytoplasmic accumulation of Hb Portland (ζ2γ2) and Hb Gower 1 (ζ2ε2) [10,12,13].

We then focused our attention on the most active compounds in order to obtain preliminary data about their mechanism of action and about the mode of cellular death induced by them.

A second part of our study was related to the well established observation that after UV-A irradiation, psoralens undergo photolysis with the formation of new species in solution, the so called photooxidation photoproducts (POPs). POPs also present some biological activity: in fact, some papers showed their antileukemic and immunosuppressive effects, which led us to hypothesize their possible biological contribution in PUVA therapy [14,15]. Recently, we also isolated and reported the erythroid differentiation induction by a specific 5-methoxypsoralen photoproduct [16].

Thus, the effect of POPs was also evaluated on the expression of embryo-fetal globin genes in K562 cells by quantititative real-time reverse transcription polymerase-chain reaction assay (RT-qPCR).

Materials and methods

Chemicals and cellular buffers and media

Psoralens and angelicins belong to the collection of the Sciences of Drug Department in Padova University [17-19]. If not specified elsewhere, all chemicals, biological buffers and cellular media were purchased from Sigma–Aldrich.

Irradiation procedure

Two HPW 125 Philips lamps, mainly emitting at 365 nm, were used for irradiation experiments. The spectral irradiance of the source was 4.0 mW cm−2 as measured at the sample level by a Cole-Parmer Instrument Company radiometer (Niles, IL, USA) equipped with a 365-CX sensor.

Cell culture

The human leukemia K562 cells were cultured in a humidified atmosphere of 5% CO2/air in RPMI 1640 medium, supplemented with 10% fetal bovine serum (FBS; Invitrogen), 100 units/mL penicillin and 100 mg/mL streptomycin.

Phototoxicity and benzidine test

Suspensions of 30,000 K562 cells/mL in complete medium were seeded in individual wells of a 24-well tissue culture microtiter plate. The plates were incubated at 37 °C for 24 h prior to the experiments. Stock solutions of furocoumarin derivatives were prepared in methanol and then diluted with Hank’s balanced salt solution (HBSS pH 7.2; the concentration of methanol was always lower than 0.5%) for irradiation experiments. After medium removal, 1 mL of the drug solution was added to each well, incubated at 37 °C for 30 min and then irradiated (1 and 2 J/cm2, which correspond to 4 and 8 min of irradiation at 0.25 J/cm2). After irradiation, the solution was replaced with complete medium and the plates were incubated for 5–7 days. The medium was never changed during this period. Erythroid differentiation was determined by counting blue benzidine-positive cells after suspending the cells in a solution containing 0.2% benzidine in 10% H2O2 and 0.5 M glacial acetic acid [7]. Cell phototoxicity was assessed by the MTT [(3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide)] test 5 days after irradiation [20]. The irreversibility of erythroid differentiation was also assessed: 6 days after irradiation, 10,000 cells were plated in a new plate with fresh medium and after further 4 days, they were counted after benzidine-staining as elsewhere described [21].

Cell cycle analysis

1 × 106 K562 cells were incubated for 24 h and then irradiated (1 J/cm2) in HBSS with or without the test compounds. After 24 h from irradiation, cells were fixed with ice-cooled ethanol (70% v/v), treated overnight with RNAse A (0.1 mg/mL) in phosphate saline buffer and finally stained with propidium iodide (PI, 0.1 mg/mL). Samples were analyzed on a BD FACS Calibur flow cytometer collecting 10,000 events. Results of cell-cycle analysis were examined using WinMDI 2.9 (Windows Multiple Document Interface for Flow Cytometry) [20].

Mitochondrial dysfunction

K562 cells (300,000 cells/mL) were seeded in 24-well microplate and incubated for 24 h prior irradiation. After medium removal, 1 mL of the drug solution was added to each well, incubated at 37 °C for 30 min and then irradiated (1 J/cm2). After irradiation, the solution was replaced with complete medium and the plates were incubated for 24 h. Cells were collected by centrifugation and re-suspended in 1 μM JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazol-carbocyanine) solution in HBSS or in 100 nM NAO (10-N-nonyl acridine orange) solution in RPMI medium. The cytofluorimetric analysis (BD FACS Calibur flow cytometer) was performed collecting green (FL1) and orange (FL2) fluorescence for JC-1 staining and only the green one (FL1) for NAO staining in at least 10,000 events for each sample [22,23].

Preparation of photoproducts

Solutions of derivatives in methanol were irradiated in a quartz cuvette with different UV-A doses (0, 8, 16 and 32 J/cm2). After the irradiation, the solution was lyophilized, suspended in a known volume of methanol and stored at −20 °C. Concentrations of unknown photoproduct mixtures in this paper were expressed as if the initial psoralen was not photodegraded.

mRNA measurements

RNA was isolated from K562 cells and measured by reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR) as described [24] using gene-specific double fluorescence labeled probes in an ABI Prism 7700 Sequence Detection System version 1.7.3 (Applied Biosystems). The following primer and probe sequences were used: α-globin forward primer, 5′-CAC GCG CAC AAG CTT CG-3′; α-globin reverse primer, 5′-AGG GTC ACC AGC AGG CAG T-3′; α-globin probe, 5′-FAM-TGG ACC CGG TCA ACT TCA AGC TCC T-TAMRA-3′; γ-globin forward primer, 5′-TGG CAA GAA GGT GCT GAC TTC-3′; γ-globin reverse primer, 5′-TCA CTC AGC TGG GCA AAG G-3′; γ-globin probe, 5′-FAM-TGG GAG ATG CCA TAA AGC ACC TGG-TAMRA-3′. The kit for quantitative RT-PCR for ζ-globin mRNA and ε-globin mRNA were from Applied Biosystems (ζ-globin mRNA: Hs00923579_m1; ε-globin mRNA: Hs00362216_m1). The fluorescent reporter and the quencher were 6-carboxyfluorescein (FAM) and 6-carboxy-N,N,N′,N′-tetramethylrhodamine (TAMRA), respectively. For real-time PCR, the reference gene was 18S; this probe was fluorescent-labeled with VIC (Applied Biosystems) [24,25].

Statistical analysis

Unless indicated otherwise, results are presented as mean ± SEM. The differences between different treatments were analyzed using the two-sided Student’s t test; p values lower than 0.05 were considered significant (* = p < 0.05).

Results

Furocoumarins induce erythroid differentiation and phototoxicity

The erythroid differentiation of K562 cells was investigated after the treatment with six psoralens and five angelicins, whose structures are described in Fig. 1. K652 cells were irradiated with two UV-A doses (1 and 2 J/cm2) and then erythroid differentiation was measured by benzidine test after 5, 6 and 7 days from irradiation. At the same time, cellular viability was evaluated by MTT, obtaining evidences suggesting antiproliferative effects and phototoxicity. Different concentrations of compounds were employed because of their different phototoxic effects; accordingly, concentrations were chosen to maximize the erythroid effect without extensive reduction of cell viability. Moreover, considering the fact that some angelicins were powerful γ-globin inducers without irradiation [26], the same tests were performed with higher concentration of furocumarins in the absence of irradiation. With the exception of 4,6,4′-trimethylangelicin (4,6,4′-TMA) [26], without irradiation all furocoumarins, when administered at 50 μM, showed a very low capability of causing an increase of benzidine positive cells (lower than 10%) with respect to control (data not shown). On the contrary, after irradiation all tested molecules were able to induce a clear increase of benzidine positive cells, as displayed in Table 1. Table 1 reports the percentages of benzidine positive cells and cellular viability after 6 days after irradiation at the highest concentration for each compound. In general, psoralens induced a higher proportion of erythroid differentiating cells (38.4–78.1%) in comparison to angelicins (24.3–58.7%), and these data confirmed the ones obtained with other furocoumarins [7]. Furthermore, the induction of erythroid differentiation was dependent on the UV-A dose with the exception of some cases in which the antiproliferative effect was a major effect (see for example 5,5′-dimethylpsoralen or 4,6,4′-TMA or 4,4′,5′-trimethylangelicin). In the panel A of Fig. 2, the concentration-dependent increase of the ratio of benzidine positive cells was illustrated for some compounds as examples. Moreover, the panel B of Fig. 2 shows representative pictures of treated cells after benzidine staining: cells irradiated in the presence of 4′,5′-dimethylpsoralen (4′,5′-DMP) clearly were blue-colored1 and became larger with respect to control (this effect is not unusual within already reported inducers of erythroid differentiation, such as cytosine arabinoside [10]). Further experiments were carried out to determine whether the induced erythroid differentiation was reversible. To this aim, 6 days after irradiation (1 J/cm2), K562 cells were incubated for additional 4 days with fresh medium, and the benzidine test was performed at this point. After 10 days, the percentage of erythroid differentiation did not present any significant variation in comparison to the ones of the sixth day, with the exception of 5,5′-DMP, which exhibited a significant increase in the proportion of benzidine-positive cells (from 50.6 ± 5.0 to 66.8 ± 2.0). Considering the fact that in erythroid-induced K562 cells the growth efficiency is lower (see Tables 1 and 2), these evidences support the concept that benzidine-negative cells at day 6 still can differentiate even in the absence of irradiated compounds in the medium (this “commitment-like” effect is present in several inducers of K562 cell differentiation). In any case, the data suggest that the induced differentiation observed at day 6 is irreversible.

Fig. 1

Furocoumarin structures.

Table 1

Differentiation and cellular viability after 6 days from irradiation with furocoumarins.

	% Benzidine positive cellsa		% Cellular viabilityb
	1 J/cm2	2 J/cm2	1 J/cm2	2 J/cm2
Control	1.5 ± 0.5	1.8 ± 0.4	98.0 ± 2.5	97.5 ± 1.4
Psoralens
5′-MP 1 μM	58.7 ± 1.9	78.1 ± 5.1	65.5 ± 2.5	45.7 ± 3.6
4′,5′-DMP 5 μM	60.7 ± 1.1	72.2 ± 5.0	73.7 ± 3.7	54.0 ± 2.5
5,5′-DMP 1 μM	50.6 ± 5.0	44.5 ± 2.7	48.8 ± 6.7	19.5 ± 2.6
8,4′-DMP 1 μM	38.4 ± 1.9	42.7 ± 1.7	60.9 ± 3.7	35.8 ± 3.9
3,4,8,4′-TMP 1 μM	43.5 ± 7.1	38.7 ± 1.7	39.4 ± 5.7	22.5 ± 1.9
5′-Br-4,4′-DMP 1 μM	40.2 ± 4.3	46.7 ± 4.5	88.7 ± 11.2	42.6 ± 3.7
Angelicins
5′-MA 5 μM	24.3 ± 3.2	28.9 ± 4.8	97.7 ± 3.2	99.9 ± 4.5
3,4,4′-TMA 1 μM	30.2 ± 2.2	39.4 ± 2.0	102.2 ± 11.2	57.8 ± 5.4
4,6,4′-TMA 0.5 μM	58.7 ± 4.0	40.7 ± 4.3	60.9 ± 4.6	15.9 ± 3.4
4,4′,5′-TMA 5 μM	47.8 ± 6.9	18.9 ± 2.5	31.1 ± 2.5	9.2 ± 2.2
6,4′,5′-TMA 1 μM	33.5 ± 4.0	42.4 ± 1.9	74.3 ± 8.2	44.4 ± 5.5

Data are expressed as media ± S.E.M. of at least four different independent experiments.

Data are obtained by MTT and are expressed as media ± S.E.M. of at least four different independent experiments.

Fig. 2

Erythroid differentiation photoinduced by furocoumarins. Panel A. Percentage of benzidine positive cells after 6 days from irradiation (1 J/cm2) of K562 cells in the presence of the indicated concentrations of 5′-MP, 4′,5′-DMP, 5,5′-DMP and 4,6,4′-TMA. Results are expressed as media ± S.E.M. of at least 4 different experiments. Panel B. Representative pictures (20×) of benzidine stained K562 cells 6 days after irradiation (1 J/cm2) by light microscope: control in the left picture, 5 μM 4′,5′-DMP in the right one.

Table 2

Percentage of K562 cells in each cell cycle phase after 24 h from irradiation (1 J/cm2) in the presence of the indicated concentration of furocoumarins.

	G1a	S	G2-M	subG1
Control	43.9	25.1	23.0	8.0
5′-MP 1b	14.0	12.0	63.2	10.8
5′-MP 0.5	22.8	14.6	50.5	12.0
4′,5′-DMP 5	7.3	15.1	59.5	18.4
4′,5′-DMP 2.5	8.9	9.3	70.8	11.0
5,5′-DMP 1	9.2	24.0	35.2	31.6
5,5′-DMP 0.5	7.1	14.3	63.6	15.0
4,6,4′-TMA 0.5	9.4	12.6	64.0	14.0
4,6,4’-TMA 0.25	17.2	17.6	44.2	21.0

Percentage of cells.

μM.

Since 5′-methylpsoralen (5′-MP), 4′,5′-DMP and 5,5′-dimethylpsoralen (5,5′-DMP) for psoralens and 4,6,4′-TMA for angelicins were the most active compounds, further experimental activity was carried out with these molecules. Moreover, the lower UV-A (1 J/cm2) dose was chosen to minimize the phototoxic effect.

Erythroid differentiation is linked to DNA damage

The mechanism by which erythroid differentiation induced by furocoumarin takes place is still unknown. However, the DNA photobinding is considered the main effect for the photoantiproliferative activity of the PUVA therapy. Thus, some preliminary experiments were carried out to verify whether furocoumarin DNA photodamage could be involved also in the erythroid differentiation process. K562 cells were irradiated in the presence of the tested compounds and of the inhibitors of some phosphoinositide kinase-related kinases, such as DNA-dependent protein kinase (DNA-PK), ataxia telangiectasia mutated (ATM) and the ataxia- and Rad3-related protein (ATR), which can be activated after different kinds of DNA damage [27]. In particular, wortmannin was used as inhibitor of the catalytic subunit of the PI3-kinase family of enzymes [28], and caffeine as inhibitor of ATM and ATR but not of DNA-PK [29]. Cell viability was not affected by the presence of these two inhibitors (data not shown). As it can be observed in Fig. 3, the amount of benzidine positive cells was significantly reduced, even if not completely abolished, for all tested compounds, when the irradiation was carried out in the presence of those inhibitors. Thus, the processes activated by DNA damage could be involved, at least in part, in the erythroid differentiation process.

Fig. 3

Involvement of DNA photodamage in erythroid differentiation by furocoumarins. Percentage of benzidine positive cells after 6 days from irradiation from irradiation (1 J/cm2) of K562 cells in the presence of 1 μM 5′-MP, 5 μM 4′,5′-DMP, 1 μM 5,5′-DMP and 0.5 μM 4,6,4′-TMA. The irradiation was carried out in the presence or absence of 10 μM Wortmannin (upper panel) or of 1 mM caffeine (lower panel). Results are expressed as media ± S.E.M. of at least four different experiments. C = irradiated control.

Furocoumarin induce increased expression of globin genes

The effects of furocoumarins on the expression of human globin genes were determined by RT-qPCR analysis using probes amplifying the α-like α-globin and ζ-globin and the β-like ε-globin and γ-globin mRNA sequences. Effects on production of β-globin mRNA were not analyzed, since it is well known that K562 cells do not efficiently transcribe the β-globin genes [10,30]. In Fig. 4, globin mRNA expression for 4′,5′-DMP and 4,6,6′-TMA is presented; these two molecules were selected as an example for linear (4′,5′-DMP) and angular (4,6,4′-TMA) most active furocoumarins in inducing erythroid differentiation (Table 1). The results were reproducible and demonstrate that: (a) when K562 cells were irradiated (1 J/cm2) in the absence of the compounds, only minor effects were observed on accumulation of globin mRNAs; (b) the induction of globin gene expression is not enhanced by low concentrations of furocoumarins without irradiation; however, (c) when irradiation was carried out in the presence of increasing concentrations of the two furocoumarins, a dose-dependent increase in globin mRNA expression was observed. The level of induction was found to be dose-dependent, all the analyzed globin mRNAs were clearly induced, the level of induction was dramatic for α-globin, ζ-globin and γ-globin mRNA sequences, but clearly evident also for ε-globin mRNA. When the experiment was repeated (n = 3) using the highest furocoumarin concentration reproducible results were observed, and if the results were compared to reference K562 cells treated with a control HbF inducer, this induction level was higher than the most effective K562 erythroid inducer available, 1-octylthymine [30]. In fact the induction of ζ-globin mRNA was 48.5-fold ± 8.5 for 4′,5′-DMP, 64.6-fold ± 8.2 for 4,6,4′-TMA and 37-fold ± 6.8 for 1-octylthymine (data not shown and Ref. [30]).

Fig. 4

Increase in α-, γ-, ε- and ζ-globin mRNA content in K562 cells irradiated (1 J/cm2) or incubated (dark) in the presence of 4′,5′-DMP (upper panel) and 4,6,4′-TMA (lower panel) at the indicated concentrations. Expression of globin genes was assayed by quantitative reverse transcriptase polymerase-chain reaction (qRT-PCR) after 6 days from the irradiation/incubation. CI = irradiated control.

Furocoumarins induce a block in G2-M phase and cause cell death by apoptosis

To further study the effects of furocoumarins on cell proliferation, a cell cycle analysis was carried out after 24 h from the irradiation of K562 in the presence of two different concentrations of the compounds (Fig. 5). This test is based on the fact that each cell cycle phase presents a different DNA content, which was quantified by propidium iodide (PI) staining. The irradiation of K562 with all tested furocoumarins caused a reduction of G1 phase together with a clear accumulation of cells in G2-M phase (see Table 2). This G2-M block was consistent with the effect of other furocoumarins in the same cell line [7]. Moreover, indications of cell death by apoptosis were detected as DNA fragments in sub-G1 phase.

Fig. 5

Representative histograms of cell cycles after irradiation of K562 cells in the presence of furocoumarins. Cells were irradiated (1 J/cm2) at the indicated concentrations of 5′-MP, 4′,5′-DMP, 5,5′-DMP and 4,6,4′-TMA and after 24 h of incubation were labeled with propidium iodide and analyzed.

Furocoumarins induce cell death and erythroid differentiation with the involvement of mitochondria

As furocoumarins are known to photoinduce apoptosis with the involvement of mitochondria, the role of these organelles was evaluated with two different flow cytometry tests [31]. Impairment in mitochondrial function is an early event in the executive phase of programmed cell death in different cell types and appears as the consequence of a preliminary reduction of the mitochondrial transmembrane potential (ΔΨM). The lipophilic cation JC-1 was used to monitor the changes in ΔΨM induced by the tested compounds in combination with UV-A irradiation. Another consequence of mitochondrial dysfunction is the production of reactive oxygen species which oxidized the mitochondrial phospholipid cardiolipin (CL). CL oxidation was monitored by staining irradiated cells with N-nonyl acridine orange (NAO) as described in Section 2.3.3.

A concentration-dependent increase of the percentage of cells with a collapsed ΔΨM can be observed after JC-1 staining (Fig. 6, upper panel): this may be an indication of the opening of the mitochondrial mega-channels also called the permeability transition pores (PTPs).

Fig. 6

Mitochondrial dysfunction photoinduced by furocoumarins. Percentage of cells with loss of mitochondrial membrane potential (upper panel) and with oxidized cardiolipin (lower panel) after 24 h from the irradiation (1 J/cm2) in the presence of the indicated concentrations of 5′-MP, 4′,5′-DMP, 5,5′-DMP and 4,6,4′-TMA. C = irradiated control. Data are expressed as mean ± S.E.M. of three independent experiments.

NAO interacts stoichiometrically with intact non-oxidized cardiolipin. In good agreement with the collapsed mitochondrial potentials, treated cells showed an increase in the oxidized CL (Fig. 6, lower panel).

Moreover, we decided to evaluate a correlation between erythroid differentiation and mitochondrial impairment. In particular, the involvement of the mitochondrial pathway via activation of caspase-3 and caspase-9 was evaluated; to this aim, K562 cells were irradiated in the presence of the pancaspase inhibitor z-VAD.fmk and then benzidine test was performed. As shown in Fig. 7, z-VAD.fmk suppressed erythroid differentiation induced by all furocumarins.

Fig. 7

Involvement of mitochondrial pathway in erythroid differentiation by furocoumarins. Percentage of benzidine positive cells after 6 days from irradiation from irradiation (1 J/cm2) of K562 cells in the presence of 1 μM 5′-MP, 5 μM 4′,5′-DMP, 1 μM 5,5′-DMP and 0.5 μM 4,6,4′-TMA. The irradiation was carried out in the presence or absence of 50 μM z-VAD.fmk. Results are expressed as media ± S.E.M. of at least three different experiments. C = irradiated control.

5,5′-DMP photoproducts induce erythroid differentiation

We also studied the possible erythroid differentiation activity of irradiated mixtures of some tested furocoumarins. These compounds were 5′-MP, 4′,5′-DMP and 5,5′-DMP and were chosen on the basis of their higher sensitivity to UV-A photodegradation (followed by UV–vis spectroscopy- data not shown).

After their irradiation in methanol solution with different UV-A doses (0, 8, 16 and 32 J/cm2), psoralens were concentrated by solvent evaporation and then resuspended in methanol. The erythroid differentiation of photoproducts was investigated by benzidine test incubating K562 with psoralen irradiated mixtures at two different concentrations (50 and 200 μM) for 5–7 days. Cell growth was also evaluated using the MTT assay after 6 days of treatment (Table 3). After 6 days of incubation, cells treated with 50 μM pre-irradiated mixtures did not show a clear increase of benzidine positive cells (Fig. 8, upper panel) nor a decrease in cellular viability in comparison to control (Table 3); on the contrary, using the higher concentration, an induction of erythroid differentiation (26–36% benzidine positive cells) (Fig. 8, lower panel) together with a reduction of cellular viability was observed only with 5,5′-DMP (Table 3); the other POP mixtures exhibited low activity or were inactive. The irreversibility of the erythroid differentiation induction by 5,5′-DMP photoproduct mixtures was also assessed. The first 6 days of treatment were sufficient for K562 cells to differentiate irreversibly since during additional 4 days of culturing in the absence of the inducer of washed cells, the population of benzidine-positive cells still increased (from 26.3 ± 3.1 to 44.3 ± 2.2 in the case of 8 J and from 35.1 ± 2.0 to 40.5 ± 1.1 in the case of 16 J).

Table 3

Percentage of cellular viability after incubation of 6 days of K562 with pre-irradiated mixtures of 5′-MP, 4′,5′-DMP and 5,5′-DMP.

	0 J/cm2	8 J/cm2	16 J/cm2	32 J/cm2
5′-MP 50a	95.0 ± 4.5	93.0 ± 3.5	92.6 ± 4.8	89.8 ± 3.5
4′,5′-DMP 50	93.8 ± 2.3	94.5 ± 5.0	94.0 ± 4.6	92.9 ± 0.4
5,5′-DMP 50	98.7 ± 5.8	94.8 ± 6.5	93.4 ± 3.5	94.3 ± 4.7
5′-MP 200	88.9 ± 7.8	80.2 ± 3.6	84.5 ± 3.9	81.9 ± 4.9
4′,5′-DMP 200	86.0 ± 2.4	86.9 ± 6.2	96.8 ± 6.3	83.3 ± 2.8
5,5′-DMP 200	99.8 ± 2.9	49.5 ± 2.3	22.4 ± 2.2	22.1 ± 1.9

μM.

Fig. 8

Erythroid differentiation induced by furocoumarin photoproducts. Percentage of benzidine positive cells after 6 days of K562 cell incubation in the presence of irradiated (0, 8, 16 and 32 J/cm2) mixtures of 5′-MP, 4′,5′-DMP and 5,5′-DMP: in the upper panel, the concentration is 50 μM while in the lower panel it is 200 μM. Results are expressed as media ± S.E.M. of at least three different experiments.

Furocoumarin photoproducts induce increased expression of globin genes

RT-qPCR was also employed to quantify the expression of globin mRNA following treatment of K562 cells with 5,5′-DMP photoproducts. There is a clear positive relationship between UV-A doses used to obtain the photoproducts and the extent of increased globin mRNAs in respect to control K562 cells (Fig. 9). As far as a possible differential activity of furocoumarin photoproducts on globin gene expression is concerned, the data clearly indicate that accumulation of both the α-like α-globin mRNA and ζ-globin mRNA are strongly induced. On the contrary, among the β-like ε-globin and γ-globin mRNAs, the γ-globin mRNA is preferentially induced. This conclusion is well in agreement with the data shown in Fig. 4 and concerning the effects of other furocoumarins on globin gene expression in irradiated K562 cells.

Fig. 9

Increase in α-, γ-, ε- and ζ-globin mRNA content in K562 cells incubated in the presence of pre-irradiated mixtures of 5,5′-DMP (0, 8 and 16 J/cm2) for 6 days. The final concentration was 200 μM. Expression of globin genes was assayed by quantitative reverse transcriptase polymerase-chain reaction (qRT-PCR).

Discussion

In this study, we reported the antiproliferative effects and the inducing activity on erythroid differentiation of some psoralen and angelicin analogs in the human chronic myelogenous leukemia K562 cell line. Some of us previously demonstrated that furocoumarins, in combination with UV-A, present the capability of inducing erythroid differentiation like other DNA binders. Thus, we decided to continue our research evaluating new derivatives, some of them chosen on the basis of some considerations about the structure–activity relationship. For instance, we focused our attention on angelicin with trimethylation as this substitution seemed to be successful for erythroid differentiation [26]. In fact, trimethylangelicins resulted to induce higher percentages of benzidine positive cells with respect to 5′-MA (see Table 1). In the case of psoralens, our aim was also to verify the role of the substitution of furan ring, considering preliminary data demonstrating that monomethylation on furan leads to a very active compound and confirmed the higher inducible power of methylpsoralens [7]. The dimethylation involving one or both furan positions led to very interesting compounds, especially when the substitution on position 8 is avoided. We also decided to evaluate new substitutions, as tetramethylation or the introduction of an halogen, but they do not seem to increase erythroid induction activity (see Table 1).

Interestingly, the most active compounds were able to induce a clear and important increase of globin mRNA expression which was much higher than that reported elsewhere for 5-methoxypsoralen [8] (Fig. 4). It should be underlined that the level of induction reached in these experimental conditions is even higher than that exhibited by the most powerful inducer described [30].

Moreover, since the mechanism of erythroid differentiation mediated by furocoumarins (in the presence or absence of UV-A exposure) is not well understood, first of all, some preliminary analyses were performed to investigate the role of DNA damage. Central to the DNA damage response are the ATM (ataxia-telangiectasia mutated), ATR (ataxia-telangiectasia and Rad3-related) and DNA-dependent protein kinases that modulate cell cycle progression, DNA repair, and sometimes, apoptosis. We observed a significant reduction of the levels of erythroid differentiation induced by furocoumarins when irradiation was performed in the presence of inhibitors of these kinases (see Fig. 3): this suggests that furocoumarin-mediated erythroid differentiation is at least partially mediated by the DNA damage activated proteins. A role of DNA damage can be also found by the analysis of cell cycle of irradiated cells in the presence of furocoumarins since a clear block in G2-M phase was observed (see Fig. 5 and Table 2). Furthermore, cell cycle studies demonstrated that furocoumarins plus UV-A induced a certain degree of cell death (see Fig. 5) by apoptosis thanks to the presence of a percentage of cells with a lower DNA content than G1 phase. The role of mitochondria in cell death was also demonstrated (Fig. 6). We also evaluated a possible role of mitochondrial dysfunction and of apoptosis in erythroid differentiation and we observed a clear suppression of the proportion of benzidine positive cells after mitochondrial pathway inhibition. These data indicate that erythroid differentiation may be a consequence of a stress response in which mitochondrial and DNA damage signaling are involved.

In this report, we also aimed at studying a possible role of photodegradation products in furocoumarin activity. The most interesting photoproducts mixtures were those obtained with 5,5′-DMP: in fact, the efficiency of these photoproducts in inducing increase of globin mRNA content is dramatic and much higher than those exhibited by other inducers of K562 erythroid differentiation, such as cytosine arabinoside, butyric acid, mithramycin. This supports the concept that this strategy might be of some interest in the design of novel agents against chronic myelogenous leukemia to be used in differentiation therapy.

The design and production of antiproliferative molecules targeting the K562 cell system might be of great interest for the development of cocktails exhibiting applications in the treatment of chronic myelogenous leukemia. For instance smenospongine [32], crambescidin 800 [33] and doxorubicin derivatives [21] were reported as molecules of possible interest for inhibiting of CML cell growth, stimulating terminal differentiation along the erythroid program. Some molecules, such as Pivanex (an HDAC inhibitor) [34] and a morpholine derivative of doxorubicin [35], are synergistic with the most common anti-CML agents, STI571 (Imatinib). In addition to synergistic effects, molecules inducing differentiation might be of great interest for treatment of Imatinib mesylate-resistant human CML cell lines, as recently demonstrated for the phytoalexin resveratrol [36].

As far as a possible differential activity of furocoumarin photoproducts on globin gene expression is concerned, the preferential effects on γ-globin mRNA might be also of interest for the development of novel HbF inducers in thalassemia. At present, one of the most promising novel approaches for the clinical management of β-thalassaemia is the treatment of patients with chemical inducers of endogenous HbF. On the basis of recent achievements obtained in this research field, several studies focusing on the mechanisms regulating reactivation of HbF production in humans have been reported. Relevant to these issues are studies showing that there is a strong negative correlation between HbF levels and morbidities.

In conclusion, the results presented in this paper demonstrate that the effects on erythroid differentiation of UV-A exposure of K562 cells in the presence of linear and angular psoralens are part of a general phenomenon. The differentiating activity of these compounds in the presence of UV-A irradiation was associated with a dramatic induction of accumulation of the α-like α-globin and ζ-globin mRNA and the β-like ε-globin and γ-globin mRNA sequences. Of particular interest is our finding that erythroid induction and accumulation of γ-globin mRNA can be also obtained with psoralen plus UVA induced photolysis products. It will be of interest to identify and characterize the active products involved.