First Symposium on Chemical-Biological Correlation, May 26-27, 1950 (1951)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

CORRELATIONS BETWEEN THE CHEMICAL STRUCTURE AND BIOLOGICAL ACTIVITY OF ARSENOSOBENZENES George O. Doak, PhD. and Harry Eagle, M. D. Syphilis Research Laboratory, USPHS, and University of North Carolina, School of Public Health, Chapel Hill, North Carolina, and the Micro- biological Institute, National Institutes of Health, Bethesda, Maryland.

I. THE CHOICE OF COMPOUNDS FOR STUDY Only three types of organic compounds containing arsenic linked directly to carbon have been used successfully in the treatment of spirochetal and protozoal diseases: the arsonic acids, the arseno compounds and the arsenoso compounds or arsenoxides (cf. Fig. 1). The simple organic radical, R, attached to arsenic is usually benzenoid, although similar compounds in which R is a heterocyclic ring, e.g. pyridine, are known to be highly active" against trypanosomes. Compounds in which R is aliphatic or alicyclic have also been tested, but are generally either completely devoid of activity8 or only weakly active. Four other types of organic arsenicals containing benzene rings are shown in Fig. 2. A few examples of these have been tested in various laboratories, but none has been found which compares in activity with the compounds illustrated in Fig. 1. ' Accordingly, only the latter will be considered in the following analysis of the relationships between the chemical structure and biological activity of organic arsenicals. An important consideration is whether all three types are active per se, or must first b« modified in the animal host. With the arsonic acids this question is easily answered. Ehrlich, in his original studies, showed that although the arsonic acids were inactive iri vitro, they were therapeutically highly active, and concluded that the arsenic was reduced to an active trivalent form by the body tissues. 28 He therefore prepared both arseno and arsenoso compounds, and found both types active in vitro The arsenoso compounds, while extremely active, were discarded because of their high toxicity; but the arseno compounds, notably arsphenamine and neoarsphenamine, found wide therapeutic application. Confirming and extending Ehrlich's classical studies, Voegtlin and Smith in 1920 demonstrated that with both arsonic acids and arseno compounds there was a latent period before trypanosomes disappeared from the blood of infected animals, while arsenoso compounds produced an immediate and striking decrease (Fig. 3). These results were interpreted to mean that the arsenoso compounds were directly trypanocidal, while the arsonic acids had to be reduced and the arseno compounds oxidized to the arsenoso form in the animal body before becoming active. Other workers, however, found that the arsphenamines as such, and without further modification in the animal body, were active against trypanosomes and spirochetes. 62 The discrepancy was resolved by later in vitro studies with Treponema pallidum. ' •* Although the arsphenamines were apparently quite active against this organism when tested by the usual anaerobic technic, when both the dissolution and testing of the drugs were carried out under an atmosphere of nitrogen, the activity of arsphenamines was reduced to a fraction of its previous value. The results of one experiment are given in Table I. The arsphenamines as such are clearly not directly spirocheticidal, but in the course of solution and dilution prior to testing there is sufficient oxidation to the arsenoso compound to make the solution active. The slight direct activity of arsphenamine is explainable on the basis of the known arsenoso content of this drug as marketed. In general, the arseno compounds are amorphous powders which are almost impossible to obtain in a state of chemical purity, and exist in several states of molecular aggregation. Both factors have been shown to influence their parasiticidal activity and toxicity to a marked degree. *^ The fact that both the arsonic acids and arseno compounds must be modified in the animal body before exerting their therapeutic action makes it difficult, if not impossible, to correlate their chemical structure and biological activity unless it is assumed that all the compounds are modified to the same degree. This assumption is certainly not justified in the case of the arsonic acids, which are excreted in part unchanged. Cohen, King and Strangeways attempted to correlate the toxicity of arsonic acids with the rate of oxidation of the corresponding arsenoso compounds, 9 on the assumption that the rate of reduction of the arsonic acids in vivo might be in the same order as the reverse effect. No such correlation was found. In later studies, Eagle, Hogan, Doak and SteinmanZO also found no regular correlation between the toxicity of arsonic acids and the corresponding arsenosobenzenes, or between their trypanocidal activities in vivo (Table II).

ARSONlC AClDS ARSENO- ARSENOSO- RAsO(OH)2 AsO(OH)2 0 RAs As AsR At RAsO AsO H2N NH2 NHCH2CONH2 TRYPARSAMIDE OH OH OH ARSPHENAM1NE BASE OXOPHENARSINE (MAPHARSEN) Fig. 1 Types of Aromatic Arsenicals Generally Used in Therapy AsO (OH) DlPHENYL ARSlN1C AC1D AsO As-O-As B1S-(DlPHENYL ARS1NE) OXlDE As TR1PHENYL ARS1NE OX1DE TRlPHENYL ARSINE Fig. 2 Illustrative Aromatic Arsenicals Which Have Not Been Generally Used

10 TRYPANOC1DAL ACT1ON OF As7 AND Asffl 1N (SEMl-DlAGRAMMATlC, AFTER VOEGTL1N AND SMlTH) VIVO I ARSACET1N II NEOARSPHENAMlNE ffl ARSENOPHENYL6LYClNE ARSPHENAMlNE I OXOPHENARSlNE Fig 3 3 4 T1ME 1N HOURS The Trypanocidal Action of Pentavalent and Trivalent Arsenicals in Vivo (after Voegtlin and Smith58)

11 TABLE I THE EFFECT OF OXYGEN ON THE TREPONEMICIDAL ACTIVITY OF ARSPHENAMINES1 3b Drug Conditions of Test Concentration necessary to immobilize 50% Relative Treponemicidal Activity * or organisms Oxophenarsine Aerobic -anaerobic^/ Anaerobics' 1 180,000 1 180,000 100 100 Neoarsphenamine Aerobic -anaerobic Anaerobic 1 84,000 1 1.100 100 1 Arsphenamine Aerobic -anaerobic Anaerobic 1 53,000 1 13,000 100 25 Sil ver- Arsphenamine Aerobic -anaerobic Anaerobic 1 80,000 1 10,500 100 13 I/Compound dissolved aerobically, and solution tested anaerobically a/Solution dissolved and tested under N2 atmosphere TABLE II THE POOR CORRELATION BETWEEN (A) THE TOXICITIES OF ARSONIC ACID AND THE CORRESPONDING ARSENOSO COMPOUNDS (B) THE TRYPANOCIDAL ACTIVITIES IN VIVO OF ARSONIC ACIDS AND THE CORRESPONDING ARSENOSO COMPOUNDS (Summarized from 18a, b, 20 and 211l) A. Toxicity B. Trypanocidal Substituent Group !.1>r,o in White Mice, mg./kg. CD50 *n White Mice, mg./kg. AsV As"I Ratio of LD50 Doses AsV As1" Ratio of CD50 Doses g-CONHCH2CONH2 1750 15 115 380 1.4 270 p-NHCH2CONH2 950 19 50 p-NHCONH2 1025 11 93 305 1. 16 267 p-OCH2CONH2 745 9. 5 78 80 0. 82 98 3-NHCOCH3-4-OH 765 6 127 565 1.91 296 3-NH2-4-OH 775 17 46 85 0. 62 137 p-SO2NH2 595 18 33 170 2.0 85 p-OH 550 2 275 "Melarsen" 260 12 217 30 0.048 625 3-NH2-4-CONH2 290 15 19

12 TABLE II (Cont. ) Substituent Group LD5g in White Mice, mg. /kg. CD5o in White Mice, mg. /kg. AsV AsIII Ratio of AsV As"1 Ratio of LD5o Doses CD50 Doses 3-NH2-4-(CH2)3COOH 140 12 12 >190 1.26 >150 p-NH2 165 1. 5 110 TABLE III REPRESENTATIVE MONOSUBSTITUTED ARSENOSOBENZENES TESTED FOR BIOLOGICAL ACTIVITY (Summarized from 18a, b, 20, 21.i, b) AsO Single Grouping on End of Substituent Chain AsO Substitution in Single Group AsO 0 O Q, (X) X-R o-, m-,p-NO2 p-OCH3 p-CONHCH3 p-CH2OH o-,m-.j,-Cl p-N(CH3)2 p-CON(CH3)2 p-OCH2CH2OH g-,m-,£-OH p-CH2NH2 p-SO2NHCH3 p-CONHCH2COOH o-, m-, p-NH2 p-C0H4NH2 p-SO2N(CH3)2 o-, m-,p-CH3 p-CONHC6H4NHCOCH3 o-,m-,p-COOH m-,£-NHCOCH3 p-(CH2)xCOOH p-CONHC2H4OH p-OCOCH3 p-CONHCH2CN o-,p-SO3H p-(CH2)xCONH2 m_-,p_-CONH2 p-NHCH2CONH2 m-,p-SO2NH2 p-CCH2CCNH2 p-F p-CCNHCH2CONH2 It is apparent that many compounds listed as X-R could just as well have been listed as R'-X, depending on which is considered the functional group.

13 The arsenoso compounds therefore appear to be the drugs of choice in an attempted correlation of chemical structure and biological activity. Unfortunately, most pharmacological studies on the aromatic arsenicals have dealt with arsonic acids and arseno compounds, on the assumption that the arsenoso compounds were too toxic for use. This assumption was questioned by Tatum and Cooper, 54 who demonstrated that m-amino-p-hydroxyphenylarsenoxide, which they named "mapharsen", possessed a higher therapeutic index against Treponema pallidum than the arsphenamines. Prior to the advent of penicillin, this compound was widely regarded as the drug of choice for the treatment of syphilis. Eagle, Hogan, Doak and Steinman later found a large series of arsenoso compounds to possess a higher therapeutic index than the corresponding arsonic acids against Trypanosoma equiperdum, although Gough and King had previously report- ed no essential difference between the two classes of compounds in a much smaller series. 33 The following sections will therefore describe the results obtained in a study of the toxicity and parasiticidal activity of a large series of arsenosobenzenes, considered in relation to the substituent groups of the phenyl ring. '8, 33 II. THE TOXICITY, PARASITICIDAL ACTIVITY IN VITRO AND THERAPEUTIC ACTIVITY IN VIVO OF ARSENOSO COMPOUNDS, CONSIDERED IN RELATION TO THEIR STRUCTURE A. Compounds Tested and Technics of Assay Table III lists some of the mono-substituted derivatives of arsenosobenzene synthesized and tested in this laboratory for the purpose of this study. " The aryl arsenoso compounds are prepared by the reduction of the corresponding arsonic acids, using a variety of reducing agents. The first column lists some of the simple substituents. The second and third columns list several representative compounds in which some of these groups were blocked by substitution (X-R), or in which the functional groups were extended on a side chain (R-X). A number of di-substituted derivatives of arsenosobenzene were also prepared. 12 When the two substituents are the same, there are six possible isomeric compounds; when they are different, there are ten possible isomers. Some of the compounds in this group which were prepared and tested are shown in Table IV. In addition to the derivatives of arsenosobenzene, a limited number of compounds were prepared using naphthalene, biphenyl, pyridine and similar ring structures. The toxicity of these compounds was tested in rabbits and mice. Their parasiticidal activity was studied with Treponema pallidum and Trypanosoma equiperdum, both iri vitro and in vivo: and a few representative compounds were further tested ill vitro with Trypanosoma cruzi and Leishmania donovani. For the assay of treponemicidal activity uj vitro, an acute testicular syphiloma was emulsified, and graded concentrations of arsenicals added to the particle-free extract. At the end of one and one-half to four hours the action of the arsenical was stopped by the addition of cysteine, and the concentration of drug which had immobilized 50 per cent of the organisms determined by microscopic observation. The activity was referred to that of a standard compound simultaneously tested, usually the unsubstituted arsenosobenzene. '5 Toxicity was determined as the LD50 dose on single intraperitoneal injection in white mice. In this way a "hybrid" therapeutic index was obtained, defined as the ratio of molar treponemicidal action in vitro:molar toxicity in white mice. A considerable number of compounds were also tested in syphilitic rabbits, and a therapeutic index obtained in terms of the LJD50/CD50. The high degree of correlation between the two indices reflects the fact (1) that the toxicity of the compounds in mice paralleled their relative toxicity in rabbits (Fig. 4), and further, that the curative action in animals paralleled and was presumably determined by their direct treponemicidal action (Fig. 5).

TABLE IV REPRESENTATIVE DISUBSTITUTED ARSENOSOBENZENES TESTED FOR BIOLOGICAL ACTIVITY12, 18b NH2 OH Cl CH3 COCH CONH2 NHCOCH3 NH2 X *6)v X JC — X OH X X X X Cl X CH3 X NO2 X X NHCOCH3 X X 'Number of isomeric compounds prepared THE CORRELATlON BETWEEN THE TOXlClTY OF ARSENOSOBENZENES IN MlCE AND RABBlTS 100 B 50 t 20 10 i 5 10 20 50 IOO RELATlVE TOXlClTY lN RABBlTS Fig. 4 The Correlation Between the Toxicity of Araenosobenzenes in Mice and RabbitslS, IBa, b, 2l

CORRELATION BETWEEN ACTlVITY OF ARSENOSOBENZES ON TREPONEMA PALLlDUM JN VIVO AND IN VITRO MG./KG. 20 0 K O -j £E uj§: gco O Ul 10 i 10 15 20 30 RELATIVE TREPONEMlCIDAL ACTIVlTY lN VITRO Fig. 5 Correlation Between Activity of Arsenosobenzenes on Treponema pallidum in Vivo and in Vitro 15' I8a' b- 21

16 With Trypanosoma equiperdum, the in vitro assay was similar to that used in the assay against Treponoma pallidum, and the therapeutic activity was tested in both mice and rabbits.'" Reasonably good agreement was again found between the trypanocidal activity of arsenoso compounds in vitro and in vivo, indicative of the fact that the therapeutic activity of these compounds is determined by their direct trypanocidal action. In view of the generally good correlation between the direct parasiticidal action of the arsenosobenzenes and their therapeutic action in vivo, the following discussion in the correlation between structure and biological activity is based largely on the in vitro data. B. Biological Activity of Substituted Arsenosobenzenes The effect of single substituents on the toxicity and treponemicidal activity of arsenoso- benzene is shown in Fig. 6 and Table V. The latter shows also their direct trypanocidal activity (T. equiperdum). Despite the wide variations encountered between closely related compounds, the data permit certain broad generalizations with respect to the correlation between chemical structure and biological activity. l . "Inert" Substituents There were a number of "inert" substituents (e.g. -Cl, -NO2, -NHZ, -OH, -CH3, -F), which did not significantly affect either the toxicity or parasiticidal activity of arsenosobenzene. The activity:toxicity ratios of the resulting compounds were therefore substantially the same as that of the highly active and highly toxic parent arsenosobenzene. Like the latter, these compounds are apparently general protoplasmic poisons, with no selective or specific effect on a particular parasite The position of such substituents on the benzene ring had relatively little effect. Thus, the ortho. meta and para arsenosotoluene compounds had relative treponemicidal activities of 84, 97 and 102, respectively (referred to the parent arsenosobenzene as 100), relative toxicities of 88, "00 and 118, and activity:toxicity ratios 0.95, 0.98 and 0.83. 2. Effects of Acidic Substituents Most acidic substit'ients markedly depressed both the treponemicidal and trypanocidal activity of arsenosobenzene, resulting in compounds with a low level of activity (cf. Fig. 1 and Tables V and VI). The toxicity of these compounds was also reduced; and although the individual compounds varied markedly in this respect, the toxicity was usually not reduced to the same degree as the parasiticidal activity. In consequence, the therapeutic indices of the acid-substituted compounds were even less than that of the parent compound. There were, however, a few import- ant exceptions to this generalization, notably with the \-(p-arsenosophenyl)-butyric acid-, > -(p_ arsenosophenyl)-valeric acid-, and ( -(£-arsenosophenyl)-caproic acid- substituted com- pounds (cf. Table VI). The first two compounds were actively trypanocidal, both m vivo and in YJirfl, although they had no significant activity against T. pallidum. Conversely, the last named compound had considerable treponemicidal activity, but was almost inert against T_- equiperdum. These exceptions emphasize the danger of generalizing even on the basis of a large number of closely related compounds. They emphasize also the specificity of the reaction of some arsenicals with particular parasites (cf. page 33). a. Evidence for Modification of Acid-substituted Arsenoso Compounds in Vivo. There is reason to believe that with these acidic compounds the observed toxicity is not that of the compound itself, but of a derivative formed in the body. (1) As is discussed in a following section, the toxicity of arsenicals is apparently determined by the degree to which they are bound by the host tissues. 37 There was a high degree of correlation between the rate of excretion of an arsenical and its toxicity, the least toxic compounds being excreted at the fastest rate. Further, the degree to which arsenoso compounds were bound by red cells in vitro proved a generally reliable measure of their systemic toxicity The acid-substituted compounds were bound to only a negligible degree by erythrocytes in vitro; and correspondingly, after intravenous injection they were at first excreted rapidly. These data implied a very low toxicity, far lower than was actually the case. However, after a few hours, the rate of excretion fell off abruptly, suggesting that the arsenical had been modified to a more toxic form, which was then retained in the body. (2) After lethal doses of acid-substituted compounds, death was usually delayed as compared with similarly toxic doses of other arsenoso compounds, not containing acidic groups (Fig. 7).

17 u. iI lI _ 1 : 1 8 1 o - *.!.. .•*•••• u ^ O CO O CO ^ r U— ^0 till CO S — Ljj > Z if a u Z M • *••+•»»• •• UJ 0 N Z H < Z O CO U] *o UJ IE CE N a -J CO ' ' Z < 0 O CO ». .. NOSOBE AoxiUity ^S ^J 1-iJ ^O i 1 Ul CO 1 0 < „ Q. Q cc c U] O « ******* *" * ^^^1^^^ <t o ce §0 N rt I X* o — o x o z O O Z O OC CE a « 'e 1 1 1 i i 1 1 UJ u l1 l1l 1 H *J n 800 OO*O cg — . O *fi W - t — — - '•2 V « B A J3 V 1/) B CD *• J! ?S ] m .S 0 _ V o- a: in B nil - S §§•§§• • ••f • f • %^k • JJ^»* O ° h tl 0 S 1 1 1 1 O o. u H 3 1*1 sl i! UJ V U] . y / »•• V a: oc o v L- KJ 1^ l^t O ^j O CO Ul .r* n ^ « — UJ cr a: i i u. « ^ x m UJ > a y 0 0 cc I- CO o § 0 • i w* 1 • «• CO i° UJ Z CO Ul O CO LL) If Is 5 J^ U 5 H < 8 £ Lll|| «J« • • * S H so X O ~ n I X 2 Q Ox O Z S O 0 2 0 fic £ 00 i i i i i i i ll Ill 1 £ o o o o m CM — o « «y -

18 o o. E o O a *t- .S8 .2 .3 V < OJ -O E o Q Q - 00 — U, uo 3uujno3Q 1eio_[

19 U m < H U Z .o 3 1| | H U .S ui <; ^ a J o rft <: *J "a -a u M S M 5" iM O ^O - si 5 ! i a* ^ fsJ ^Q r g iM ^- r-' °° "3 >, 5 ; O 00 O r^ iM* go ± * .2 > ** O -' O o rt tj OS A jl r^ ** ^* .t^ J o 00 r- o *t" ^O fs3 ^ sO in i~ * X fO 00 . So iA « o S o .2 S " O o o iM' £3 | u § « .£ "2 .rt S f_^ O S'TS ^-* ' ^ ri ^• r g 'u H '5 C .^ « j r- m m *"* O ^ * M H - u i4 * ? 1 11 C" ^ m ^ sr * oo 00 c." ° § dg. - 7 °° rn N C-t- V o - 1 H1 u H— h H .^ Is j m — . 7 oo ^[IS so in i so .*i " " ™. o a d. iM r- H .5 fO sO in " ?ly X ^ E iM m s?s u o i i IK ml V X U w 8O X u y CSH, SO ,CSH <n i titueets fd5? x z ri iM X U X O Z X Z U A Z ' O zo o u O X z z .i«. U ffj 9l i 1 i ^^ 1 1 *•"•*• Q J3 3 lA V V >, - 3 2 •£ "2 v ." H « it X > M OO '^ .a TJ TJ g ^ u r U) a ^1 s-i <I ri — - TJ V 1 o a > * o o a 11

00 it U Q § U z U 0, H 10 n a a u > o 3 S < > H H U J 9 u D H y H « o 2 U U IX u t- c 113 rg „, i— . + i rg in 0 O rg m rs3 O in 00 — — C +- 1 0 o O O rg O rg o — . ^ o so o o so -a o V m m ^r •M Jl 4-1 aeum C H a M1 rg 25 « fO O m i L. I o o Urtizat o o o r- r g O O •*r o o rg C1 en o V rs] ^O *J E '3 .3 *J •o u *! N in in + 1 ffl a O •* r^ oo m ^* fM m o ,^ CJ 3 o o" • .r - i *- i - i - O r n r- ^ ^ o so "S a" o V in rg — rg a! H E 3 -o m r- rg — c* r- rg 5T "rt o ^ so" ^' ^t ^" rg O r^ 00 0O V O *0 a o i rs] — i*J — rg i * H i. r- m r- ^* — . r- rg 'u 0 in « — r-' 00 r- oo — ^£i r- r- m m r-' r-' i o .* .* .* - r g rg f M — H in fn m in m m ^* a fM V X X 0 v 0 0 -o ^H 0 L^ c X 8o§|? Comprt esubstituted pheeyl arseeo X X X K ^ o o o o S o o o o o u u u u o i^ m^Tj^i 'J O u u ~' I -COOH -CSH .CSH iMl iM U iM iT O U I I I 2 O Z Z Z CO D Ol ai < '-i cxi ai a| at cxi E| 01 m CX|fl (X|m .o 51 E g i £ '2 « * 7. c ~ - S 0n > * S.

These several lines of evidence indicate that the acidic group may be modified in the animal body, possibly by esterification or conjugation, to form a more toxic arsenical. The body mechanisms for defense against acid in this case apparently increase the reactivity of the compound with the tissues and increase its toxicity. b. The Effect of pH on the Parasiticidal Effect of Arsenoso Compounds. There is reason to believe that such slight parasiticidal activity as acid-substituted arsenpso compounds possess is largely a function of the non-ionized molecule, and that the ion is usually relatively inactive. In preliminary studies there was a rough correlation between the pK of the acid and its treponemicidal activity in vitro. 27 The effect of pH on both treponemicidal and trypanocidal action was, therefore, studied in a larger series of compounds. The parasiticidal activity of arsenosobenzene and of derivatives containing non-acid sub- stituents was independent of the pH of the testing medium over the total range of viability of the organism. With acid-substituted compounds, however, the activity against both Trypanosoma equiperdum and Treponema pallidum usually increased strikingly with the hydrogen ion concentra- tion (Fig. 8). This effect of pH was referable to the fact that the ions were relatively inactive as compared with the undissociated acid. The activity changed with pH in relation to the degree of ionization of the compound, and was predictable from the pK of the acidic group. With strong acids, such as the p-SO3H compound (pK = 2t), there was no demonstrable change in activity in the range of pH 8. 5-5. 5, throughout which the compound would be essentially completely ionized. With weak acids, however, the activity varied as much as a hundred-fold over the range pH 8. 5- 5. 5 (cf. right hand portion of Fig. 5). With several of the acids, the ionized form, while less active than the undissociated molecule, nevertheless had a definite parasiticidal action. With such compounds, the total parasiticidal activity at a given pH was therefore a sum of the (slight) activity of the ionized fraction and the activity of the non-dissociated free acid, the proportion of the latter at a given pH depending on the pK of the compound. Thus, the ions of the p-(CH2)3COOH, 3-NH2-4-(CH2)3COOH and 3-NO2-4-COOH arsenosobenzenes had an unusually high trypanocidal activity which contributed significantly to the total activity noted at e. g. pH 7. 0. One may, therefore, conclude that in a series of acid-substituted arscnoso compounds, the most effective against a given organism would be that compound with the highest pK (i.e. with the largest proportion in undissociated form at body pH), the ion of which was also moderately active against the parasite. In a series of compounds with substituent groups R-COOH, or R-SO3H, some interest therefore attaches to the effect of the radical R on the dissociation constant of the terminal acidic group (cf. Table VI). In the series -(CH2)nCOOH, there was a progressive increase of approximately 0. 25-0. 35 pH units in the pK of the compound with each additional methylene group. The high activity in vitro of some of these acid-substituted compounds against cultures of Schizotrypanum cruzi and Leishmania donovani is particularly to be noted. Infections with these two organisms have proved generally difficult to treat. In the case of T_. equiperdum the direct trypanocidal activity m vitro had proved an accurate index to therapeutic activity in vivo (cf. page 16). This unfortunately was not the case with these two organisms, perhaps because their intracellular localization in the infected animal rendered them inaccessible to the drug. In addition, preliminary and unpublished data indicate that the susceptibility of the intracellular parasites to arsenoso compounds may be less than that of the cultured flagellate. The mechanism of action of arsenoso compounds in general, and of these acid compounds in particular, and the explanation for the extraordinary difference between the parasiticidal activities of the free acid and its ion, will be discussed in a following section. 3. The Effect of Acid Amide and Ester Substituents. A third group of substituents concerning which it was possible to make some generaliza- tions with respect to biological activity were the acid amides Compounds so substituted were remarkably uniform in their toxicity and parasiticidal activity. As shown in Table VII, their direct trypanocidal and treponemicidal activities were intermediate between those of the generally inactive acids and the highly active arsenosobenzene. However, the most important effect of the amide substituents was a regular and marked decrease in toxicity, so that the amide-substituted compounds had activity:toxicity ratios as much as six times higher than that of the parent

22 too 50 20 1.0 o c s. as .» 0.2 cr O.1 0.05 - 0.02 v o-C0OH (5.5S) -* P-CH-CHC00H (4.7) 3-NH2-4-CoOH (4.7! (4 .01) * x-x p-OCH2C00H (3.3) p-50,H v o-S03H J. 90 85 " 8.0 7.5 7.0 6.5 6.0 5.5 5.0 pH 4.S Fig. 8 The Effect of pH on the Trypanocidal Activity of Ac id-substituted Arsenosobenzenes 17 Figures given in parenthesis are the pK1 values of the various acids

23 1 PQ Q 3 3 H S h O H U X J , Q H y § O 2 U B j: o .o v '3 o m oo oo o- r- m iM in m in in o moo^ moo fMrno orsi o —.-.—. 'O o ooo "^oo iin^O i i i o o O O O O i J o c cd .f i a in r- •& ^* oo oo in ^, N rMO fO — O O — rg O O oo J- ^j if l hi o OO' l00OOiOOOt ' *Oi V V O i O O 1 C S "" X U a C O i4 "c o rC E H 3 *O *^ n .3 •3 V in^* inr-oo rs3r- s o u a r-m ^O,O om — rsi — . ^f r^ > ' '3 o « o" ' i 00* rsl O i • rsJ O* i ' ' O i O O O rsJ i > S D o V u • 1* H1 -a ^ U d E v 3 rt .o V. a, o in cr 1 O o m^r-msom-i NiM. it-j- 00 T m °" - S ? o in f-i E 3 "H ^ o a H o inrgoc<i^^m fMcorsJin o^r- s C miUe o cosor^oin^^e^o-ino^coin co i*] in ^O .M .S l o « U i O X O x 1 « > r0 r^] XX N eZ Z X N 0 Unsubstituted prteeyt arseeoxide r g rg Z iM 00 fMQ 5 u u o z p-CONHC2H4OH p-SO2NHC2H4OH p-CONHCH2CN 3-NH2-4-CONH2 3-OH-4-CONH2 a 1 U El ai al a| a; al ai H li al ai ai ai El ai ai

24 arsenosobenzeneZ' (Table VII). A similar effect had been noted by Gough and King in studying amide -substituted arsonic acids in mice infected with Trypanosoma equiperdum. ." The favorable therapeutic effect of the amide groups was essentially independent of the particular acid used; sulfonamides and benzamides were similar in activity and toxicity. The position of the groups on the benzene ring was also unimportant. However, with both the p-CONH2 and p-SOZN?^ compounds, the integrity of the terminal amide grouping was usually essential for the favorable effect on toxicity. Substitution of the amide hydrogens with methyl or ethyl groups successively increased the toxicity at a greater rate than the activity, so that the therapeutic index of these substituted amides was considerably less than that of the amides. The £-SO2NH2, £-SO2NHCH3 and £-SO2N(CH3)2 arsenoso compounds had treponemicidal activities ±n vitro of 29, 72 and 1 12; relative toxicities of 4. 8, 18 and 93; and activity :toxicity ratios of 6. l, 4. 0 and 2. 3, respectively. Similar effects of substitution of the p-CONr?2 compound are summarized in Table VIII. In contrast to the CONHCH3 or -SO2NHC2H5 compounds, when the terminal grouping was one which was itself "eutherapeutic", i. e. depressed toxicity more than activity, the resulting compound was just as useful as the parent amide. Thus, the -CONHC2H4OH, -CONHCfcH4NHCOCH3 and -SO2NHC2H4OH compounds compared favorably with the parent amides with respect to toxicity and parasiticidal activity (cf. Table VIII). Unlike amide -substitution, the esterification of acid compounds greatly increased both their parasiticidal activity and their toxicity. In consequence, such ester-containing compounds, when stable, were no better therapeutically than the parent arsenosobenzene. Many hydrolyzed in aqueous solution, so that their biological activity was the same as that of the acid-substituted compound. 4. Miscellaneous Substituents (-NHCOCH3, -OCOCH3, -RNHCOCH3. -RNH2J Erratic results were obtained with arsenosobenzenes containing this group of substituents. Their parasiticidal activity varied between 3 and 78; their toxicity between 6. 7 and 76; and their activity :toxicity ratio varied between 0. 38 and 2. 6 times that of the parent arsenosobenzene. In no instance was a compound obtained which was sufficiently active and non-toxic to suggest the possibility of its therapeutic use for the particular infections tested. 5. The Importance of the Terminal Grouping in a Substituent Single substituents have thus been shown to have widely varying effects on the toxicity and parasiticidal activity of arsenosobenzene. In a number of the compounds, the substituent was a side chain of varying length. In general, and regardless of the length or nature of the side chain, the activity and toxicity of the compound were usually determined by the nature of the terminal functional groupings. Thus, the -CH3 group had been found to be "inert", while the -CONH2 and -SO2NH2 groups had a highly favorable effect on toxicity. The £-CH2CONH2, -(CH2)2CONH2 and p-(CH2)3CONH2 arsenosobenzenes all behaved as amides (Table VII). However, in the -CONHCH3 and the -CON(CH3)2 compounds the favorable effect of the amide grouping was diminished or abolished by substituting methyl groups for the amide hydrogens (cf. above). Similarly, when an amide hydrogen was replaced by a group carrying a terminal acidic group (e.g. the -CCNHCH2COOH), the properties of the compound were determined by the terminal acidic group, and not by the amide. 6. The Effects of Multiple Substituents1 8b The effects of multiple substitution on the toxicity and parasiticidal activity of arsenoso- benzenes were difficult to interpret. Only a relatively small number of compounds were tested (cf. Table IV); and in no case could the effect of double substitution be anticipated from the effects of the two groups acting singly. Thus, six of the ten possible arsenosoaminophenols were prepared and tested for treponemicidal activity (Table IX). The .-".!!., and -OH substituents separately, whether o-, m-, orj>- to th^ arsenoso group, had resulted in a uniform series of compounds, with treponemicidal activities varying only from 72 to 98, toxicities of 49 to 85, and activity:toxicity ratios of 1 . 0 to 1.46. The treponemicidal activity of the six aminophenol compounds was also remarkably uniform, varying only between 39 and 57; but their toxicities varied between 6. 94 and 78. 9, so that the activity rtoxicity ratios varied ten-fold, from 0. 54 to 5.5. The combination 3-NH2-4-OH, which is the well-known compound oxophenarsine

25 TABLE VIII THE EFFECT OF SUBSTITUTION IN THE -CONH2 GROUP ON THE TOXICITY AND TREPONEMICIDAL ACTIVITY OF P-ARSENOSOBENZAMIDE21a AsO Molar Toxicity Molar Treponemicidal Activity Ratio of Activity: Toxicity o R -CONH2 9.6 45 4. 6 -CONHC2H4OH 4. 8 25 5. 2 -CONHCH3 15 54 3.6 -CONHCH2COOH 16 0. 7 0.44 -CON(CH3)2 19 48 2. 5 -CONHC2H5 26 59 2.3 -CON(C2H5)2 64 53 0.83 -CONHC6H5 101 97 0.96 TABLE IX THE TREPONEMICIDAL ACTIVITY AND TOXICITY OF AMINOPHENOL-SUBSTITUTED ARSENOSOBENZENES l 8b Substituent Groups Molar Toxicity Molar Treponemicidal Activity Ratio of Activity: Toxicity 3-NH2-4-OH 6.9 42 6. 1 3-OH-4-NH2 10 41 3.9 2-NH2-3-OH 10 34 3. 3 2-OH-5-NH2 22 39 1.8 3-OH-5-NH2 74 57 0.78 2-OH-3-NH2 79 43 0.54

26 ("mapharsen") was the most favorable combination in the entire series. At the other extreme, the combination 3-NH2-2-OH gave a less favorable therapeutic index than arsenosobenzene itself. Any change in the 3-NH2-4-OH combination, either by the introduction of a third substituent, extending either group on a side chain, or substitution of the hydrogen in either group, abolished its highly favorable properties. Similarly variable and unpredictable results were obtained with the arsenosoamino- benzoic acids. Nine of the ten possible isomers were tested for direct trypanocidal activity. Unlike the example of the aminophenols just cited, the trypanocidal activities of these compounds varied 40-fold, from 0. 6 to 23, with the 3-NH2-2-COOH compound showing the highest activity. 20. 17 While no combination of two wholly inert substituents (-CH3, -NO2. -Cl, etc. ) resulted in compounds with a highly favorable activity:toxicity ratio, one cannot say unequivocally that further search would not have revealed such a combination. 7. Thioarsenites The reactivity of arsenicals with compounds containing thiol groups to form the corresponding thioarsenites was the subject of two patents by Kharasch in 1928 and 1934. " A number of thioarsenites have been used experimentally against trypanosomiasisZ4i amebiasis and filariasis, 44 |,,lt none has as yet gained general clinical use. A thorough study of the chemistry and pharmacology of the thioarsenites was undertaken by Barber2 and by Cohen, King and Strangeways. 1O These authors prepared a number of such compounds and determined their efficacy in experimental trypanosomiasis in white mice. A number of thioarsenites were also prepared in this laboratory by the condensation of arsenoso compounds with various thiols (cf. page 29). They are listed in Table X, together with their treponemicidal activity in vitro and their toxicity in white mice, both activities expressed relative to that of the unsubstituted arsenosobenzene as 100. As there shown, the thioarsenites were regularly less toxic and less activity parasiticidal than the parent compound; and in the treatment of rabbit syphilis their therapeutic index (the rap^tic^ activity) was no more favorable, and in several instances less favorable, than that of the corresponding arsenosobenzene. We have been unable to confirm the finding4Sc that the condensation of "mapharsen" with HA! . (2, 3-di- mercaptopropanol) results in a compound more toxic than mapharsen itself. The LD5o °f tne addition compound prepared in this laboratory was 230 mg. /kg. and 60 mM/kg. , as compared with 43 mg./kg. and 12.4 mM/kg. for mapharsen. Whether the activity and toxicity of the thioarsenites is determined solely by the amount of hydrolysis to the corresponding arsenosobenzene, or whether a thioarsenite may act directly has not been determined. It is true that an excess of thiol abolishes the parasiticidal activity of an arsenoso compound, both in vitro and in vivo (cf. Fig. 9), strongly suggesting that the hydrolysis of the thioarsenite may be essential for parasiticidal activity. It is, however, conceivable that the thioarsenite reacts directly with a tissue thiol, according to the equation: RAs(SR')2 + 2R"SHi=*RAs(SR")2 + 2R'SH Thioarsenite Tissue Tissue Free thiol thioarsenite thiol In such case an excess of thiol would inhibit parasiticidal activity by competing with the tissue thiol for the thioarsenite. C. The Selective Parasiticidal Action of the Arsenosobenzenes The parent arsenosobenzene, and those arsenoso compounds with "inert" substituents, were highly toxic to the animal host, and were uniformly active in vitro against all the organisms tested. When substituents were introduced which modified the biological activity of the compound, they often developed a high degree of specificity. The amide-substituted compounds were relatively non-toxic for the mouse or rabbit host, uniformly treponemicidal, highly variable in their action on Trypanosoma equiperdum, and only negligibly active against Leishmania donovani or

27 TABLE X THE RELATIVE TOXICITY AND TREPONEMIC1DAL ACTIVITY OF THIOARSENITES1 AND THE CORRESPONDING ARSENOSO COMPOUNDS183 Substituent Group Relative Molar Treponemicidal Activity in vitro Relative Molar Toxicity in White Mice SRl/ OA/ ^SR SRl/ Q.,o Ratio OA,o OA<*. Ratio 0-CH3 84 109 1. 30 88 90 1 . 02 £-OCH2CONH2 52 32 0. 61 9.0 7.3 0. 80 p-CONH2 45 39 0. 87 9.6 4.6 0.48 (a) 23 (b) 21 0.51 (a) 6. 1 (b) 5. 1 0. 64 0.53 0.47 p-NHCONH2 38 21 0. 55 8. 1 2.9 0. 36 p-CONHCH2CONH, : 32 17 0. 53 5. 7 1. 6 0. 28 m-SO2NH2 21 22 1.05 6. 1 3. 3 0. 54 p-S02NH2 29 24 0.83 4.8 3. 3 0. 70 3-NH2-4-OH 38 29 0. 76 6.9 4. 1 0.60 (c) 0.86 0.023 (c) 1.4 0. 2 m-COOH 13 1 1 0.85 16 15 0. 94 P-SO3H 3.4 3. 6 1.06 29 24 0. 83 .' All were dicysteinyl compounds, with the exception of the compounds indicated as (a), (b) and (c), which were the dithiosalicylate, dithioglycolate and dimercaptopropanol, respectively.

2S TABLE XI THE VARYING PARASITICIDAL ACTIVITY OF ARSENOSOBENZENES AGAINST A NUMBER OF ORGANISMS (Summarized from I8a, b, 20 and 21b) Substituent Group (R Q AsO) T. p.iliidum T. equiperdum L. donovani Schizotrypanum Normal "Arsenic -fast" cruzi p-CH3 102 102 70 96 47 p-CONH2 45 45 0.34 0. 08 0.07 p-OCH2CONH2 52 26 2 4. 30 0. 18 3-NO2-4-COOH 18 17 - 40 24 p-COOH 6. 7 0.45 - 0. 21 0.04 p-(CH2)3COOH 1.9 54 38 65 32 TABLE XII THE RELATIONSHIP BETWEEN THE TRYPANOCIDAL ACTIVITY OF ARSENOSOBENZENES AND THEIR COMBINING AFFINITY WITH THE ORGANISMS (AFTER EAGLE AND MUSSELMAN22) Original As concentration = 0. 166 mgs.%; No. of trypanosomes = 250 x 10°/cc; Vol. of 10^ sedimented organisms = 0.48 cc. Substituted Arsenosobenzene R-AsO Relative Trypanocidal Activity (referred to arsenosobenzene as 100) Conc, of As in Trypanosomes , mg. % Average ratio of ' trypanosomes 'Asi supernatant p-SOjH 0.06 0.41 2.4 p S02NHCH2CONH2 1.4 0.68 5.2 p-OCH2COOH 4. 5 1. 5 1 1 p-CONHCH2CONH2 15 5.7 53 3-NH2-4-OH 27 6. 5 87 p-S02N(C2H5)2 35 7.4 93 p-CONH2 45 7.7 113 m-OH 66 8.2 119 o-CH3 91 9.5 186 Unsubstituted arsenosobenzene 100 9. 7 224

Schizotrypanum cruzi. The acid-substituted compounds were generally only slightly parasiticidal. A few were, however, highly active against Trypanosoma equiperdum, Schizotrypanum cruzi and Le:.shmania donovani, and others were active against Treponema pallidum (cf. Table XI). The basis for this selective action is discussed in the following section. 111. MODE OF ACTION OF ARSENOSO COMPOUNDS A. The Reactivity of Arsenicals with Thiols Ehrlich originally suggested that drugs were effective only to the degree to which they were bound by "chemoreceptor" groups in the parasite, and laid down the thesis that "corpora non agent nisi fixantur. "Z6 He further suggested that thiol groups might determine that fixation, although no definite chemical reaction was postulated. 25 (In 1908 Friedberger noted that arsanilic acid when mixed with thioglycollic acid increased markedly in both toxicity and trypano- cidal activity. '" In the light of subsequent developments, it seems clear that this effect was due to the reduction of the pentavalent compound to the more active trivalent form, and is not relevant to the mode of action of the latter. ) The reaction between arsenosobenzenes and thiol compounds to form thioarsenites has been briefly discussed in a preceding section. In 1923 Voegtlin, Dyer and Leonard5? demonstrat- ed that trypanosomes contained -SH groups, that organic compounds such as cysteine and glutathione containing -SH groups inhibited the trypanocidal action of trivalent arsenicals in vitro. and that the injection of such compounds immediately before an otherwise fatal dose of an arsenoso compound prolonged the life of the animal. They concluded that the effect of the arsenoso compounds in vivo was based on the reaction: RAsO + 2R'SH »RAs(SR')2 + H2O and that the specific sulfhydryl compound which acted as the "chemoreceptor" in determining toxicity was primarily glutathione. The possibility that other sulfhydryl-containing compounds might be involved was considered by Rosenthal. ^?c The inhibiting effect of sulfhydryl compounds on arsenicals was extended to T.. pallidum by Eagle in 1939. who demonstrated that cysteine, glutathione and thioglycollic acid in excess abolished the treponemicidal action of arsenicals in vitro, as well as that of mercury and bismuth compounds. 14 in vivo also, thiol compounds in excess inactivated the arsenicals and abolished their trypanocidal and treponemicidal activity (cf. Fig. 9). The inactivation of arsenicals by thiols as outlined in the foregoing paragraphs involves two quite different mechanisms. Thiol compounds in excess react with arsenicals to form thioarsenites, inhibit the hydrolysis of the thioarsenites to the free arsenoxides, and prevent the interaction of these arsenicals with the host tissues or with parasites, thereby decreasing their toxicity and parasiticidal activity. In addition, however, even after arsenicals have combined with the tissue cells or parasites, thiol compounds may reverse that combination, and remove the arsenical from its combination with the cell. 15 When cysteine was added to a suspension of T_. pallidum which had previously been incubated with arsenicals and in which the larger proportion of the organisms had already been immobilized, not only was the treponemicidal action of the arsenical immediately halted, but a large proportion of the already immobilized organisms were revived. '5 The same phenomenon has been even more strikingly demonstrated with trypanosomes "killed" by arsenical, and then exposed to BAL (2, 3-dithiopropanol). 23 In vivo also, it is apparent from Fig. 9 that the parasiticidal action of arsenicals in vivo may be successfully reversed by thiol compounds given hours after the arsenical had been injected, and after the organisms had been given ample opportunity to combine with the arsenical injected.

30 The ability of thiol compounds to abstract arsenicals from the cell after they have already entered into combination with cell component is manifested not only in the revival of organisms already seriously damaged by the arsenicals, but may be demonstrated by the direct chemical analysis of the cells affected. Z3 The arsenical presumably leaves the cell in the form of the thioarsenite. Cur«d by Ar««nical in Absence of BAL CD 7O TRYPANOSOMlASlS lN WHlTE MlCE Rate of Cure by Mopharsen inoculum - 10s Age of infection — 24 hours Dosage of Mapharsen- 6mgYkg. Dosage of BAL used to inactivate arsenical - 40 mg./kg (Compos/It ol 7 individual npiramnts ) Total in tach group . : J « 3 « s i n n Time (Hours) When Therapeutic Action of Arsenical Was Terminated by lnjection of BAL Control. no BAL Fig. 9. The Inhibition of the Trypanocidal Action of Oxophenarsine in White Mice by 2. 3-Dithiopropanol (BAL)17b

31 B. The Reactivity of Arsenicals with -SH Groups in Enzymes The fact that arsenicals react strongly with thiol compounds to form thioarsenites, and the fact also that thiol compounds are able not only to prevent the combination of arsenicals with cells, but to remove the arsenic after it has already entered into combination with cellular constituents, do not however prove that a similar combination with cell thiols is primarily responsible for the toxic 'action of arsenicals. More recent work has indicated that this may actually be the case, and that the reactivity of arsenicals with -SH groups in protoplasm is the factor which determines their biological effects. It now seems unlikely that glutathione as such is necessarily or even primarily the cellular grouping affected. In 1933 several workers reported on the role of thio1 groups in the reversible deactivation of enzymes. Hellerman, Perkins, and Clark studied the inactivation of crystalline urease with mercurials such as phenylmercuric hydroxide;.^ and Bersin and Logermann noted that the activity of papain was destroyed by many oxidizing agents. ^ In both instances, the enzymatic activity was restored by the addition of either hydrogen sulfide or potassum cyanide. Maschmann had suggested that such reactivation of papain or cathepsin by H2S involved reductions of disulfide linkages in the enzyme to active thiol groupings essential for activity. Bersin then correlated the work of Cohen, King and Strangeways1O on the reaction between arsenicals and thiol compounds in vitro and the several previous indications as to the importance of thiol groups in the activity of many enzymes, and systemically studied the effects of a number of arsenic compounds on papain. * He showed that papain was inhibited by both arsenoso compounds and arsonic acids, and that this inhibition was reversed by glutathione. In explanation, he proposed the following alternative reactions: (1) 4 Enz-SH + RAsO3H2^± RAs(S-Enz)2 + Enz-S-S-Enz (2) 2 Enz-SH + RAs(OH)2;r± RAs(S-Enz)2 + H2O A marked difference in the inhibitory actions of p-CHsCONHC6^AsC^H2 and £-H2NCoH4AsO3H2 was explained on the basis of a difference in the equilibrium constant in reaction (1) above. In the United States, Barron and his co-workers have thoroughly investigated the inhibiting effect of arsenicals on a number of enzyme systems, 3 including the enzymes involved in nitrogen, fat, and carbohydrate metabolism Many were reversibly inhibited by arsenoso compounds and by the war-gas "lewisite" (C1CH = CHAsCl2); and in such cases, -SH groups in the enzyme protein were believed to be essential to their enzymatic activity. In England, working chiefly with lewisite, Peters and his co-workers have also investigated the reaction of enzymes and trivalent arsenicals. ^ They considered the primary site of attack of the arsenicals in the animal body to be on the pyruvate-oxidase system, which is particularly sensitive to trivalent arsenicals. Stocken and Peters^b reacted sodium arsenite and lewisite with kerateine in buffered solution and analyzed the precipitated protein for arsenic, sulfur and nitrogen. From 73 to 92 per cent of the bound arsenic had combined with sulfur in the ratio 1 : 2. Further, the protein sulfur was now resistant to oxidation, while the thiol groups in kerateine are normally highly susceptible. These workers therefore postulated that the arsenic had combined with two neighboring thiol groups on the protein molecule, i.e. that the protein contains a dithiol which is the primary point of attack by the arsenicals. Further, they stated that on physicochemical grounds the ring compound formed by the reaction of a dithiol with an arsenical (a) should be more stable than the straight chain thioarsenites formed with two molecules of a monothiol (b):

HS C RAsO + 1 •RAs HS C HS-C'-R RAsO + ' —• RAs<T | (b) HS—<J'-R ^^S—C'R There may be several gaps in this argument. No dithiol such as that postulated by Stocken and Peters has yet been identified as a constituent of kerateine or any other protein in the amounts implied by their experimental data. It seems unlikely also that arsenicals would react with disulfide groups in proteins, or that the major portion of the free monothiols in a protein molecule would be arranged in pairs. The fact that the arsenic:sulfur ratio of the compound formed between thiols and kerateine in solution approaches 1 : 2 is not conclusive evidence for the presence of a dithiol, since the arsenic molecule could be bridging two molecules of kerateine. Further, in the absence of resonance there seems to be no basis for assuming that a bond between two atoms in a ring is necessarily stronger than the same bond in a straight chain compound. In a recent paper, Slater has furnished strong evidence that the reaction between arsenicals and enzyme systems is probably more complex than that suggested by Peters and his co-workers. He has pointed out that while the reaction is supposed to be: RAsO + 2-HS-Enz ^±RAs (S-Enz2) + H2O, there may be no reactivation when the arsenical is removed or the reaction mixture is greatly diluted. However, whether or not the reasoning of Stocken and Peters was valid in its entirety, it led to a significant advance in the therapeutic use of arsenicals. Following up the previously cited data, Stocken and Thompson^ found that the dithiols were more effective than monothiols in protecting the pyruvate oxidase system in pigeon brain from arsenicals. In animals suffering lewisite burns or injected with toxic arsenicals, the dithiols greatly increased the rate of excretion of arsenic. ". " Further, Whittaker found that dithiols which he believed to react with arsenic to produce strained rings were much less effective than dithiols which formed a stable ring structure. ->° As an outcome of these studies, the English workers developed 2, 3-dimercaptopropanol (HS-CH2CHSH-CH2OH) for the treatment of lewisite burns. This compound (BAL = British Anti- Lewisite)5' has found extensive use not only in the local treatment of arsenical skin burns, but also in the systemic treatment of arsenic poisoning^, 41 and poisoning due to such heavy metals as mercury^, antimony'9 and gold.47a n ;5 ineffectual against cadmium and of questionable value in lead poisoning. 31 The use of BAL has been thoroughly reviewed by Stocken and Thompson. 5 Summarizing the above evidence, one may conclude that (a) arsenoso compounds combine with sulfhydryl compounds to form reasonably stable thioarsenites; (b) many enzyme groups contain sulfhydryl groups which are necessary in the intact state for the action of the enzyme, and such enzymes are inactivated by arsenoso compounds and other arsenicals; (c) this inactivation can be reversed by thiols, and in particular dithiols; and (d) thiol compounds, and particularly dithiols, protect animals and microorganisms against the toxic effects of arsenicals, can on occasion reverse the toxic action of arsenicals after they have already become manifest, and in such cases actually reverse the combination of the arsenicals with the cell or tissue. On the basis of these facts it is now generally believed that the action of the arsenicals rests on their ability to combine with 'hiol groups in essential enzyme systems, both in the host (toxicity) and in the cell of the invading microorganism (therapeutic activity). This may be expressed schematically as: RAsO + 2R'SH •RAs(SR')2 t H2O (A) arsenoso enzyme in host tissue or compund tissue or in cell thio- parasitic cell arsenite

33 Possible Explanations for the Varying Reactivity of Differ ,j Arsenicals Against the Same Organism, and of the Same Arsenical Against Different Organisms How does the theory of reactivity of arsenicals with enzyme -SH groups explain the experimental observations cited in detail in the preceding section (1) that chemically closely related arsenoso compounds may vary 100-fold in toxicity; (2) that different arsenicals may vary in their effect on a given microorganism, and (3) that the same arsenical may vary in its effect on different organisms? Thus, Tables VI, VII and XI compare the trypanocidal, treponemicidal, and leishmanicid.il activity of a number of arsenosobenzenes. The striking effect of \-(p_- arsenosophenyl)-butyric acid against trypanosome infections'^ and its inactivity against Treponema pallidum is but one example of the parasite-specificity of many arsenic compounds (cf. page 26). Kuhs and Tatum4O found that arsenicals capable of curing T_- lewisi infections possessed aliphatic side chains with acidic groups, while the arsenicals which cured Trypanosoma equiperdum infections usually contained basic groups. 1. The Possibly Varying Reactivity of Different Arsenicals and Thiols as the Basis of Their Parasite Specificity Reaction (A) above is freely reversible, being shifted to the left in alkaline solution and to the right in acid solution. 'O It is therefore conceivable that the variation in toxicity or activity of a series of arsenoso compounds might be a function of the hydrolysis constant of the formed thioarsenite. It is also possible that the rate of the above reaction is the determining factor. In either case, the usual relationships between chemical structure and reactivity should apply. In the particular case of equation (A), it should be possible to predict both the relative rate and the hydrolysis constant by means of Hammett's equation: log k- log k' = -V, where k is the rate or hydrolysis constant for the substituted aryl groups, k* the rate constant for the unsubstituted group,/* is a constant for the particular chemical reaction studied and which is independent of the groups, and o is a constant for each substituent group and is independent of the reaction. '^ Under these circumstances, if either toxicity or parasiticidal activity were quantitatively dependent solely on the above reaction, that activity should be a function of the <"*values. Actually, no correlation whatever was found between the Hammettlvalue's, and either the toxicity or the parasiticidal activity of the arsenoso compounds as previously described in Section II B. Again on the basis of the Hammett equation, if the parasiticidal action of arsenicals and their toxicity to the host were quantitatively dependent upon their reactivity with thiol groups in the parasite and in the host, then the therapeutic index of a series of arsenoso compounds in a given infection should be reasonably uniform. The relative reactivity of an arsenoso compound with the two groups of thiols, one group in the parasites and one in the host cell, should be independent of the particular substituent on the benzene ring. Instead, that ratio varied as much as 60-fold in the arsenoso series alone. Even when the same compound was tested against the same organism (T_. rhodesiense) but in different animal hosts, Tatum, Pfeiffer and Kuhs'' obtained therapeutic indices of 11, 5 and 0 in rats, rabbits and dogs respectively. It seems clear that possible variations in the rate or equilibrium constants of the reaction between arsenoso compounds and thiols to form thioarsenites are themselves inadequate to explain the known relations in the parasiticidal action or toxicity of these compounds. Barron and Singer3 studied the relative inhibitory action of various arsenicals against a number of enzymes in vitro, and their results are pertinent in this connection, although the quantitative aspects of their data are partially obscured by the varying amounts of impurities in the enzyme preparations capable of combining with and inactivating the arsenicals. The susceptibility of the different enzymes to the same arsenical was generally of the same order of magnitude. 3 Using eighteen different enzymes systems, they' found that under the particular experimental conditions, most of the enzymes were almost completely inhibited (by 80 to 100 per cent), but that pyruvate dismutation was inhibited by only 35 per cent and pyruvate condensation by 56 per cent. The degree of inhibition obtained with five different arsenoso compounds acting on the same enzyme, succinoxidase, were also not significantly different. 3D However, from the same laboratory it was later shown that d-amino acid oxidase and yeast carboxylase were inhibited by p-aminophenyldichlorarsine but not by lewisite. Also, trans-

54 o „ ^ o «o V. ° t I o S £ * < 2 0 < LJ fc ,, o OdllA Nl S31AOOdHlAd3 A9 ONHOa OIN3SWV JO SINOOWV 1 .S1 o h X f. 8i3 ri o >.i. * .c ° " H § « v, • O i; H H

aminase was almost completely inhibited by p-arsenosobenzoic acid, but only 33 per cent inhibited by the same concentration of lewisite. The authors suggest that these differences may arise from a different spatial arrangement of the thiol group on the protein molecule 2. The Possibly Varying Permeability of the Cell In possible explanation of the varying biological effects of different arsenoso compounds, one may postulate as an alternative hypothesis that although their reactivity with a given -SH compound is reasonably uniform, the arsenicals vary widely in their ability to penetrate the cell wall, either in the microorganism or the host, and that it is these differences which largely determine the varying parasiticidal action and toxicity of these compounds. This thesis is strongly supported by the data on the binding of arsenicals by cell suspensions. It was first shown by Thuret that a suspension of red blcod cells in vitro_ bound the highly active arsenoso compounds, but not the relatively inactive arsonic acids. ° This finding was extended by Hogan and Eagle, who demonstrated that in a large series of arsenoso compounds the amount of arsenic bound by red blood cells ill vitro under standard conditions was roughly proportional to the systemic toxicity of the arsenical, 3^ strongly suggesting a causal relationship (Fig. 10). The corollary of this finding was also demonstrated: with compounds of widely varying toxicity, at dosages which produced equivalent toxic effects in vivo, essentially similar amounts of arsenic had been bound by the tissues. (The binding of arsenicals by erythrocytes was also studied by Fink and Wright. 29 Using a single arsenical, mapharsen, these workers found an approximate linear relationship between the logarithm of the amount in the cells and the logarithm of the amount in the plasma, and suggested that the binding of arsenicals was largely physical in nature. They also found that from 60 to 90 per cent of the bound arsenic could be released from the red blood cells by simple resuspension in fresh plasma. These important findings are difficult to reconcile with the thesis that arsenoso compounds react with -SH groups in the cells as previously discussed, and further study seems indicated.) Eagle and Magnuson^ subsequently showed that, just as the degree to which arsenicals were bound by red blood cells was related to the systemic toxicity of the compounds, so the binding of arsenicals by trypanosomes was a measure of their parasiticidal activity. Their results with a large series of the arsenoso compounds and a normal strain of Trypanosoma equiperdum, are given in Table XII. The correlation between the amount of the arsenical bound under standard conditions, and its trypanocidal activity, was again so regular as to suggest a causal relationship. This quantitative study confirmed and extended the previous qualitative findings of Hawking35, Yorke, Murgatroyd and Hawking63 and Reiner, Leonard and Chao47b, that actively trypanocidal compounds were bound by the organisms, while inactive compounds were not. The wide differences in the amounts of arsenical bound by a trypanosomal or erythrocyte suspension could perhaps be due to the corresponding differences in the reactivity of the several arsenicals with a given cellular enzyme system. Differences of this order of magnitude have, however, not been demonstrated in the reaction of arsenoso compounds with thiols in vitro: and it seen.s improbable that the hydrolysis constant of a series of thioarsenites involving the same thiol and different arsenoso compounds would differ by a factor of e.g. 10Z-103(cf. page 27 et seq.). The results are, however, consistent with the thesis that different arsenicals vary in the ease with which they can pass through the cell wall, and that it is this selective permeability which determines the varying activity of a series of closely related compounds against the same cell. The results obtained with "arsenic-resistant" strains of trypanosomes offer additional strong evidence that the varying amounts of arsenical bound by the organisms reflect the varying permeability of the cell wall to the several compounds, rather than their varying reactivity with a cell thiol or thiols. Hawking35 had shown that arsenosobenzene was as actively trypanocidal against a "resistant" strain of T_. equiperdum as against a normal strain; but that arsenicals containing certain substituents, notably the -NHCH2CONH2 group were not effective against the resistant organism, and were not bound by it. Similar results were obtained by Yorke, Murgatroyd and Hawking, *>3 using an a.-senic-resistant strain of T_. rhodesiense. King classified arsenoso compounds into three separate groups on the basis of their action on "resistant" strains of trypanosomes. ' (1) arsenosobenzene, arsenosoxylene, and similar compounds which were active against both normal and "arsenic-resistant" strains, and which fall into the group of compounds with "inert" substituents as defined on page 16; (2) carboxyl substituted compounds and (3) amide substituted arsenicals. Compounds in either of the last two groups were relatively

36 inactive against one type of "arsenic-resistant" trypanosome, but normally active against the other. King concluded that while the actual intracellular process determining the death of the organism was the same in each instance, the several classes of compound entered the trypano- some by a different mechanism. Eagle and Magnuson,22 working with a strain of Trypanosoma equiperdum which had spontaneously become arsenic-"resistant", found that arsenosobenzene and many of its simple derivatives (chloro, methyl, carboxy, etc. ) was bound just as strongly by, and were just as active against, this resistant variant as the parent normal strain. On the other hand, arsenoso compounds containing amide and amino substituents were less active against the resistant strains by a factor of 1/Sth to l/200th with a corresponding reduction in the amount of arsenic bound by these organisms (Table XIII). TABLE XIII THE CORRELATION BETWEEN THE TRYPANOCIDAL ACTIVITY AND COMBINING AFFINITY OF ARSENOSOBENZENES FOR NORMAL AND "ARSENIC-RESISTANT" TRYPANOSOMES 20* 22 Compound Tested Relative Trypanocidal Activity vs . Percentage of Arsenic Bound* by (R<^)AsO) Normal Trypanosomes "Arsenic -Resistant" Trypanosomes Normal Trypanosomes 11 Arsenic-resistant" Trypanosomes p-S02NH2 15-21 0. 1 1 45-50 0 p-CONHCH2CONH2 9 0. 3 37-39 1. 8 p-(CH2)3COOH 35-39 38 49-55 59 p-CI 75 81 53-59 70 Unsubstituted phenyl arsenoxide 100 100 71-79 80 * 200-300 x 10 trypanosomes per cc. ; 1. 66/"g. As/cc. The selective susceptibility of "arsenic-resistant" organisms to certain types of arsenicals, associated with their ability to bind those arsenicals, is thus well documented. If this selective action is not determined by a selective permeability, one must then assume (1) that each class of arsenical can combine only with certain thiols in the cell, and (2) that the normal and resistant trypanosomes differ qualitatively in the type of cellular thiol which is susceptible to arsenicals, and the inactivation of which determines the death of the cell. 35 The former assumption in particular seems improbable. The importance of cell permeability is further indicated by the effect of pH on the parasiticidal activity of acid substituted arsenoso compounds (Eagle'7). With aryl arsenoso compounds not containing acid substituents the trypanocidal activity proved to be independent of the pH of the solution. With acid substituted compounds, however, the activity, both against T. equiperdum and Treponema pallidum, decreased strikingly as the pH of the testing solution increased. As discussed in a preceding section, this was related to the fact that the ion was usually almost wholly inactive, while the undissociated molecule was highly trypanocidal. The effect of pH on the binding of acid-substituted arsenoso compounds paralleled its effect on their trypanocidal activity. One must conclude that the ion was usually inactive because it was bound by the organism to only a limited degree, while the highly active undissociated acid was highly active because it was strongly bound.

37 The striking differences shown in Tables XII and XIII and in Fig. 8 can hardly be related to the varying reactivity of the salt and the free acid with thiols: the salts of acid substituted arsenoso compounds reacted as readily with cysteine and other thiols as did arsenoso compounds with non-ionic substituents. The results are, however, consistent with the assumption that the ionized form passes through the cell wall of the trypanosome into the interior of the cell only to a limited degree as compared with the undissociated molecule. The many analogies in other cell types and compounds have been summarized by Eagle^, by Albert ' and by Simon and Blackman41'. (The possibility that the arsenicals are bound on the outside wall of the organism, rather than within the cell, has been discussed by Reiner, Leonard and Chao*7b and by Eagle'?. The latter concluded the quantitative relationships between the amount of arsenical bound and the available surface area make it unlikely that the major portion of the arsenic bound by trypanosomes attaches to the cell membrane. ) On the permeability theory, the cell wall would be impermeable to certain arsenoso compounds, while others can penetrate with ease, and having penetrated are so firmly bound to some of the cellular constituents (thiols) that they no longer figure in the intracellular-extra- cellular equilibrium. In consequence, the concentration builds up inside the cell, and may attain 50, 100 and even 200 times that outside the cell. The factors determining which compounds can penetrate the cell wall and which cannot are not yet clear; and similarly unresolved are the nature of the changes in the cell which make it impermeable to, and thus, resistant to compounds which are normally highly active. The phenomenon of drug antagonism, which has become so important of recent years in relation to the mode of action of bacteriostatic and bactericidal agents (the sulfonamides, anti- biotics) and of growth factors, is of significance in this connection. Williamson and Lourie°0 showed that while the trypanocidal action of most arsenical drugs were not antagonized by £-aminobenzou acid (P. A. B. A. ), y-(p-arsenosophenyl)-butyric acid was so antagonized. Similarly, melarsen oxide, a drug which is highly active against arsenic-resistant strains of trypanosomes, is antagonized by "surfen C",°O a non-arsenic-containing compound somewhat similar to melarsen oxide in structure. Schleyer and Schnitzer have shown that many esters and amides of isocyclic and heterocyclic acids antagonize the action of oxophenarsine against Trypanosoma equiperdum, while the free acids do not. 48 (These drug antagonists did not, however, affect the toxicity of the arsenicals. ) Unfortunately, there are no data as to whether these antagonists prevent the concentration of the arsenical within the cell. In such case, the antagonists might either prevent the passage of the arsenical across the cell wall, or by blocking the union of the arsenical with an intracellular compound, prevent its rapid accumulation and parasiticidal action. Work and Work, commenting on the striking antagonism of P. A. B. A. for y-(£-arsenosophenyl)-butyric acid concluded that "the only explanation for this action is to assume that the drug acts ultimately in the same manner as other trypanocidal arsenicals, but that P. A. B. A. may be preventing or limiting its admission into the trypanosome cell. "°1 SUMMARY 1. Of the three types of arsenicals used in the treatment of spirochetal and protozoal diseases, only the arsenoso compounds have a direct parasiticidal activity, the arsonic acids and arseno compounds being active by virtue of their conversion to arsenoso compounds in the animal body. 2. Araenosobenzene was one of the most toxic and most actively parasiticidal compounds in the entire aeries tested, with no demonstrable selective action among the organisms studied. (a) Substitution with a single -CH3, -NO2, -Cl, -NH2, -OH or -F groups either did not significantly affect the toxicity or parasiticidal activity of the parent arsenosobenzene, or reduced them both to the same slight degree.

(b) Acid-substituents strikingly decreased both the direct parasiticidal and toxic action of arsenosobenzene The ionized form was generally inactive, undissociated acid highly active, and the effect of pH on parasiticidal activity could be related to its effect on dissociation. There were, however, important exceptions to the general inactivity of the ionized compounds (e.g. , the 3-NO2-4-COOH and £-(CH2)3COOH arsenosobenzene). (c) Amide substituents caused a slight decrease in treponemicidal and trypanocidal activity, but a striking decrease in toxicity. Substitution in the amide hydrogens usually diminished its favorable effect on toxicity, exceptions being noted in groups which in themselves also reduced the toxicity of arsenosobenzene (e. g. -C2H4OH, -C6H4NHCOCH3). (d) Unlike amidification, esterification of acid substituents resulted in compounds approaching arsenosobenzene in their high toxicity and general parasiticidal action. Many of these compounds readily hydrolyzed in aqueous solution, their biological activity then corresponding to that of the free acid. (e) The effect of complex substituents was usually determined by the nature of the terminal group in the substituent (e.g. (CH2bCONH2; CON(CH3)2). (f) Two substituent groups had an effect which could not be predicted from those of the substituents taken singly. Unlike the case of the single substituents, in these di-substituted compounds the position on the benzene ring profoundly modified the activity of the compound. (g) Both the toxicity and direct tryponemicidal activity of thioarsenites were somewhat less than those of the corresponding arsenosobenzene; and in the treatment of rabbit syphilis, the therapeutic index was also no better than that of the parent compound. 3. There is considerable evidence to support the view that the toxicity and therapeutic activity of arsenoso compounds are determined by the amounts which enter into combination with cellular components. The widely varying activity of a single arsenoso compound against different organisms, and of a series of such compounds against the same organism, would be related to the amount of arsenical bound. 4. Data have been summarized which indicate that arsenicals combine with -SH groups in protoplasm. Many essential enzymes contain sulfhydryl groups; and most such enzymes can be inactivated by arsenoso compounds and reactivated by sulfhydryl or other agents which can compete with the enzyme protein for the arsenical. It is a reasonable working hypothesis that arsenoso compounds combine reversibly with sulfhydryl groups in essential enzymes, and that their toxic effect on cells is due to that combination. 5. The selective action of particular compounds against certain cells could be explained on the basis of either (a) the varying affinity of different enzymes for the same arsenical, or of different arsenicals for the same enzvme; or (b) differences in the permeability of a given cell to different arsenicals. While the evidence is far from conclusive, the latter assumption appears best to explain the experimental data.

39 BIBLIOGRAPHY 1. Albert, A., Nature, 165. 12(1950). 2. Barber, H. J. . J. Chem. Soc. , 1020(1929). 3 (a) Barron. E.S. G. and Singer, T. P. , Science, 9_7, 356 (1943); (b) J. Biol. Chem., 157, 221 (1945); (c) Barron. E.S.G., Miller, Z. B. . Bartlett, G.R., Meyer, J. and Singer, T. P. Biochem. J. , 4J, 69 (1947). 4. Bersin, T.S., Z. phvsiol. Chem., 2£i, 177(1933). 5. Bersin, T. S. and Logermann, W. , Z. physiol. Chem., 220, 209(1933). 6. Binz, A., Rath, C. and Wilke, G. , Biochem. Z. , 223. 176(1930). 7. (a) Blicke, F. F. . Oneto. J. F. and Webster, G. L. , J. Am. Chem. Soc., 59.. 925(1937); (b) Rosenthal, S. M. and Bauer, H. . Pub. Health Rep. , 54, 1317(1939). 8. (a) Castelli, G. , Arch. Schiffs-Tropen-Hyg. , .16, 605(1912); (b) Schamber, J. F. , Raiziss. Geo. W. and Kolmer. J.A. , J. Am. Med. Assoc. , 78. 402 (1922); (c) Nichols. H. J. , ibid. , 76, 1335 (1921). 9. Cohen, A., King, H. and Strangeways, W. I. , J. Chem. Soc., 2866 (1932). 10. Cohen A.. King. H. and Strangeways, W. I. . J. Chem. Soc., 3043(1931). H. (a) Doak. G. O. , Eagle, H. and Steinman, H. G. , J. Am. Chem. Soc., 62, 168(1940); (b) ibid , 62. 3010 (1940); (c) ibid. , 66_. 194 (1944). 12. Doak, G.O. , Steinman, H. G. and Eagle, H. , J. Am. Chem. Soc., 63, 99(1941). 13. (a) Eagle, H. , J. Pharmacol. and Exp. The rap. , 64, 164 (1938); ibid. , £6, 423(1939). M. Eagle, H. , J. Pharmacol. and Exp. Therap. , 66, 436(1939). 15. Eagle, H., J. Pharmacol. and Exp. Therap., 69, 342(1940). 16. (a) Eagle, H., Science, 101, 69 (1945); (b) Pub. Health Rep. , 6J, 1019(1946). 17. Eagle, H. , J. Pharmacol. and Exp. Therap., 85, 265(1945). (a) Eagle, H. , unpublished observations, (b) Eagle, H. , Blood Levels, Renal Clearance and Chemotherapeutic Activity, with Particular Reference to Arsenicals and Penicillin. Chapter in "Evaluation of Chemotherapeutic Agents", New York Academy of Medicine, Section on Microbiology, Symposium No. 2, Columbia University Press, New York City, 25-43, (1949). 18. (a) Eagle, H. , Doak, G. O. , Hogan, R. B. and Steinman, H. , J. Pharmacol. and Exp. Therap., 70, 211, 221 (1940); (b) ibid. , 74, 210(1942). 19. Eagle, H. , Germuth, F.G. , Magnuson, H.J. and Fleischman, R. , J. Pharmacol. and Exp. Therap., 89, 196(1947). 20. Eagle, H. , Hogan. R. H. , Doak. G.O. and Steinman, H. G. , Pub. Health Rep. , 59.. 765 (1944). 21. (a) Eagle, H. , Hogan, R. B. , Doak, G.O. and Steinman, H.G. , J. Am. Chem. Soc., 65. 1326 (1943); (b) J. Pharmacol. and Exp. Therap. , JU, "2 (1944).

40 22. Eagle, H. , and Magnuson, H. J. , J. Pharmacol. and Exp. Therap. . 82, 137(1944). 23. Eagle, H. , Magnuson, H. J. and Fleischman, R. , J. Clin. Invest., 25, 451 (1946); Amer. J. Syphilis, 30, 420 (1946). 24. Eckler, C. R. andShonle, H. A. , Am. J. Syphilis, Neurol. , 19., 495 (1935). 25. Ehrlich, P., Ber., 42, 17(1909). 26. Ehrlich, P. , Lancet, 2, 445 (1913). 27. Ehrlich, P. and Bertheim, A., Ber., 43, 917(1910). 28. (a) Ehrlich, P. and Hata, S. , The Experimental Chemotherapy of Spirilloses, Englished., Redman, N. Y. , 1911. (b) Ehrlich. P., Ber.. 42, 17(1909). 29. Fink, L. D. and Wright, H. N. , J. Pharmacol., 94, 445(1948). 30. (a) Friedberger, E. , Berlin, klin. Wochschr. , 45, 1714 (1906); (b) Friedheim, E. A. H. , Ann Inst. Past., i5, 108 (1940), J. Am. Chem. Soc. , 66, 1775 (1944). 31. Germuth, F.G. and Eagle, H. , J. Pharmacol. and Exp. Therap., 92, 397(1948). 32. Gilman, A., Philips, F. S. , Allen, R. P. and Koelle, E. S. , J. Pharmacol., 87, 85(1946). 33. Gough, G.A.C. and King, H. , J. Chem. Soc., 669(1930).. 34. Hammett, L. P. , Physical Organic Chemistry, McGraw-Hill Co. , New York, N. Y. , Chap. VII., 1940. 35. Hawking, F. , J. Pharmacol., 59, 123 (1937). 36. Hellerman, L. , Perkins, M. E. and Clark, W. M. , Proc. Nat. Acad. Sci. , U.S., 1.7, 855 (1933). 37. Hogan, R. B. and Eagle, H. , J. Pharmacol. and Exp. Therap., 80, 93(1944). 38. (a) Kharasch, M.S., U.S. Patent 1, 677, 392 (1 928); (b) U. S. Patent 1. 959, 958 (1934). 39. King, H. , Trans. Faraday Soc. , 39, 383(1943). 40. Kuhs, M. L. and Tatum, A. L. , J. Fharmacol. , 6.1, 451 (1937). 41. Longcope, W. T. , Leutscher, J. A. , Wintrobe, M. M. and Jager, V., J. Clin. Invest., 25, 528 (1946). 42. Longcope, W.T., and Luetscher, J.A.,'J. Clin. Invest., 25, 557(1946). 43. Maschmann, E. , and Helmert, E. , Z. physiol. Chem., 219, 99(1933). 44. (a) Otto, G. F. and Maren, T. H. , Science, 106, 1 05 (1947); (b) Thetford, N. D. , Otto, G. F. , Brown, H. L and Maren, T. H. , Am. J. Trop. Med. , 28, 577(1948). 45. (a) Peters, R. A. , Sinclair, H. M. and Thompson, R. H. S. , Biochem. J. , 40, 516(1946); (b) Stocken, L. A. and Peters, R. A. , ibid., 40, 529 (1946); (c) Peters, R. A. and Stocken, L. A. , ibid., 41, 53 (1947); (d) Spray, G. H. , Stocken, L. A. and Thompson, R.H.S., ibid., 41,' 362 (1947). 46. (a) Probey, T. F. and McCoy, G. W. , Pub. Health Rep. , 45, 1 716 (1930); (b) Wright, H. N. , Biedermann, S. , Hanssen, E. and Cooper, C. I. , J. Pharmacol., 73, 12(1941).

41 47. (a) Pagan, C. and Boots, R. H. , J. Am. Med. Assoc. , 133. 752 (1947); (b) Reiner, L. , Leonard, C.S. and Chao, S. S. , J. Pharmacol. and Exper. Therap. , 43, 186 (1942). 48. Schleyer, W.I. and Schnitzer, R.J., J. Immunol. , 60, 265(1948). 49. Simon, E. W. and Blackman, G. E. , Toxicity and Antibiotics, Symposia for the Society of Experimental Biology, Academic Press, New York, N. Y. , p. 253 (1949). 50. Slater, E.C., Biochem. J. , 45, 130(1949). 51. Stocken, L. A. and Thompson. R. H. S. , Physiol. Rev., 2^, 168(1949). 52. Stocken, L. A. and Thompson, R. H. S. , Biochem. J. , 40, 535(1946). 53. Stocken, L. A. and Thompson, R. H. S. , Biochem. J. , 40, 548(1946). 53a. Strangeways, W.I. , Ann. Trop. Med. , ^1, 387 (1937). 54. Tatum. A. L. and Cooper, G. A. , J. Pharmacol. , 5_0, 198 (1934). 55. Tatum, A. L. , Pfeiffer, C. C. and Kuhs, M. L. , J. Pharmacol., 5_9, 241 (1937). 56. Thuret, M. J. , J. pharm. chim. , 28, 22(1938). 57. (a) Voegtlin, C. , Dyer, H. H. and Leonard, C. S. , Pub. Health Rep. , 38, 1882 (1923); (b) Rosenthal, S. M. and Voegtlin, C. , J. Pharmacol. , 39, 347 (1930); (c) Rosenthal, S. M. Pub. Health Rep. , 47, 241 (1932); (d) Voegtlin, C. , Dyer, H. H. and Leonard, C. S. , J. Pharmacol. , 25, 297 (1925). 58. Voegtlin, C. and Smith, H. W. E. , J. Pharmacol., 1_5, 475(1920). 59. Whittaker, V. P. , Biochem. J. , 4J , 56(1947). 60. Williamson, J. and Lourie, E. M. , Nature, 161, 103(1948). 61. Work, T.S. and Work, E. , The Basis of Chemotherapy, Interscience Publishers, Inc. , New York, R Y. , p. 229 (1948). 62. (a) Yorke, W. and Murgatroyd, F. , Ann. Trop. Med. , 24, 449 (1930); (b) Schamberg, J. F. , Kolmer, J. A. and Raiziss, Geo. W. , Am. J. Syphilis, J., 1 (1917). 63. Yorke, W. , Murgatroyd, F. and Hawking, F. , Ann. Trop. Med. , 2^, 351 (1931).

DISCUSSION DR. CARROLL KING (Northwestern University, Evanston, Illinois); I would like to ask Dr. Doak just how they plotted this information, in making an attempt to correlate this data with the sigma function. DR. DOAK: Actually, no plot was possible between the Hammett-sigma values. One has, in the Hammett table, the £-amino compounds at the top, and a hydrogen in the middle, and a nitro at the bottom. We found that the activity or the toxicity of the p_-amino compounds and p-nitro compounds were essentially the same. In attempting to plot either the log of the activity or the log of the toxicity, which must be a function of this sulfhydryl reaction, we made several attempts to isolate the actual reaction and study either the rate or the hydrolysis constants, and those studies are still going on. We have not been able, so far, actually to isolate the reaction and study either the rate or the hydrolysis. DR. CARROLL KING: I would like to call attention to one factor in connection with the application of this sort of data. If you consider the Hammett equation, it states that the log of the ratio of two constants, is equal to sigma rho. Both sigma and rho are constants. The validity of the equation has been established by application to fifty -two different reactions. I would just like to point out one thing. The sigma values have been established by many reactions. The rho value, is simply the slope of a line, and about all you can tell from the Hammett equation is whether the points fall on a line. So, if you plot log K against the sigma values, you may get a line whose slope is one in which case you have direct correlation of the type which, perhaps, you have searched for here. But you may get a line whose slope is minus. in which case, you correlate the other way, and, slopes may be much steeper, either negative or positive. Specific examples of reactions in which the slope is large are known, and in which the slope is small are known, with the same sigma values. It may be that you simply have a case where the slope is very small. DR. MAXWELL SCHUBERT (New York University College of Medicine. New York, New York): This is an enormously interesting paper which we have just listened to, and there are a great many things which would be worth discussing. There is just one point I would like to discuss at the present time, though, and that is the remark which Dr. Doak made that the BAL compounds with arsenic are not necessarily more stable than the open-chain ones. I am going to base this discussion on a paper which recently came out by Fennell and Carmack, in which they studied ultraviolet absorption spectra of organic sulfides and mercaptols The type of compound they had to deal with, if we take the mercaptal, was this type of / X S-R compound. It has generally been assumed, and this is a point which Dr. Doak was very careful to make, that in the absence of resonance, a compound of this structure would not necessarily be any less stable than a compound in which these R's were independently linked together. In the particular case of these mercaptals, as a result of the study of the ultraviolet absorption, a conclusion was arrived at that there is a resonance which may be described as follows : © 0 H .S-R

43 One electron pair, which binds the carbon to the hydrogen, is pulled in by the carbon and shared with the sulfur. This particular sulfur atom becomes negative, and has, in effect, ten electrons around it. But the hydrogen becomes positive and is ionized under those conditions. If we follow this thought and substitute our metal, whether it is arsenic or antimony does not make any difference at the moment, we would have a compound of this character. If the same kind of resonance is possible here, there are two ways in which such electrons sharing could be imagined; one is by pulling in an electron from the sulfur; another is by pulling an electron pair out of the metal and sharing it with the sulfur. The set of possibilities could constitute a resonance system. In the second case, you would get an ionization effect, in which © 0 -S-R the sulfur would become negatively charged. In that particular case, if you have an open-chain compound, this would be ionizable. In the presence of competing systems, which might want to combine with the metal, this would be more readily displaced. However, in the case where the R's were combined independently, if one did become somewhat ionizable as a result of such resonance, it could not move away and, it would block the introduction of other elements there and give an appearance of greater stability. DR. JOSEPH H. BURCHENAL (Sloan-Kettering Institute, New York, New York): This problem of resistant strains interests me tremendously, and I gather that the resistant proper- ties of the trypanosomes are correlated with their ability to bind arsenic. I would like to ask two questions of Dr. Doak. In the first place, were these resistant strains naturally occurring or induced by suboptimal therapy or treatment in vitro? Second, have you any suggestions as to the reason behind the ability of the resistant trypanosomes to bind one arsenical and not another type. DR. EAGLE: If I may answer Dr. Burchenal's question this was a spontaneously occurring resistant variant, which developed in the absence of exposure to an arsenical. I wish we had some explanation of the fact that certain arsenicals are bound by these resistant strains just as actively as they are by normal trypanosomes, and are correspondingly active, while others are not bound at all, and are correspondingly inactive. Therein lies the crux to this problem, it seems to me: why certain substituents affect the permeability to the cells to a rsenosobenzenes. DR. D. W. WOOLLEY (Rockefeller Institute for Medical Research, New York, New York): In speaking of the effects of PAB on toxicity of arsenicals, there is one point which I think is worth bearing in mind, and that is an experiment which Sandground did some eight or ten years ago, in which it was shown that you could prevent, to a considerable degree, the toxicity of atoxyl and arsphenamine and neoarsphenamine, and various arsenical compounds, in animals by giving p -aminobenzoic acid. The point I want to bring up is that, once the arsenic compound had been given, no amount of PAB would protect the animal. If PAB was given before the arsenical, then a good protection could be had. This might bear on the point brought up about penetration of the cell.