Molecular basis for mustard-induced vesication

FUNDAMENTAL

AND

APPLIED

TOXICOLOGY

Molecular

5, s 134-s

149 ( 1985)

Basis for Mustard-Induced

Vesication’s’

BRUNO PAPIRMEISTER,~ CLARK L. GROSS, HENRY L. MEIER, JOHN P. F%TRALI, AND JOHN B. JOHNSON U.S. Army

Medical

Research

Institute

of Chemical

Molecular Basis for Mustard-Induced

Defense,

Vesication.

Aberdeen

Proving

PAPIRMEISTER,

Ground,

Maryland

B., GROSS,

C. L.,

21010-5425

MEIER,

A biochemical hypothesis explaining the generation of pathology in human skin by mustard gas (HD) is presented whichlinksthe initiation of DNA damages to local alterations of metabolism and subsequent development of blisters. The proposed sequence involves HD alkylation of purines in DNA which are processed to form apurinic sites. Apurinic endonucleases act at these sites to produce backbone breaks in DNA which cause activation ofthe chromosomal enzyme poly(ADPribose)polymerase. This enzyme utilizes NAD+ as a substrate and, at vesicating doses of HD, would deplete the cells of their NAD+ content. The depletion in NAD+ would cause inhibition of glycolysis, and the resulting accumulation of common intermediates would stimulate the NADPCdependent hexosemonophosphate shunt (HMS). Such stimulation of the HMS has been associated with DNA damage and enhancement of protease synthesis and release. These proteases could be responsible for development of subepidermal blisters which result from fluid accumulation in the cavity created by separation of the moribund basal cell layer from the basement membranea characteristic feature of HD-exposed human skin. Partial validation of this biochemical hypothesis has been achieved. DNA alkylated with either monofunctional or bifunctional sulfur mustards, followed by spontaneous or enzymatic depurination, was shown to be sensitized to degradation by apurinic endonuclease. Studies on the effect of HD on human skin grafted to athymic nude mice demonstrated dose- and time-related decreases in NAD+ levels. These decreases in NAD+ levels preceded and correlated to the predicted severity of pathology. The participation of poly(ADPribose)polymerase activity in the HD-induced NAD’ loss was substantiated by prevention of this loss in the presence of inhibitors of the enzyme. Additional supporting evidence for the proposed mechanism was obtained at the cellular level by studies which utilized human leukocytes. The subsequent involvement of the HMS and proteases in HD-induced vesication is discussed. H.

L., PETI~LI,

J. P., AND

JOHNSON,

J. B. ( 1985).

Fundam.

Appl.

Toxicol.

5, S134-S149.

0 1985 Society of Toxicology.

Mustard gas (HD),“ the vesicam CW agent first used in World War I and, according to press reports, more recently, in the Iraq-Iran con’ The opinions or assertions contained herein are the private views of the authors and are not to be construed as reflecting the views of the Department of the Army or the Department of Defense, USA. * This work was presentd in part at the Third Annual USAMRDC Chemical Defense Bioscience Review, 2-3 June 1983, Aberdeen Proving Ground, Md.; 68th Ann. Mtg Fed. Am. Sot. ExptI. Biol., April l-6,1984, St. Louis, MO.; Fed. Proc. 43(3), 2452,1984; and Pmt. Army Science Conf., June 1984, West Point, N.Y. 3 To whom reprint requests should be addressed: Commander, USAMRICD, AT’IN SGRD-UV-CT/Dr. Bruno Papirmeister, Aberdeen Proving Ground, Md. 2 1010-5425. 4 Abbreviations used: HD, bis-2-chlorcethyl sulfide, mustard gas, sulfur mustard, poly(ADP-tibose) polymerase, poly adenosine diphosphoribose polymerase; HMS, 0272X)590/85 $3.00 Copyright

0 1985 by the Society of Toxicology.

All rights of reproduction

in any form reserved.

flict, was the first chemical compound shown to possess mutagenic activity (Auerbach and Robson, 1946). It possesses additional genotoxic properties, such as the ability to produce chromosomal aberrations, cancers, and a variety of DNA damages, which are thought to be responsible for the high sensitivity of proliferating cells (Fox and Scott, 1980). HD and its nitrogen-containing analogs have radiomimetic properties which are being exploited in cancer chemotherapy (Ross, 1962; Connors, hexose monophosphate shunt; NAD+, oxidized form of nicotinamide adenine dinucleotide; NADP+, oxidized form of nicotinamide adenine dinucleotide phosphate; CEES, Z-chloroetbyl ethyl sulfide; Semi-HN, 2-chIomethyl diethylamine.

s134

.MULl _^_ _^___ . - .BASIS . ,..,. --XlJLAK i-UK

1975). The data demonstrating that cells incapable of repairing HD-induced DNA crosslinks and apurinic lesions were considerably more sensitive than repair capable strains were the most compelling evidence that DNA rep resents the most sensitive cellular target (Papirmeister and Davison, 1964; Lawley and Brookes, 1965; Kohn et al., 1965). Experiments performed in the late 1960s suggested that DNA damage might also be important in the cutaneous HD injury (Papirmeister et al., 1969). When DNA repair in HD-exposed rab bit skin was inhibited by topical application of caffeine or proflavine, skin injury was produced at lower doses and was more severe. The possibility that the cytotoxic, vesicant, and other pathologic effects of HD were due to metabolic disturbances, unrelated to the genotoxic actions of this alkylating agent, received much attention. Thus, while recognizing the highly genotoxic potency of HD and its much lower potency for inducing metabolic effects, Wheeler ( 1962) nevertheless concluded that a correlation exists between the inhibitory action of alkylating agents on glycolysis and their carcinostatic, carcinogenic, and vesicant properties. Dixon and Needham ( 1946) stated that a direct connection exists between HD’s characteristic inhibition of glycolysis and the production of skin injury. All other vesicants also produced this glycolysis inhibition whereas non-vesicants, even though closely related chemically, did not. Additionally, the degree of glycolysis inhibition and the time at which such inhibition occurred correlated with production of visible skin damage. The objective of this presentation is to propose a biochemical hypothesis-a model which links DNA damage, alterations in metabolism, and pathology-for HD-induced vesication in human skin. Following the rapid fixation of HD, several hours may elapse before visible injury can be detected (Renshaw, 1946), a time sufficiently long for developing strategies designed to attenuate, modify, or prevent the production of highly incapacitating slow-healing skin lesions. Three distinct models were employed to validate portions of

MUSTARD

VESICATION

s135

our biochemical hypothesis: (a) HD-treated DNA in vitro-to identify the major DNA lesions produced and how such lesions are processed by nucleases; (b) HD-exposed human skin grafts in athymic nude mice-a human skin model which provides pertinent morphological, biochemical, and other in vivo data on the development of skin lesions (Papinneister et al., 1984a,b); and (c) HD-treated human leukocytes-to study relevant biochemical aspects at the cellular level. BIOCHEMICAL HYPOTHESIS HD VESICATION

FOR

The severity of the cutaneous HD injury is dependent on the degree of alkylation of skin constituents (Renshaw, 1946). In man, alkylation occurs within minutes and is followed, after a latent period, by slowly developing pathology (Renshaw, 1946). We propose a biochemical model in which DNA damage is linked to sequential changes in metabolism, followed by morphological changes in cells, which eventually results in a vesicating lesion. For vesicating doses of HD, the following sequence is postulated (Fig. I): (a) Pathogenesis is initiated by rapid alkylation of DNA, especially in the keratinocytes of the basal layer (Renshaw, 1946); (b) alkylated DNA purines are unstable and subject to both spontaneous and enzymatic depurination, producing a large number of apurinic sites which, upon cleavage by constitutive apurinic endonucleases, yield DNA breaks (Papirmeister et al., 1970; Lindahl, 1979); (c) accumulation of DNA breaks causes activation of the chromosomal enzyme poly(ADP-ribose)polymerase, which utilizes NAD+ as a substrate to ADP-ribosylate a variety of nuclear proteins and causes a severe lowering of cellular NAD+ (Rankin et al., 1980; Shall, 1982); (d) depletion of NAD+ results in inhibition of glycolysis which was previously found to parallel injury production by HD (Renshaw, 1946); (e) stimulation of the NADP+dependent I-IMS pathway follows due to accumulation of glucosed-phosphate, a

S136

PAPIRMEISTER ALKYLATION OF

~*~rrlmtion

DNA

INHIBITION GLYCOLYSIS

DEPURINATED

r*‘c”on’

OF

~

ET AL. a..,,.~. •~d.I~01~.SP

DNA

CELLULAR DEPLETION

NAD+

l

,

DNA WITH BREAKS

‘ACTIVE’POLY (ADP-RIBOSE) POLYMERASE

HEXOSE SHUNT

MONOPHOSPHATE ACTIVATION

+

PROTEASE RELEASE

, PATHOLOGY

FIG. 1. Biochemical hypothesis for the cutaneous HD injury. The proposed mechanism describes sequential events which link DNA alkylation to metabolic disturbances and the development of pathology.

common precursor for both glycolysis and HMS (Lehninger, 1979); (f) protease release (e.g., plasminogen activator) is stimulated, a process which has been reported to be caused by both production of DNA damage (Miskin and Reich, 1980) and enhancement of HMS (Schnyder and Baggiolini, 1980); and finally, (g) pathological changes are produced, characterized by basal cell necrosis, breakage of anchoring filaments of hemidesmosomes, increase in the colloid osmotic pressure at the dermoepidermal junction, accumulation of edema fluid within the cavity, which culminates in formation of subepidermal microvesicles and, following coalescence, produces the characteristic HD blister (Papirmeister et al., 1984a,b). PRODUCTION OF BREAKS IN HDTREATED DNA BY APURINIC ENDONUCLEASE The first premise of our hypothesis suggesting that HD-treatment of DNA initiates the formation of breaks had to be established. Activation of poly(ADP-ribose)polymerase and the following steps of our hypothesis for HD-induced vesication are dependent on the formation of such DNA breaks. Although al-

kylation of DNA by HD rarely produces breaks directly, such breaks are readily formed by incubation of HD-treated DNA with crude extracts from bacteria or mammalian cells. It appears that breaks in HD-treated DNA are primarily caused by formation of apurinic sites and cleavage of these sites by apurinic endonucleases. Degradation to acid-soluble products results from the further action of exonucleases present in crude extracts. The major sites of DNA alkylation by HD are monofunctional adducts at the NT- position of guanine (about 60%) and the Nj- position of adenine (about 16%). Bifunctional adducts account for another 16% and involve the NT- positions of two adjacent guanines (intrastrand crosslinks) or of two guanines situated on opposite strands (interstrand crosslinks) (Brookes and Lawley, 196 1; Papirmeister, 196 1). In addition, several trace monofunctional adducts on DNA purines and pyrimidines have recently been identified which may be important in the mutagenic and/or carcinogenic properties of HD (Ludlum et al., 1984). Previous data using DNA and synthetic polydeoxynucleotides treated with several bifunctional and monoRmctional mustards demonstrated that the N3-alkyladenine adducts had the greatest sensitizing effect to endonucleases present in crude extracts

MOLECULAR

BASIS FOR MUSTARD

PHD CEES

20

i;

0

5

O MUSTARD

10 CONCENTRATION

I-

100

IX104#)

FIG. 2. Degradation of alkylated DNA by a crude extract of E. coli. Two micrograms of [3H]thymidine-1abe1ed E. coli DNA ( lo6 dpm/& was alkylated with 100 ~1 of the bifunctional sulfur mustard, HD (0), the monofunctional sulfur mustard, CEES (a), or the monolimctional nitrogen mustard, semi-HN (Cl), at the indicated concentrations. Alkylation was for 1 hr at room temperature in buffer containing 10 mM potassium phosphate (pH 7.65)-150 mM NaCl. Alkylated DNA was incubated at 37°C for IO min with a crude extract (from E. coli Bs-1) and the radioactivity soluble in ice-cold 0.3 N perchloric acid was determined. Incubation mixtures contained per 1.5 ml: 100 rmol Tris buffer (pH 7.9, 10 pmol MgClr, 50 pg tRNA, and 2.0 mg crude extract protein.

from bacteria and mammalian cells (Papirmeister et al., 1970; Lindahl, 1979). The high sensitizing effect of N+lkyladenine adducts produced by sulfur mustards is confirmed by current studies. Figure 2 shows that crude extracts from Escherichia coli preferentially degraded DNA that had been freshly alkylated by either the bifunctional HD or a monofunctional analog, chloroethyl ethyl sulfide (CEES). In contrast, alkylation of DNA by a monofunctional nitrogen mustard, which causes the formation of N+lkylguanine but does not produce N3-alkyladenine adducts (Price et al., 1968), did not sensitize the polymer to degradation. This specific sensitization of DNA containing N+lkyladenine adducts was attributed to the action of another enzyme,

VESICATION

s137

N+lkyladenine DNA glycosylase, which is present in several different extracts from bacteria and mammalian cells (Lindahl, 1979). This depurinating enzyme was found to eliminate N+lkyladenine from DNA treated with sulfur mustards (Fig. 3A), but failed to remove the more prevalent product, N+lkylguanine (Fig. 3B). Also shown in this figure is spontaneous depurination of the unstable major monoadducts, N+lkyladenine and N+lkylguanine. Once formed by either enzymatic or spontaneous processes, apurinic sites in DNA are rapidly cleaved by apurinic endonucleases. Figure 4 demonstrates that CEES alkylated/ depurinated DNA is a much better substrate than fresh CEES alkylated DNA for the highly purified apurinic endonuclease from E. coli, endonuclease VI/exonuclease III. For these investigations, sonicated [3H]thymidine-labeled DNA treated with the monofunctional sulfur mustard, CEES, was used. Freshly alkylated and alkylated/depurinated (i.e., produced by preincubation of alkylated DNA for 48 hr at 37°C) DNA was incubated with apurinic endonuclease. Duplicate samples were treated with alkali to determine the total number of available alkali-labile (apurinic) sites. As shown in Fig. 4, freshly alkylated DNA exhibits a dose-dependent increase in alkali lability (3.2, 8.2, and 13.8% acid solubility for control, 5 mM CEES-treated, and 10 mM CEES-treated samples, respectively). Alkali lability is greatly enhanced if the alkylated DNA is incubated for 48 hr (3.4, 31.3, and 40% for the corresponding alkylated/depurinated samples, respectively). While the alkalilabile sites of freshly alkylated DNA are derived primarily from spontaneous depurination of N3-alkyladenine, the alkali-labile sites of the alkylated/depurinated DNA are derived from spontaneous depurination of both N3-alkyladenine and N,-alkylguanine (not shown). Also to be noted is the smaller dosedependent increase in acid solubility, even in the absence of the alkali treatment (1.8, 2.6, and 4.5% for the corresponding alkylated

S138

PAPIRMEISTER

ET AL.

FIG. 3. Removal of alkylated purines from DNA treated with a monofunctional sulfur mustard by crude cellular extracts. E. coli DNA (1 mg/ml) was treated with 5 mM [‘%]2-chloroethyl, 2-hydroxyethyl sulfide (semi-HD) for 30 min at 37°C in 10 mM potassium phosphate buffer (pH 7.65)-150 mM NaCl. Alkylated DNA was dialyzed at 4°C against several changes of buffer to eliminate radioactive hydrolysis products. Alkylated DNA (45 Fmol semi-HD bound/mmol DNA-P) was incubated at 37°C with crude extracts derived from the indicated E. coli strains. The incubation mixtures contained per milliliter of KA buffer (50 mM potassium phosphate, pH 7.5, 5 mM 2-mercaptoethanol, 1 mM EDTA): 500 pg alkylated DNA and 3.5 mg crude extract protein. At the times shown, 20-al aliquots were removed, spotted on Whatman 3M paper and chromatographed in ascending direction for 16 hr. Peaks corresponding to N3-alkyladenine and N7alkylguanine were located with the aid of appropriate radioactive markets, and assayed for radioactivity. The total amount of NJ-alkyladenine or N,-alkylguanine was determined by measuring the radioactivities of corresponding peaks which were released by mild acid hydrolysis of the alkylated DNA. Extracts used were no extract (O), AB 1157 (0), BW 2001 (0), AB3027 (A), AB 1884 (A), AB 1885 (Cl), and AB 1886 (H). (A) Release of N3-alkyladenine; (B) release of NT-alkylguanine.

samples, respectively; 2.5, 6.6, and 12.9% for the corresponding alkylated/depurinated samples, respectively). With freshly alkylated DNA, these effects may reflect the production of a small number of strand breaks due to direct action of a&dating agent on DNA, while the somewhat larger breakage in alkylated/depurinated samples may be due to hydrolysis of a few apurinic sites. Samples incubated with purified apurinic endonuclease had most of their alkali-labile sites cleaved enzymatically, indicating that DNA breaks are produced almost exclusively at apurinic sites. When corrected for direct CEES-induced breaks, the number of alkali-induced and apurinic endonuclease-induced breaks was identical. En-

zyme activity was significantly greater with alkylated/depurinated DNA than with freshly alkylated DNA. The dose-response relationship for the sensitization of HD alkylated/depurinated DNA to degradation by purified apurinic endonuclease VI/exonuclease III of E. coli is shown in Fig. 5. Rate measurements were carried out by incubating samples for 30 min since the rate of formation of acid-soluble products was linear. The extents of degradation were determined following incubation for 120 min, when degradation was complete. The following results were obtained: (a) Only double-stranded, DNA was degraded, the small amount of degradation of unalkylated DNA being due to

MOLECULAR

ND CEFS

BASIS FOR MUSTARD

5mM CEES lomM CEES

VESICATION

NO CEES

5mM CEES mm

s139

CEES

4. Endonucleolytic action of apurinic endonuclease at alkali-labile (apurinic) sites. [‘H]Thymidinelabeled sonicated E. coli DNA (0.05 aCi/ag from Miles Laboratories, Elkhart, Ind.) was treated with the monofunctional sulfur mustard, CEES, at 5 and 10 mM concentrations. One-half of each preparation was incubated for 48 hr at 37’C to provide alkylated/depurinated samples. Unalkylated controls were handled in the same manner. All preparations were incubated for 30 min at 37°C with purified apurinic endonuclease from E. coli, endonuclease VI/exonuclease III (Boeringer Mannheim, Indianapolis, Ind.). Incubation mixtures contained per milliliter: 65 pmol Tris buffer (pH 8.0) 0.65 pmol MgClr, I pmol 2-mercaptoethanol, 250 pg bovine serum albumin, and, when indicated, 1.O pg enzyme. Following incubation, preparations were subjected to filtration on Sephadex G-15 and the fractions excluded by the gel were pooled. Duplicate aliquots were then either precipitated with perchloric acid directly or following a 15-min incubation at 37°C with 0.2 N NaOH, and the percentage of the radioactivity rendered acid-souble was determined. The bar graph compares the degradation produced by the apurinic endonuclease with that resulting from cleavage of alkali-labile sites as follows: solid bars-minus enzyme, minus alkali; stippled bars-minus enzyme, plus alkali; open bars-plus enzyme, minus alkali; cross-hatched bars-plus enzyme, plus alkali. FIG.

exonucleolytic activity of the enzyme; (b) the rate of degradation was linearly related to the dose of HD; and (c) the extent of degradation was also dose-related, but never exceeded 4045%. The remaining DNA was single-stranded

and subject to further degradation by addition of exonuclease I (data not shown)-an enzyme specific for hydrolyzing single-stranded DNA. Conclusions. (a) The vast majority of DNA breaks are not produced directly by the alkyl-

PAPIRMEISTER

EXTENT

RATE

0

0

2

4

6

8

10 4

HD

CONCENTRATION

x10

MOLAR

FIG. 5. Dose-response relationship for sensitization of HD alkylated/depurinated DNA to degradation by purified apurinic endonuclease VI/exonuclease III of E. coli. [‘HlThymidine-labeled E. coli DNA was alkylated by the indicated concentrations of HD and depurinated by incubation for 48 hr at 37°C. Aliquots were then incubated with enzyme (2.2 &ml) for 30 and 120 min and the acidsoluble radioactivity was determined. Other incubation conditions were as described in Fig. 4. Samples which were incubated for 30 min were used to calculate the rate of degradation (a), while samples which were incubated for 120 min were used to determine the extent of degradation (0). The degradation of alkylated/depurinated DNA to acid-soluble products is due to the sequential actions of the apurinic endonuclease function of the enzyme, followed by exonuclease III action at 3’ ends.

ating agent but result from sensitization to enzymatic breakage by apurinic endonucleases; (b) apurinic sites are formed mainly by enzyme-induced and/or spontaneous depurination of the major monoadducts, N+&yladenine and N+lkylguanine; (c) DNA crosslinks make, at best, only a minor contribution to sensitization; (d) apurinic endonuclease quantitatively converts apurinic sites to DNA breaks, creating good substrates for further degradation by exonucleases. The results are consistent with our biochemical hypothesis

ET AL.

according to which the production of DNA breaks plays a prominent role in initiating the HD injury. The abilities of monofunctional sulfur mustards to sensitize DNA to enzymatic breakage is of interest since these alkylating agents are also potent vesicants (Renshaw, 1946). PATHOLOGICAL CHANGES IN HDEXPOSED HUMAN SKIN GRAFTS IN ATHYMIC NUDE MICE After establishing that HD induced DNA breaks, the initial assumption of our hypothesis, it was felt that the first visible pathology should be at the nuclear level. In order to study the pathogenesis of the cutaneous HD injury, we employed a new animal model for human skin-human skin grafts in congenitally athymic nude mice. Injury produced in this skin model was similar to that reported in humans, including the formation of microblisters (Papirmeister et al., 1984a,b). After a latent period of several hours, vesicating doses of HD produce overtly visible pathology in human skin grafts. Consistent with our biochemical hypothesis, light microscopic and ultrastructural evaluation showed that the earliest morphological changes do, indeed, appear in the nucleus and are especially pronounced in the cells of the basal layer of the epidermis. Nuclear pathology is soon followed by cytoplasmic injury-swelling of both the smooth and rough endoplasmic reticulum, formation of debris filled cytoplasmic vacuoles, loss of rosettes of polysomes, breakage of the plasma membrane that abuts the basal lamina, extrusion of intracellular debris into the lamina lucida, and loss of mitochrondrial integrity. All of these necrotic changes in basal keratinocytes precede microblister formation. Typical light microscopic changes 12 hr af ter exposure of a human facial skin graft to a vesicating dose of HD (635 &cm*) are shown in Fig. 6A. Many of the basal cells were pyknotic. These focal areas of pyknotic cells were more numerous than those seen at 6 hr postexposure, when nuclear changes are first noted,

FIG. 6. Pyknosis in basal kerotinocytes due to HD exposure of human skin grafts in athymic nude mice. (A) Facial skin graft I2 hr after exposure to a vesicating dose of HD 635 (Ilp/cm’). Many basal cells have pyknotic nuclei, which were first seen after a latent period of about 6 hr. After 24 hr practically all the basal cells were pyknotic (Papirmeister et al., 1984a). Some of the cells in the spinous layer are hypochromic (H & E, X 340). (B) Electron micrograph of a pyknotic basal cell nucleus (n) 24 hr after exposure of a human foreskin graft to a subvesicating dose of HD (60 &cm’). Note the decrease of euchromatin, condensation of heterochromatin, extensive damage and blebbing of the nuclear membrane (arrows), and formation of debris-filled perinuclear and cytoplasmic vacuoles (v). The nuclei of neighboring basal cells show less damage. Reprinted from Papirmeister et al. (I 984a,b) by courtesy of Marcel Deklcer, Inc.

S142

PAPIRMEISTER

but not as extensive as at 18-24 hr, when all of the basal keratinocytes are pyknotic (Papirmeister et al., 1984a,b). Occasionally, the epidermis was beginning to separate from the basement membrane at the epiderma.l-derrnal junction, giving rise to a small cleft which was filled with debris and a few floating cells. However, overt acantholysis was not a prominent feature after 12 hr. While some of the differentiated keratinocytes in the spinous layer were vacuolated, a large number of these epidermal cells appeared unchanged. Dermal changes were less pronounced. The collagen was well organized, although some separation of the bundles indicated the presence of dermal edema. Ultrastructural features of a basal kerotinocyte undergoing pyknosis 24 hr after exposure of a neonatal foreskin graft to a subvesieating dose of HD is shown in Fig. 6B. At this lower dose of HD (60 &cm2), basal cells with pyknotic nuclei coexisted with neighboring basal cells containing normal appearing nuclei. The pyknotic nucleus was characterized by a preferential decrease in euchromatin (containing active DNA) and condensation of heterochromatin (containing inactive DNA). The nuclear membrane lost its integrity and blebbing of the nuclear envelope was a prominent feature. The shrunken nucleus created a perinuclear vacuole which contained debris. Although the number of pyknotic basal cells increases with both the dose of HD administered and the post exposure time, the nuclear changes just described are typical and precede the appearance of cytoplasmic injury and blister formation (Papirmeister et al., 1984a,b). Blister formation is a unique characteristic of HD-exposed human skin but usually does not occur in the skin of experimental animals (Renshaw, 1946; McAdams, 1956). The occurrence of microblisters in HD-treated human skin grafts in athymic nude mice validates the use of this model for studying the pathogenesis induced by this vesicant (Papirmeister et al., 1984a,b). Separation of the graft epidermis from the basement membrane just

ET

AL.

above the epidermal-dermal junction begins 12-24 hr following exposure with formation of extracellular vacuoles below the moribund basal cell layer. These vacuoles become more numerous with time, fill with fluid and debris, coalesce, and cause widening of the cleft. Figure 7A depicts a full-fledged subepidermal microblister 48 hr after exposure of a human facial skin graft to a vesicating dose of HD. The roof of the blister consisted of necrotic basal cells while the base of the blister contained the basement membrane which overlaid the slightly edematous, but otherwise intact, dermis. Analysis of ultrastructural features of the HD-induced microblister detailed the sequence and anatomy of vesication. Prevesication lesions 24 hr after HD exposure had shown that the basal cell plasma membrane adjacent to the junction sustains numerous breaks through which intracellular substances leak into the lamina lucida, causing the disruption of epidermal-dermal adhesion (Papirmeister et al., 1984b). Figure 7B shows an electron micrograph of a typical microblister 48 hr after exposure of a human facial skin graft to a vesicating dose of HD. An intact basal lamina formed the base of the blister cavity which contained debris and some degenerating cellular organelles which appeared to have been extruded through the highly damaged basal cell plasma membrane. Intact hemidesmosomes and the remnants of the plasma membrane formed the roof of the blister. Anchoring filaments which were attached to hemidesmosomes were seen dangling into the cavity. Other structures, such as tonofilaments and anchoring fib&, appeared relatively undamaged. The pathology results stongly support our biochemical hypothesis by showing that skin injury by HD is characterized by early nuclear damage that progresses to basal cell necrosis and vesication. Such early nuclear damage is also consistent with the data demonstrating the ability of HD to sensitize DNA to enzymatic breakage.

MOLECULAR

BASIS FOR MUSTARD

VESICATION

. Formation of microblisters in HD-treated human skin grafts in athymic nude mice. (A). Facial t 48 hr after exposure to 635 pg HD/cm’. The epidermis has completely separated from the (dermis fe the basement membrane with formation of subepidennal blister cavity. The roof of the blister just the moribund cells of the basal layer (Humphrey stain, X 160). (B) Electron micrograp Nhof a cont micr*obli!rter which developed in a facial skin graft 48 hr after exposure to 635 pg HD/cm’. The blister cavity was fern led by breakage of anchoring filaments (af) from their attachment sites on the basal lami na (bl) Fl

s144

PAPIRMEISTER

LOSS OF NAD+ IN HUMAN SKIN GRAFTS DUE TO HD EXPOSURE According to our biochemical hypothesis, the production of DNA breaks by HD would activate the chromosomal enzyme poly(ADPribose)polymerase, as was previously demonstrated for other alkylating agents (Hayaishi and Veda, 1982). With extensive DNA damage, caused by vesicating doses of HD, activation of poly(ADP-ribose) could be sufficiently great as to consume the available NAD+ in exposed skin. Depletion of this vital cofactor could be responsible for the previously reported depression of skin glycolysis by HD (Renshaw, 1946; Dixon and Needham, 1946) and the pathologic consequences leading to vesication. Consequently, we investigated the effects of HD treatment on the NAD+ content of human skin grafted to athymic nude mice (Gross et al., 1985; Meier et al., 1984). The effect of various doses of HD and exposure times on NAD+ levels in human skin grafts is shown in Fig. 8. Three doses of HD were chosen, based on their abilities to produce mild ( 127 Kg HD/cm*), moderate (3 18 pg HD/cm*), and severe (1270 /Ig HD/cm*) injuries, respectively (Papirmeister et al., 1984a,b). All three doses of HD lowered the level of NAD+, which began to decrease within one hour following exposure. During the first 4 hr following exposure, NADf levels showed a dose-dependent decrease, reaching near minimal values. At the two higher doses, NADf levels plateaued near their minimum 4-hr levels for the entire 72 hr of the experiment. The NAD+ levels at the lowest dose recovered, reaching near control levels by 72 hr. It appears, therefore, that NAD+ levels have to decrease below a critical value-approximately 25% of control-before the NAD+ loss

ET AL.

80

7. 6. g 5 so g i 4. 5 2 i 3o *’ ’o 0 ”

4 Hours

8

12

After

Exposure

16

20

72

to HD

FIG. 8. The effect of various concentrations of HD and exposure time on NAD+ levels in skin grafts. Human skin grafted to athymic nude mice was exposed to varying concentrations of HD. Animals were sacrificed at the designated postexposure times, skin grafts were removed, and 4-mm biopsy punch skin samples were taken immediately. The skin samples were quickly frozen in Iiquid nitrogen, weighed, and then extracted with 0.5 M HClO, overnight at 4°C. The samples wre centrifuged and supematants were removed and neutralized with 2.0 M KOH-0.66 M potassium phosphate, pH 7.8. The skin samples were again extracted with 0.5 M HCIO, overnight and neutralized as above. Insoluble KCIO, was removed by centrifugation and the supematants were pooled and assayed for NAD+, using the enzymatic cycling assaydescribed by Jacobson and Jacobson ( 1976). Each time point was the average of two separate skin punches assayed in triplicate.

becomes irreversible. It should be noted that the HD-induced fall in skin NAIS content preceded the development of pathology ob served in grafts treated similarly (Papirmeister

which forms the base of the blister. Hemidesmosomes remain attached to the damaged basal cell plasma membrane which forms the roof of the blister. Basal cell debris (arrows) can be seen to extrude into the blister cavity (b). The adjacent dermis (d) appears well organized. Reprinted Born Papirmeister et al. (1984a,b) by courtesy of Marcel Dekker, Inc.

MOLECULAR

BASIS FOR MUSTARD

s145

VESICATION

in human skin, in accordance with our molecular model. The severity of this injury in humans was shown previously to depend on the degree of alkylation of skin constituents (Renshaw, 1946). The fixations of 0. I- 1.O pg HD/cm2 result in mild (erythematous), of 1.0-2.5 pg HD/cm2 result in moderate (vesicating), and of >2.5 pg HD/cm’ resulting in severe (necrotizing) skin injuries, respectively (Renshaw, 1946). To relate the loss of NAD+ to the severity of skin injury, human skin implants in nude mice were exposed to varying doses of “C-labeled HD, and the NAD+ levels after 4 hr were compared to alkylation levels in skin. ”

2

0 Hours

4 After

Exposure

6 to

6

HD

3 2 Ei

z

00 T

pa

HD

Cl

HD

+

S-aminobenzamide

FIG. 9. The effect of 3-methoxybenzamide pretreatment on the HD-induced decrease of NAD+ in human skin g&s. Athymic nude mice bearing human skin grafts were injected with 3-methoxybenzamide (3.75 mg, ip) I hr before exposure to HD. Human skin grafts were exposed to HD and mice wre sacrificed at the designated times. The skin samples were processed as described in legend to Fig. 8. Error bars represent the standard error of the mean for three biopsy skin punches (measured in duplicate) from each animal.

I-

et al., 1984a,b), consistent with our biochem-

ical hypothesis. If activation of the poly(ADP-ribose)polymerase were responsible for the observed NAD+ loss, then specific inhibitors of this enzyme (Purnell and Whish, 1980) should prevent the loss of NAD+. A preliminary experiment was performed in which 3-methoxybenzamide was injected ip into nude mice 30 min prior to exposure of human skin grafts to a moderate dose of HD (3 18 clg/cm2). Figure 9 demonstrates that this poly(ADP-ribose)polymerase inhibitor protected against loss of NAD+ for up to 6 hr after HD exposure. This finding implicates poly(ADP-ribose)polymerase activity in HD-induced NAD+ depletion

(0.1 )rg HD

0.1-1.0 Fixed

1 .o-2.5 /cm2

of

Skin

> 2.5 Graft

FIG. 10. The effect of “fixed” HD on NAD+ levels in biopsy punches from human skin grafted to athymic nude mice. Mice were either pretreated with 3aminobenzamide (50 me/kg, ip) or sham treated with vehicle for 30 min before exposure of human grafts to varying concentrations of r4C-labeled HD. Animals were given an additional injection of 3-aminobenzamide (50 mg/k& ip) 90 min after exposure to HD. The amount of HD fixed/cm* of human skin graft was correlated with the severity of skin damage previously described in man (Renshaw, 1946). Nude mice were sacrificed 4 hr postexposure, skin grafts were removed, and 4-mm biopsy punch skin samples were taken immediately. The skin samples were processed as previously described in legend to Fig. 8. The skin biopsy punches were combusted in a Packard Tri-Carb Sample Oxidizer and the “fixed” radioactivity contained in these punches was determined by scintillation counting.

S146

PAPIRMEISTER

The results of this study are shown in Fig. 10. The decrease in skin NAD+ levels 4 hr after exposure to HD correlated well with the expected severity of the injury. The fixation of >O. 1 pg HD/cm*, which does not produce any skin injury, did not alter the NAD+ content (compare with 0-hr controls in Fig. 8). The approximate losses of skin NADf content after 4 hr for the mild (fixation of 0. I- 1.O pg HD/cm*), moderate (fixation of 1.0-2.5 PLg HD/cm*), and severe (fixation of >2.5 pg HD/ cm*) injury groups were 55, 63, and 77%, respectively. While all of the 4-hr NAD+ losses sustained by the mild injury group can be expected to recover, most of the NAD+ losses in the moderate and all of the NAD+ losses in the severe injury groups are deemed to be irreversible (see Fig. 8). These findings show that 4-hr NAD+ values correlate with the severity of the skin injury produced by HD and could serve as an early biochemical marker of such injury. Also shown in Fig. 10 are the effects of 3aminobenzamide, an inhibitor of poly(ADPribose)polymerase (Pumell and Whish, 1980). The data demonstrate that the inhibitor increased the 4-hr NADf levels from 45 to 58% in the mild, from 37 to 68% in the moderate, and from 23 to 50% in the severe, HD-injury groups, respectively. Studies are underway to optimize treatment regimens. If increases in NAD+ levels can be achieved and sustained to maintain skin viability, this may prevent and/or attenuate the HD injury. Conclusion. The present results showing that HD exposure to human skin grafts in nude mice cause a dose-dependent lowering of NAD+ levels which precedes the onset of pathology and the ability of poly(ADP-ribose)polymerase inhibitors to attenuate the loss of NAD+ lend strong support for our biochemical hypothesis for HD-induced vesication. DISCUSSION The biochemical hypothesis for HD-induced vesication, which links DNA damage

ET

AL.

to metabolic disturbances and the development of pathology (Fig. l), has been partially validated by the present study. We have shown that monofunctional alkylations of DNA purines sensitize this macromolecule to nucleases, leading to DNA breaks. Most degradation of HD-treated DNA is caused by apurinic endonuclease which cleaves apurinic sites created by enzymatic and/or spontaneous depurination (Figs. 3-5). HD-induced DNA crosslinks are not primarily involved in the production of DNA breaks (Fig. 2) which initiate vesication. Such adducts would severely inhibit DNA replication (Roberts et al., 197 1) and prevent the division of basal kemtinocytes. These consequences could contribute greatly to the characteristic slow healing of cutaneous HD injuries (Renshaw, 1946). According to the hypothesis, these DNA breaks would activate the chromosomal enzyme poly(ADP-ribose)polymerase, which consumes NADf by ADP-ribosylating nuclear proteins (Hayaishi and Veda, 1982). At vesieating doses of HD, the cellular NAD+ would be rapidly depleted. We confirmed this speculation by demonstrating that HD exposure of human skin grafts in athymic nude mice causes a dose-dependent decrease in NADf levels (Fig. 8). The levels of NAD+ could be conserved by inhibitors of poly(ADP-ribose)polymerase (Figs. 9 and 10). Depression of skin NAD+ content, taking place during the asymptomatic latent period, correlates with the severity of the injury produced (Fig. 10). This observation could serve as an early biochemical marker of the injury. The appearance of nuclear damage in basal epidermal cells of HD-treated human skin grafts, which precedes cytoplasmic damage and blister formation, is consistent with our model. DNA damage may be a causal factor in the initiation of pathology and supports the hypothesis. Human leukocytes were employed to extend the biochemical hypothesis at the celhrlar level (Meier et al., 1984). Exposure of mixed human leukocytes to HD caused dose- and

MOLECULAR

BASIS

FOR

time-dependent decreases in the NAD+ content, which were prevented by the poly(ADPribose)polymerase inhibitor 3-methoxybenzamide. Although the maximum NAD+ depression in mixed human leukocytes was only approximately 40%, the maximum NAD+ decrease in purified lymphocytes was greater than 90% (unpublished data). Lymphocytes, which comprise about 40% of the total population, are the only known leukocytes containing poly(ADP-ribose)polymerase (Hayaishi and Veda, 1982). These findings provide further support for the involvement of this enzyme in the HD-induced NAD+ depletion. Additional studies with leukocytes were performed to determine if the loss of NAD+ content by HD can also be prevented by enhancing the synthesis of this cofactor. Both nicotinamide, an inhibitor of poly(ADPribose)polymerase and a precursor of NAD+ synthesis, and niacin, a precursor of NAD+ synthesis which does not inhibit the enzyme (Purnell and Whish, 1980; Hayaishi and Veda, 1982) were able to boost the NAD+ content of HD-treated leukocytes above control levels. The ability of niacin to prevent the HD-induced decrease in NAD+ without inhibiting poly(ADP-ribose)polymerase activity, which is required for DNA repair (Shall, 1982), may have therapeutic implications. Niacin could permit cells to continue repair of their HDdamaged DNA without interfering with critical energy-dependent metabolic processes, and thereby promote cell recovery. A major consequence of low NAD+ levels would be inhibition of glycolysis (Hayaishi and Veda, 1982) which was found by earlier investigators to be associated with HD injury (Dixon and Needham, 1946; Renshaw, 1946). Glycolysis inhibition in HD-treated skin is delayed by several hours and is followed shortly thereafter by the appearance of first morphologic signs of cutaneous injury. Dividing cells require more NAD+ than nondividing cells (Hayaishi and Veda, 1982) and basal keratinocytes, the propagating epidermal cells, would be expected to be greatly affected by

MUSTARD

VESICATION

s147

low NAD+. Skin pathology, similar to that ob served by HD exposure is also a prominent feature in pellagra, a niacin deficiency disease. The involved pellagra skin shows NAD+ losses similar to the depletion reported here in HDtreated skin (Findlay, 1963). Glycolysis inhibition and associated cytotoxicities were also noted when cultured human and other mammalian cells were exposed to mustards (Roitt, 1956; Fraser, 1960). These reactions could be prevented by nicotinamide, an inhibitor of poly(ADP-ribose)polymerase and a precursor for NAD+ synthesis. Although these findings strongly implicate NAD+ depression in glycolysis inhibition, further investigations are needed to determine this exact relationship in HD-treated skin. Loss of cellular NAD+, which results from stimulated poly(ADP-ribose)polymera~ activity, would tend to enhance NADP+-dependent metabolic reactions since this cofactor is not consumed by the enzyme (Hayaishi and Veda, 1982). Enhanced activity of the NADP+-dependent hexosemonophosphate shunt could be expected due to accumulation of glucosed-phosphate (Lehninger, 1979). An increase in shunt activity has been shown to result in a high rate of protease release, including that of a plasminogen activator (Schnyder and Baggiolini, 1980). The formation of plasmin from the interaction of plasminogen activator with plasminogen [which is concentrated at the epidermal-dermal junction (Isseroff and R&in, 1983)] could be responsible for vesication caused by HD. A similar mechanism has been implicated in blister formation in bullous skin diseases such as bullous pemphigoid (Hashimoto et al., 1983). An understanding of the molecular and pathogenic mechanisms of vesication by HD is critical to the development of rational approaches to therapy. The long latent period which precedes the development of irreversible pathology lends itself to exploitation for therapy. The biochemical hypothesis presented here could be the basis for devising therapeutic

S148

PAPIRMEISTER

strateties to attenuate. modify . and/or .- txevent the production of highly incapacitating, slow healing HD lesions. Although portions of the hypothesis are supported by current findings, further work is required for complete validation.

ET AL. for the loss of epidermal cohesion and blister formation. J. Exp. Med. 157,259-272. HAYAISHI, L., AND UEDA, K. (1982). ADP-Ribosylation Reactions, Biology, and Medicine. Academic Press,New Y0rk ISSEROFF,R. R., AND RIFKIN, D. B. (1983). Plasminogen is present in the basal layer of the epidermis. J. Invest Dermatol.

ACKNOWLEDGMENTS The authors thank CPT Warren Jederberg and Specialist Victoria A. Smith for helpml discussions and for providing the athymic nude mice bearing human skin grafts. Thanks are also due to MAJ Clifford J. Hixson for his pathology support, to Mrs. J. E. Kilduff and Specialist L. M. Graham for expert technical assistance, and to Mrs. Donna M. Brown and Mrs. Linda E. Brown for help in manuscript preparation.

REFERENCES

80, 297-299.

JACOBSON,E. L., AND JACOBSON,M. K. (1976). Pyridine nucleotide levels as a function of growth in normal and transformed 3T3 cells. Arch. Biochem. Biophys. 175, 627-634. KOHN, K. W., STEIBIGEL,N. H., AND SPEARS,C. L. ( 1965). Crosslinking and repair of DNA in sensitive and resistant strains of E. coli treated with nitrogen mustard. Proc. Natl. Acad. Sri. USA 53, 1154- 1160. LAWLEY, P. D., AND BROOKES, P. (1965). Molecular mechanism of the cytotoxic action of difunctional alkylating agents and of resistance to this action. Nature (London)

206,480-483.

LEHNINGER, A. L. (1979). Biochemistry. Chap. 16. Worth, New York. LINDAHL, T. (1979). DNA glycosylases,endonucleases for apurinic/apyrimidinic sites, and base excision-repair. Prog. Nucleic

AUERBACH, C., AND ROBSON, J. M. (1946). Chemical production of mutations. Nature (London) 157, 157. BROOKES, P., AND LAWLEY, P. D. ( 196 1). The reaction of mono- and dihmctional alkylating agents with nucleic acids. Biochem. J. 80,496-503. CONNORS,T. A. (1975). Mechanism of action of 2-chloroethylamine derivatives, sulphur mustard, epoxides and a&dines. In Antineoplastic and Immunosuppressive Agents II, (A. C. Sartorelli and D. Cl. Johns, Eds.), pp. 18-34. Springer-Verlag, Berlin. DIXON, M., AND NEEDHAM, D. M. (1946). Biochemical research on chemical warfare agents. Nature (London) 158,432-438. FINDLAY, G. H. (1963). Epidermal diphosphopyridine nucleotide in normal and pellagrous Bantu subjects. Brit. J. Dermatol.

75, 249-253.

Fox, M., AND SCOTT, D. (1980). The genetic toxicology of nitrogen and sulphur mustard. Mutat. Rex 75, 13 l168. FRASER,I. M. (1960). Effect of mechlorethamine on tumor glycolysis and diphosphopyridine nucleotide content. Fed, Proc. 19, 345. GROSS, C. L., MEIER, H. L., PAPIRMEISTER, B., AND BRINKLEY, F. B. (1985). Sulfur mustard lowers NAD+ levels in human skin grafted to nude mice. Submitted for publication; preliminary report in Fed. Proc. 43, 2452. HASHIMOTO, K., SHAFRAN, K. M., WEBBER, P. S., LAZARUS,G. S., AND SINGER,K. H. ( 1983). Anti-cell surface pemphigus autoantibody stimulates plasminogen activator activity of human epidermal cells. A mechanism

Acids Res. Mol.

Biol. 22, i35- 192.

LUDLUM, D. B., TONG, P. A., MEHTA, J. R., KIRK, M. C., AND PAPIRMEISTER,B. (1984). Formation of 06ethylthioethyl deoxyguanosine from the reaction of chloroethylethyl sulfide with deoxyguanosine. Cancer Res. 44,5698-570 1. MCADAMS, A. J., JR. (1956). A study of mustard vesication. J. Invest. Dermatol. 26, 317-325. MEIER, H. L., GROWS,C. L., AND PAPIRMEISTER,B. ( 1984). The use of human models for validating the biochemical mechanism of mustard-induced injury and for developing and evaluating therapeutic regimens to prevent mustard gas incapacitation. Proc. Army Science Con&, West Point, New York. MISKIN, R., AND REICH, E. (1980). Plasminogen activator. Induction of synthesis by DNA damage. Cell 19,2 17224. PAPIRMEISTER,B. ( 196 1). On the mechanism of inhibition of T2 bacteriophage by mustard gas. Edgewood Arsenal Spec. Publ. 2-45.

PAPIRMEISTER, B., AND DAVISON, C. L. (1964). Elimination of sulfur mustard induced products from DNA of E. coli. Biochem. Biophys. Rex Commun. 17, 608617. PAPIRMEISTER,B., WESTLING, A. W., AND &HROER, J. (1969). Relevance of DNA damage to the vesicant action of sulfur mustard. Edgewood Arsenal Tech. Pub/. No. 4294. PAPIRMEISTER, B., DORSEY, J. K., DAVISON, C. L., AND GROSS, C. L. (1970). Sensitization of DNA to endonuclease by adenine alkylation and its biological significance. Fed. Proc. 29, 726.

MOLECULAR

BASIS FOR MUSTARD

PAPIRMEISTER, B., GROS, C. L., PETRALI, J. P., AND HIXSON, C. J. (1984a). Pathology produced by sulfur mustard in human skin grafts on athymic nude mice. I. Gross and light microscopic changes. J. Toxicol.-Cut. & Ocular Toxicol. 3, 311-393. PAPIRMEISTER, B., GROSS, C. L., PETRALI, J. P., AND MEIER, H. L. (1984b). Pathology produced by sulfur mustard in human skin grafts on athymic nude mice. II. Ultrastructural changes. J. Toxicol-Cut. & Ocular Toxicol. 3,395-410. PRICE,C. C., GAUCHER,G. M., KONERO, P., SHIBAKOWA, R., SOWA, J. R., AND YAMAGUCHI, M. (1968). Relative reactivities for monofunctional nitrogen mustard alkylation of nucleic acid components. B&him. Biophys. Acta 166,327-359. PIJRNELL, M. R., AND WHISH, W. J. D. (1980). Novel inhibitors of poly(ADP-ribose) synthetase. Biochem. J. 185,775-777. RANKIN, P. W., JACOBSON,M. K., MITCHELL, V. A., AND BUSBEE, D. L. (1980). Reduction of nicotinamide adenine dinucleotide levels by ultimate carcinogens in human lymphocytes. Cancer Res. 40, 1803- 1807. RENSHAW, B. (1946). Mechanisms in production of cutaneous injuries by sulfur and nitrogen mustards. In Chemical Warfare Agents and Related Chemical Prob-

VESICATION

s149

lems, Vol. 1, Chap. 23, pp. 479-5 18. U.S. Office of Scientific Research and Development, National Defense Research Committee, Washington, DC. ROBERTS, J. J., BRENT, T. P., AND CRATHORN, A. R. (197 1). Evidence for the inactivation and repair of the mammalian DNA template after alkylating by mustard gas and half-mustard gas. Eur. J. Cancer 7,5 15-524. ROITT, I. M. (1956). The inhibition of carbohydrate metabolism in as&es-tumor cells by ethyleneimines. Biochem. J. 63, 300-307. Ross, W. C. J. (1962). Biological Alkylating Agents: Fundamental Chemistry and Design of Compounds for Selective Toxicity. Buttenvorths, London. SCHNYDER,J., AND BACGIOLINI, M. (1980). Induction of plasminogen activator secretion in macrophages by electrochemical stimulation of the hexose monophosphate shunt with methylene blue. Proc. Nat/ Acad. Sci. USA 77,414-417. SHALL, S. (1982). ADP-ribose in DNA repair. In ADPRibosylation Reactions, Biology, and Medicine (0. Hayaishi and K. Veda, Eds.), pp. 477-520. Academic Press, New York. WHEELER, G. P. ( 1962). Studies related to the mechanism of action of cytotoxic alkylating agents. Cancer Res. 22, 65 l-688.