The psoralen family of compounds was originally discovered as natural products with the ability to cause intense photosensitization reactions (exacerbated sunburn) in human skin. Under controlled conditions, they became a major tool in the dermatologic armory for the treatment of several skin disorders. Simultaneously, it was found that they could also be used to unravel the details of molecular contacts between DNA and proteins, as well as between DNA strands. Among the wide variety of compounds that can be photoactivated, the photochemistry of psoralen is probably the most studied and the best understood.
1. Properties
Psoralens are members of the furocoumarin family. They are tricyclic structures (Fig. 1) with an extended aromatic system that gives rise to strong ultraviolet absorption bands near 220, 250 and 300 nm (Fig. 2) (1). Also shown in Figure 2 is the output of the ultraviolet lamps typically used to activate psoralen. The spectroscopic properties of representative psoralens are listed in Table 1. After absorption of a photon by a molecule in the ground state, an electron can be promoted to an excited singlet state. It may then return to the ground state by the emission of a photon or by the release of energy in the form of heat (radiationless collisional deactivation) or light (fluorescence or phosphorescence). Excited state forms of the molecule can react in different ways: photo-addition, photo-dimerization, and photo-oxidation of nearby moieties (nucleic acids, proteins, or membranes). Alternatively, energy may be transferred to molecular oxygen, leading to highly reactive singlet oxygen (^2) (2). Although many molecules may absorb UV light, photochemistry rarely occurs.
Furthermore, it has been shown that psoralen photochemistry in a solution can differ drastically from that within the confines of a cell.
Figure 1. The chemical structures of psoralen (a) and angelicin (b).
Figure 2. The UV absorption spectrum
or 8-methoxypsoralen (8-MOP; inset) and the typical output of UVA lamps used to photoactivate psoralens
Table 1. Spectroscopic Properties of Psoralens(l)
|
Absorbance |
Fluorescence |
|||
|
Compound |
lmax (nm) |
Extinction Coefficient (M-1cm-1) |
lmax (nm) |
Quantum Yield |
|
Psoralen |
295 |
10,600 |
450 |
0.019 |
|
8-methoxy |
303 |
11,500 |
495 |
0.013 |
|
5-methoxy |
308 |
14,500 |
- |
- |
|
3-carbethoxy |
320 |
11,500 |
490 |
0.025 |
|
4,5′,8- |
293 |
7,950 |
460 |
|
|
Trimethylpsoralen |
||||
|
4′-aminomethyl |
298 |
10,000 |
450 |
|
|
Angelicin |
300 |
9,350 |
||
|
3-methyl |
300 |
10,900 |
||
|
5-methyl |
305 |
11,410 |
||
|
4,5′-dimethyl |
298 |
9,350 |
||
|
4,6,4′-trimethyl |
298 |
8,650 |
||
2. Photoadduct Formation
2.1. Nucleic Acids
The best understood psoralen photochemistry is that which occurs when psoralen is intercalated between DNA base pairs. A 2 + 2 photocycloadditon product is formed between pyrimidine (primarily thymidine at 5′-TpA sites) when the intercalation complex absorbs UV radiation (3). Furan side monoadducts can absorb additional long-wavelength ultraviolet radiation (UVA, 320400 nm) and, if located at a proper site, undergo a second photoaddition reaction, thereby forming an interstrand crosslink. The psoralen isomers (angelicins) are angular psoralens (Fig. 1) that have been studied because of their ability to form only mono-addition photoadducts on one DNA strand (4). Due to its angular structure, the angelicin 4′,5′-monoadduct cannot form crosslinks.
2.2. Proteins, Lipids
Although DNA psoralen photochemistry has attracted the most attention, psoralens also react with proteins as well as other cell constituents (5). In studies of subcellular fractions of rat epidermis after treatment with 8-methoxypsoralen (8-MOP) and UVA light, it was found that 17% of the 8-MOP was bound to DNA, whereas a substantial amount was bound to proteins (57%) and lipids (26%) (6). Proteins photomodification is known to alter the ability of endopeptidases to recognize the usual sites of incision (7). Much greater doses of 8-MOP and UVA are required to affect proteins than doses necessary to damage DNA. Even low levels of protein modifications, however, could participate in the concerted cellular events that lead to clinical effects. Recently, the first psoralen-amino acid photoadduct was described (8).
The ability of 8-MOP and trimethylpsoralen (TMP) to react in an oxygen-independent way with unsaturated fatty acids was first shown by Kittler et al. (9). Specht et al. (10, 11) and Caffieri et al. (12) independently isolated the lipid-psoralen adducts. NMR studies indicated that these adducts resulted from cyclo-additions of the 3,4-double bond of the psoralen to the central double bond of the fatty acid. Recently, Dall’Acqua et al. (13) suggested that these adducts have structures similar to diacylglycerol (DAG) and proposed that they could play a role in cell signal transduction events. In more recent studies, it has been observed that some photoactivated psoralens can crosslink DNA and proteins (14).
3. Impact on Cells
Over the nearly three decades that psoralen photochemotherapy has been employed by dermatologists, its efficacy has been explained on the basis of cytotoxicity, immune modulation, and/or apoptosis, somewhat depending on what has been in vogue. The initial observation of psoralen crosslinking led to the potent antiproliferative/cytotoxic paradigm that persisted until the evidence arose for gene induction and immune modulation.
The formation of psoralen-DNA adducts has been studied extensively in different cell types under in vitro conditions. [ H]8-MOP has been used to quantify the number and type of adducts formed, using HPLC and radioactivity incorporation analysis (15). In Figure 3, the number of photoadducts induced in a variety of cells, and treated in vitro is shown as a function of the combined dose of psoralen and light (8-MOP concentration in ng/ml multiplied by UVA dose in J/cm ). Adduct formation is directly proportional to the dose ( r = 1.0). The effect of 8-MOP and UVA in response to mitogenic stimulation by phytohemagglutinin (PHA) and exclusion of trypan blue has been studied in human lymphocytes. The former response is abrogated at relatively low doses of 8-MOP and UVA, with a complete block at a combined dose greater than ~3050ng J/ml cm (Fig. 4). Much greater doses are required to affect membrane integrity; hence, a distinctly different curve is seen for trypan blue exclusion (16). The antiproliferative effect of 8-MOP and UVA on murine keratinocytes was studied in vitro by Tokura et al. (17), who correlated the number of adducts with inhibition of [ H]thymidine incorporation after PHA stimulation. A minimum number of 0.9 adducts per million bases was required to inhibit DNA synthesis. Moreover, it was shown that the photoadduct removal rate in keratinocytes was dependent on the extracellular calcium concentration, being reduced in low-calcium media (17).
Figure 3. Dose dependence for 8-MOP photoadduct formation in cells treated with 8-MOP and UVA light. Adduct numbers per megabase pair (mbp) are plotted vs combined dose of 8-MOP (ng/ml) and UVA light (J/cm2). A, human lymphocytes; O. murine keratinocytes;, bovine aorta smooth muscle cells; /, Jurkat cells.
Figure 4. Dose dependence for cellular responses to psoralen and light. linear fit of the photoadduct data from Figure 3. The biological effects of these adducts are most significant at the lower modest doses (1 to 100), over which the response to PHA stimulation (I_ as measured by tritiated thymidine incorporation) is completely abrogated by a combined dose of 100. The impact on cell viability (as measured by trypan blue exclusion (heavy dash) measured three days after treatment) only occurs at much higher doses, reflecting the lower efficiency of 8-MOP photochemistry with proteins and lipids. Superimposed on these results are data (solid lines) illustrating the induction of class IMHC molecules on murine lymphoma cells, either total (/), (48). or surface-bound (O) ((49),(50)). Finally, the induction of apoptosis in human lymphocytes is shown (^) (22). Taken together, these data illustrate that any selected property is likely to have its own characteristic dose dependence. Furthermore, these different effects are likely to reflect the combined effects of 8-MOP/UVA at different cellular sites involving a range of biomolecules. The range of these dose-dependent effects may be represented by response to mitogen (perhaps best reflecting the impact of DNA photoadducts) and cell viability (reflecting the effects of photomodification of lipids and proteins). Regarding the latter, it is interesting to note that the impact on viability takes several days to develop. If viability is measured immediately following the 8-MOP/UVA treatment, there appears to be little difference between cell populations treated with different doses of 8-MOP and UVA.
Extensive DNA repair studies have been carried out in bacterial systems as well as in mammalian cells. Several repair mechanisms have been described, including excision repair (the most common), postreplication repair, and photoreactivation (18). (See Photolyase/Photoreactivation.) It has often been assumed that crosslinks would not be repaired, but an excision-recombination mechanism has been proposed to account for crosslink repair in bacteria and bacteriophage (19, 20).
Repair mechanisms in mammalian cells have not been described as completely, but it has been shown in various human and murine cells that DNA-psoralen adducts induced by 8-MOP and UVA can be removed (see Table 1). Freshly isolated human lymphocytes do not repair 8-MOP photoadducts when treated with 100 ng/ml 8-MOP and 1 J/cm of light, but do remove photoadducts after treatment with only 10 ng/ml 8-MOP. In PHA-stimulated lymphocytes, relatively more adducts are formed, but more of these are repaired in 24 hours (16). Repair in murine keratinocytes seems to be more efficient than in lymphocytes, which may be due to the increased metabolic activity of keratinocytes.
4. Apoptosis
Although modest doses of 8-MOP and UVA are antiproliferative, higher doses can be toxic to cells. Thus, depending on the dose employed, the resulting effects can range from cytostatic to apoptotic to necrotic. Marks et al. (21) demonstrated the induction of apoptosis in human lymphocytes by using 300 ng/ml 8-MOP and 10 J/cm UVA light (21). This dose is much greater, however, than the therapeutic dose used in patients (100- 200 ng/ml 8-MOP and 1-2 J/cm2 UVA). More recently, Vowels et al. (22) described the induction of apoptosis with more therapeutically relevant doses of 8-MOP and UVA (see Fig. 3).
5. PUVA Therapy for Psoriasis and Skin Cancer
If apoptosis fails, mutations may occur, which can lead to skin cancer. Thus, the therapeutic use of psoralens and UVA (PUVA) is not without drawbacks. PUVA therapy for psoriasis increases the incidence of squamous cell carcinoma (23). This is thought to be due to the formation of mutagenic psoralen photoadducts in the keratinocytes that are repaired incorrectly or not at all. Based on mutation frequencies in human fibroblasts treated with physiologic doses of 8-MOP and UVA, Burger and Simons (24) calculated the number of induced mutations in human hprt genes per phototherapeutic session (1.2*10 ) and per 30 years of maintenance therapy (1.3*10 mutants per cell). This is two orders of magnitude greater than the spontaneous mutation rate. Petersheim et al. observed chromosomal aberrations and sister chromatid exchanges in treated lymphocytes immediately after photopheresis (25). Although the damage was repaired in 72 hours, no molecular information was presented to demonstrate the accuracy of the repair processes. The implications of these data are not clear, but close monitoring for mutations occurring in treated cells would appear to be prudent.
It has been thought that crosslinks would primarily lead to the observed mutations, because the mutations occur at crosslinkable sites (5′TpA). Recent evidence indicates that this is probably not the case. Chiou et al. (26) studied the induction of mutations in the hprt gene in human fibroblasts. Using 8-MOP plus low doses of UVA (0.0060 J/cm ) to induce primarily monoadducts, mutations at 5′-TpA sites were observed. Fewer hprt mutations per 10 6 clonable cells were detected when a protocol favoring crosslinks (second irradiation of treated cells in the absence of any free psoralen) was employed. Because the numbers of adducts were not reported, drawing firm conclusions is difficult, but monoadducts might be at least as mutagenic as crosslinks.
Gunther et al. (27) studied the mutations induced by UVA and 8-MOP (producing both crosslinks and monoadducts) and those induced by UVA and 5-methylangelicin (5-MA; producing only monoadducts), related to adduct distribution. Mutagenesis was examined in a mouse fibroblast cell line carrying a recoverable, chromosomally integrated lambda phage shuttle vector, using the supF gene as a mutation reporter gene. Both psoralens generated predominantly T:A to A:T transition mutations with some T:A to G:C transversions. Most of the mutations occurred at TpA and ApT sites, both of which are conducive to crosslink formation. However, 5-MA induces only monoadducts. In addition, it was shown by HPLC analysis that, under the 8-MOP/UVA conditions used, only 20% crosslinks and 80% monoadducts, were formed, strongly indicating that monoadducts, as well as crosslinks, are critical premutagenic lesions. Earlier, different studies concluded that crosslinks were more mutagenic than monoadducts, showing one mutation hot spot that correlated with crosslink formation (28, 29). These studies were performed by irradiating isolated vector DNA containing the supF gene with very high doses of 8-MOP and UVA. Gunther et al. (27) irradiated cells containing DNA with a chromosomally integrated supF transgene. Thus, the different results of these two kinds of studies could be attributed to the extra- and intracellular PUVA treatments used, which would be expected to lead to different adduct yields and distributions and, hence, to different mutation frequencies and spectra.
6. p53 Mutations in Murine vs Human Skin Cancer
Mice exclusively exposed to a regimen of psoralen and UVA radiation developed skin cancers over a 42-week period (30). Analysis of p53 mutations in the resulting tumors showed evidence of psoralen photochemistry at the expected 5′-TpA hot spots. In humans, however, the picture is significantly different. In two independent studies, a paucity of PUVA-type mutations at 5′TpA sites were found, and UVB-type mutations at dipyrimidine sites were more prevalent (31, 32). These data suggest that the incidence of skin cancers in PUVA patients may be traceable to some degree of immunosuppression in the PUVA-treated skin. In fact, these cancers may arise from previous solar damage. It may be possible to dissect the relative contribution of previous actinic damage and therapy-induced damage by analysis of p53 mutations in these patients.
Although it is common to think that DNA damage is mutagenic and cytotoxic, it has also been shown that it can lead to gene induction. In fact, many studies have illustrated how DNA damage can induce immune-modulating gene expression. An example of this was demonstrated by Bernstein et al. (33) in a transgenic murine system. Transgenic mice containing a chromosomally integrated reporter gene construct with the promoter for human elastin linked to the gene for chloramphenicol acetyl transferase (CAT) were used to demonstrate the dose-dependent induction of CAT activity (33). Events such as these link DNA damage to gene induction and can explain changes in cytokine production in phototreated cells (34, 35)
7. Therapeutic Applications
Psoralen photochemotherapy has been widely used for the treatment of the hyperproliferative skin disease psoriasis (36). 8-MOP is administered orally or topically and, after an appropriate time, the affected skin is exposed to UVA light. Clearing can be achieved after 8-12 weeks, with three treatments each week. The PUVA therapy can be gradually decreased, so that only maintenance therapy at monthly intervals is required. For vitiligo, a similar therapy can be followed, leading to repigmentation of the affected skin after 200-300 treatments (37)
It has long been thought that 8-MOP DNA photoadducts inhibit cell division and lead to the beneficial effects of PUVA therapy. However, immunosuppression due to PUVA also appears to play an important role (38). Although PUVA therapy has a high efficacy, it also exhibits some undesirable side effects, such as skin phototoxicity and increased risk of squamous cell cancer (see text above), both of which are thought to be due to the ability of psoralens to undergo photoreactions with DNA.
8. Other Applications
A completely different application of psoralens and UVA light is the sterilization of blood products, mainly platelet concentrates (38, 39). It has been shown that 8-MOP and 4′-aminomethyl-4,5′,8-trimethylpsoralen (AMT) are able to reduce the titers of enveloped viruses by factors of 105-10 6 while preserving platelet function. AMT has the advantage of being water soluble, and it is not harmful to the DNA-lacking platelets in the presence of scavengers of reactive oxygen species (40). AMT exerts some mutagenic effects in the dark, however, so it must be removed before transfusion. Other psoralens with undisclosed structures have reported to be very efficient for platelet concentrate sterilization (41), but the lack of published data makes evaluation of these claims very difficult.
9. A Tool for the Study of Macromolecular Structures
In studies employing adventitious interactions of proteins and nucleic acids, the ability of psoralen to be photoactivated in situ and to lock in structures has been used to identify molecular contact points between the interacting molecules (42). In other studies, psoralens have been chemically linked to oligonucleotides that recognize specific DNA sequences (including double helical structures) to deliver psoralen adducts either to specific nucleic acid base sequences or to proteins in contact with oligonucleotides (43). The internal structure of 16 S ribosomal RNA (rRNA) in the small ribosomal subunit has been studied using a site-directed crosslinking approach (44, 45). Psoralen derivatives were transferred into specific positions in the 16 S rRNA aided by complementary deoxyoligonucleotides, which were completely removed after the transfer reaction. Another photoreactive group, azidophenacyl, was then attached to the psoralen moiety before reconstitution of the 16 S rRNA into the 30 S subunit. Re-irradiation of the derivatized subunits with UV radiation produced intramolecular crosslinks in the 16 S rRNA, as evidenced by gel electrophoresis under denaturing conditions. Crosslinking points in the 16 S rRNA were determined after preparative separation of the products by gel electrophoresis, followed by reverse transcriptase analysis.
In another study by the same group, this technique was refined by pre-forming monoadducts in an oligonucleotide (46). Thus, monoadducts were targeted to specific nucleotides in the pre-mRNA, leading to the subsequent formation of intramolecular RNA-RNA crosslinks after another round of photoactivation. This decreased the number of crosslinked products in comparison to nonspecific psoralen crosslinking and confirmed the locations of previously determined free psoralen crosslinks in the human precursor mRNA. New crosslinks consistent with an alternative secondary structure were also observed, as were a small number of crosslinks representative of higher-order interactions. The use of psoralen isomers such as angelicins, and/or activation with other UV and visible wavelengths, could lead to more selective photochemical reactions that have the potential to reveal other binding modes and sites. The recent elucidation of a psoralen amino acid photoadduct is noteworthy in this context (see text above).
Table 2. Repair of 8-MOP Photoadducts in Different Cell Types®
|
Cell Type |
8-MOP (ng/ml)/UVA (J/cm2) |
Adducts/mbp |
% repaired (24 h) |
|
Human lymphocytes (15) |
|||
|
Resting |
1 0/1 |
0.40 |
25 |
|
PHA stimulated |
10/1 |
0.80 |
52 |
|
Murine keratinocytes |
1 5/1 |
0.48 |
54 |
|
(17) |
|||
|
Murine fibroblasts |
1080/0.1 |
4.4 |
ND |
|
(27) |
|||
|
1080+60’400nm light |
2.9 |
66 |
|
Murine lymphoma (46) |
1 00/1 |
4.2 |
54 |
|
Bovine SMC (47) |
1000+12J/cm2 419 nm light |
13.5 |
25 |
10. Historical Acknowledgments
Five figures stand out for their contributions to the field of psoralen photochemistry, photobiology, and photomedicine. A half century ago, Aaron Lerner and Tom Fitzpatrick pioneered the application of psoralen photochemotherapy for vitiligo. In the 1970s, this was expanded by John Parrish to include therapy for psoriasis. It was in this same era that John Hearst developed techniques that led to a fuller development of psoralen photochemistry. In the 1980s, largely through the efforts of Margaret Kripke, we came to a better understanding of the immune-modulating effects of photoactivated psoralen.
