J. M. Johnson and F. Bergel

CHAPTER 2 3

Biological Alkylating Agents

J. M. Johnson and F. Bergel

I. Introduction 1 6 1

II. Newer Developments 1^4

A. Nitrogen Mustards ^{1 6 4}

B. Dimethanesulfonates 174

C. Ethylenimines 1 7 5

D. Epoxides and Others 1 7 8

III. Mechanism of Action 1 7 9

IV. Summary and Outlook · 1 8 6

References 1 8 7

I. INTRODUCTION

Biological alkylating agents belong to a group of compounds which, chemically speaking, attack nucleophilic (electron-rich) centers. The details of these mechanistic aspects have been fully discussed by Ross

(1, 2). In contrast, the mechanistic aspects of their biological action, a summary of which will be given later in this review, have been only partly elucidated during recent years. The fact will emerge that these compounds are only in an indirect way "metabolic antagonists," for on the whole they do not participate in biochemical processes of the cell in the same manner as true antimetabolites, i.e, as "fraudulent" substrates, products, or coenzymes, but seem to achieve their effects by interacting chemically with certain constituents of living cells and altering them in such a way that the change in the structure and functional properties of these constituents leads finally to cell death. However, it should be men- tioned here that some alkylating agents appear to influence amino acid incorporation and thus may, under certain conditions, act as metabolic inhibitors (3).

A great number of reviews and papers read during symposia have been 161

CHART I—ALKYLATING AGENTS A. Sulfur-mustards SXS / \ CH2 CH2 I I CH2 CH2 (122) Hal Hal Nitrogen-mustards / Aromatic * Μ = —Ν \ 1. Purely aromatic

^CH^CH^Hal NCH2-CH2-Hal CH2— CH2—CI CH2—CH2—CI e.g., CB 1048 Γ T| active (140) 2. Aromatic aliphatic / \ u a? υ η = 1, 2, 3, 4 Deer. act.— 3. With latent activity (a) Azobenzenes X = C02H; Y = Me; CB 1414, highly active (59) (b) Acyl groups Μ

(c) Aminoacyl groups (cx) RCHCONH—ft y—M NH2 R = H, PhCH2, inactive (150) (c2) CH—NH tf/Nk==/ CO CO I e.g., CB 1655; η = 0, R = Η, Μ is tucta; active (40) 4. Nitrogen mustards Aromatic + substituent (a) Fatty acid or O-fatty acid (a,) Μ—(CH2)nCQ2H e.g., Chlorambucil, η ~ 3, highly active (145) (a2) Μ—^—0(CH2)/zC02H η - 2, highly active (144) (b) Amino acids (b,) M—fl \-(CH2)„CHC02H ^=/^ NH2 ii = 1 L-CB 3025, highly active; D-CB 3026, much less active; DL-CB 3007 = Sarcolysine, active (142,143) η = 0 less active; n=2 or 3, highly active; CB 1385, 1494 (144) CH2CHC02H NH2

(c) Alkylamines (CH2)„NR'R" R' = R" = H. highly active: R' = Me. R" = H, highly active: R' = R" = Me, active (23), +NR3, η = 2, inactive (d) Peptides C02H NH inactive I COPh j ν C02H (d2) Μ—<^ \— CH^H-CO-NHCHjj NH inactive (d3) Μ­

ΟΒ 3084

I COPh COzH I 2 CH2CHCO-NHCHR (14 tf) COOEt (L) ,CH-NH'COCHR (47) (L) NH2 e.g., CB 3262; R = CH2CH(CH3)2, active (ds) Μ

COOEt CH2CH'NH-COCHR (147) (L) NHCOCHJ e.g., CB 3224; R = —CH2—^ ^) active R = C13C, R = Me or Ph (149) CB 1431, active (5S) CB 3051, inactive (146)

C. Nitrogen mustards; Aliphatic 1. MeM, HN2, active 2. C1CH2CH2M, HN3, active 3. MeN(CH2CH2Cl)2, active Ο 4. RCH M, R = H, Me, active COR'R' = OH or N< (53)

162 J. M. JOHNSON AND F. BERGEL

CHART I (Continued) 5. M(CHjn CHCO,H NH, (56, 57a) NCH2CH2C1 6. HC^ ^CH2 CH, (CH.),,, CI 7. Carbohydrates NH-CHj-CH^-Cl CH, I (CHOH)« CH, NH-CHa—CH2—CI BCM, active (124) D. Nitrogen mustards; Amides RCOM 1. R = Organic acid res. (125) 2. R = EtO—, active (126) 3. R = "N

σ

CB 3159, active (128) (151)

3. Antimalarial structures Μ

(b) Quinones NHCH(CH2),M OMe CB 3117, inactive (66) Quinacrine Μ (153) F. Other alkylating agents 1. Tosyl derivatives ^CHj-CHj-Tos / Ph-N CH2-CHa-Tos 2. Methanesulfonoxy X MeSOjO^ \>S02Me (a) X = (CHa)„ η = 4 = Myleran (70) (b) X = -CH1CH=CHCH2- (c) X=-CH(CH,)^H- Me Me (d) η = 2 » Dimethyl Myleran X = —CHa(CHOH)«CH2— CB2511, active (130) 3. Ethylenimines (a) Phosphoramides

CH.-CH,

Ν I X

O(S) fcN—P—N<3 Ν TEPA, active; ThioTEPA, etc. (131)

•χχ "

V

\ I

23. BIOLOGICAL ALKYLATING AGENTS 163

164 J. Μ. JOHNSON AND F. BERGEL published, and it would only be a waste of precious space to attempt to summarize these here. One could perhaps start with a reference to some of this published material and then concentrate in the main on newer developments so far not available in a predigested form. A symposium (4) on biological alkylating agents was presented at the New York Academy of Science in 1957. A table presented there by Bergel (5), showing the various groups and subgroups of alkylating agents, has been brought up to date (Chart I ) . In addition to this report on the 1957 symposium, there are reviews by Haddow (6), Davis ( 7 ) , Montgomery (8), Timmis (9), and R. B. Ross (10), which adequately cover the older work. More recently, and principally from a clinical point of view, the use and func­

tion of these agents was discussed during a conference in Washington (11). Of course, reviews (12) are being prepared in increasing numbers, and some will become available not very much later than this article.

Even textbooks on medicinal chemistry (18) allow space for this group of chemotherapeutic compounds.

During the last few years, the search for improved alkylating agents has been carried out with reference to the following cytotoxic groups:

(A) nitrogen mustards, (B) dimethanesulfonates, (C) ethylenimines, and (D) epoxides and others.

A. Nitrogen Mustards

Due to the fact that simple dichloroethylamine derivatives such as H N 2 [ ( I ) , R = C H³] , nor-HN2 [ ( I ) , R = H ] , do not possess any pronounced selectivity of action vis-a-vis normal and abnormal dividing cells, work in this field has revolved around the modification of the

"carrier" part of the drug molecules, consisting in the first place of aro­

matic residues (II). These aromatic rings can be substituted by side chains R which may be (a) acidic, (b) basic, or (c) of an amino acid-

II. NEWER DEVELOPMENTS

N — R

"Warhead" "Carrier

(I) (Π)

23. BIOLOGICAL A L K Y L A T I N G A G E N T S 165 peptide character. In addition, a brief reference will be made to so-called

"latent compounds" (58 and 59).

1. W H E R E T H E S U B S T I T U E N T R I S A N A C I D I C R E S I D U E

Work in this field led to the development of the butyric acid deriva­

tive called CB 1348, chlorambucil, or Leukeran [ ( I I I ) , η = 3, Μ = ( C 1 C H²C H²)²N — here and in all succeeding formulas], which was found to be active against the Walker carcinoma 256 (14) and produced a fall in circulating neutrophiles and more so of lymphocytes (15). The higher and lower homologues of chlorambucil [ ( I I I ) , η = 0, 1, 2, and 4) were

found to be less active against the same experimental tumor. It has found use in clinical medicine against lymphocytic leukemia and lymphosar­

coma.

The introduction of an oxygen atom between the aromatic ring and the fatty acid side chain, gave rise (16) to a series of compounds of the general formula (IV). The compound in which η = 2 was found to be twice as active as CB 1348, but also twice as toxic, whereas the homo­

logues in which η — 0, 1, 3, and 4 were again less active. It is interesting to note that both these compounds with pronounced antitumor activity are isosteric, which suggests that their activity is in some way connected with the length of the side chain.

Another series of closely related, but nonacidic compounds of general formula (V) has recently been studied (17) because of their relation to schistosomicidal drugs (18-20). Only the compound (V) (n = 7) was

slightly active when tested against the Walker carcinoma, whereas the compound (V) (n =z 6) was inactive at the same dose level. In addition to this, Baker et al. at the Stanford Research Institute have begun a study

(21) of a wide range of analogues of chlorambucil, which includes, for example, such compounds as (VI, VII, and V I I I ) , as well as the ortho and meta mustard isomers of chlorambucil. The cinnamic acid isomers

(VIII) have shown encouraging antitumor activity against the mouse tumor sarcoma 180, adenocarcinoma 755, and leukemia L 1210, the stand-

(III) (IV)

(V)

1 6 6 J . Μ . J O H N S O N A N D F . BERGEL

ard test tumors of the Cancer Chemotherapy National Service Center, Bethesda. Their effects, however, are more like phenylalanine mustard than chlorambucil (22).

C H2C H2C O O C H3 C H2C H2C O O H

C H2C H2C I C H2N H C H2C H2C I - H C I

(VI) (VII)

C H = C H C O O H

^ C H2C H2C 1

<

XC H2C H2C 1 ο-, m-9 and />-Isomers

(vin)

2 . W H E R E T H E S U B S T I T U E N T R I S A B A S I C R E S I D U E

Some compounds of the general formula ( I X ) have been prepared and tested against the Walker carcinoma 2 5 6 (23). The compounds ( I X ) ,

R . Me ^ M — / V-(CH²)^W Ν Μ — / V- i C H ^ N t - M e i l " o r B r '

R' M e *

(IX) (X) η = 2 , R = R^R = Η and η = 2 , R = Η , R⁷ = Me were found to be

active, while compound ( I X ) , in which η = 2 , R == R^R = M e was less active, and ( Χ ) , η = 1 and 2 , a quaternary derivative, was found to be ineffective.

3 . W H E R E T H E S U B S T I T U E N T R I S O F A N A M I N O A C I D - P E P T I D E C H A R A C T E R

From the recent literature it appears that more attention has been paid to this group of compounds than to the two groups already mentioned, the

23. BIOLOGICAL A L K Y L A T I N G A G E N T S 167 reason being that the amino acid moiety might become involved in the metabolic processes of the cancer cell and thus act as an efficient carrier of the mustard "warhead." One of the first representatives of this class which had an appreciable antitumor activity (24) was the DL-phenyl- alanine mustard called merphalan ( X I ) , which was also independently

synthesized by Russian workers (25) and called sarcolysine. D u e to the interesting properties of merphalan and sarcolysine, it was decided to prepare the L - and D-enantiomorphs, which have become known as melphalan (name derived from raustard-L-pftenylaZanine) and medphalan, respectively. Biological tests of these three optical forms on the Walker tumor and on the rat sarcoma 45 (26) have shown the highest activity to be produced by the L-isomer, with the D-isomer being the least active compound. These differences among the sterically related compounds have also been verified by other workers (27-30) using other biological sys­

tems. On the other hand, when Zamenhof et al. (31) used the three phenyl­

alanine mustards in vitro on the transforming D N A of Hemophilus influenzae, there was no difference among them in their inactivating power. These findings suggest that the differences in the biological activity of each isomer may be due to transport or permeability properties, in that the L-enantiomorph may more effectively penetrate the cells. The use of labeled L - and D-forms would test this hypothesis.

Another modification of the phenylalanine molecule carrying the mus­

tard residue involved the length of the chain between the aromatic ring and the α-amino acid end group (16), giving rise to compounds of general formula ( X I I ) . To cite one example, called aminochlorambucil [ ( X I I ) , η =z 2 ] , it was found to be less toxic and less active than merphalan, but

its more favorable effect on circulating blood elements warranted its clinical study. The resolution of aminochlorambucil by Smith and Luck

(82) and the subsequent biological testing of the isomers on the S 91 Cloudman mouse melanoma showed that one of the isomers was more

NH² (XI)

NH2

(XII)

168 J. Μ. JOHNSON AND F. BERGEL effective than the other, and by analogy with the phenylalanine series it was assumed that this had the L-configuration. However, the measure­

ment of the optical rotations of the isomers under acidic conditions [Lutz and Jirgensons' rule (33)] indicated that the more active compound, surprisingly, possessed the D-configuration. Further evidence by Bergel et al. (34) supported this interesting difference between the phenylalanine and phenylaminobutyric acid series.

Until 1959, all these compounds carried the mustard group in the para position to the other residues of the aromatic ring. However, a number of laboratories (35-88) have now synthesized the meta isomer of merphalan or sarcolysine [ ( X I I I ) , η = 1 ] , which showed an improved antitumor activity (39) and therapeutic index as compared with the compounds belonging to the para series. This also applies to the para and meta isomers of phenylglycine [ ( X I I ) , rt — O] prepared by Ross et al. (40) and Connors (41). The same workers (41, 42) have even taken this story

a stage further by preparing the ortho isomer of merphalan or sarcolysine ( X I V ) , which is more active than either the meta or para isomers but has a somewhat greater toxicity.

Another possible way of varying structural changes on the phenyl­

alanine molecule is to produce compounds in which the α-amino group is transferred to the /^-position. This was first suggested by Professor J.

Murray Luck of Stanford University, and taken up by Bergel et al. (48), (CH2)wCHCOOH

NH2

(ΧΠΙ) (XIV)

NH2

(XV)

who prepared the para-mustard derivative of /?-phenyl-/?-alanine (XV) which was found to be less effective than melphalan, but at the same time

23. BIOLOGICAL ALKYLATING AGENTS 169 less toxic. The difference in biological activity among the para, meta, and ortho isomers in the β-phenyl-a-alanine series prompted the synthesis (38, 44) of the meia-mustard in the β-alanine series ( X V I ) , which was found to be more biologically effective than the corresponding para isomer, while the toxicity was about the same. It seems therefore that in

both the a- and β-amino series the general trend is toward increased activity, but about the same or slightly diminished toxicity, as the mus­

tard group is moved from the para position of the aromatic ring to the meta position and thence to the ortho position. The future preparation and testing of the ortho isomer in the β-alanine series would, so one hopes, prove this specific activity-structure relationship. The last kind of modi­

fication of the prototype amino acid derivative which should be men­

tioned here refers to both simple acyl or peptide derivatives of the α-amino group and esters, amides, or peptides of the carboxyl group. The two iV-acyl compounds which first received attention were iV-formyl

[ ( X V I I ) , R = CHO] (45, 46) and iV-aeetyl derivatives of merphalan and sarcolysine [ ( X V I I ) , R = C O C H³] (45). The investigators in Russia claimed that antitumor activity was not abolished in these compounds, although a very much higher dose was required to produce the same effect as with merphalan or sarcolysine. The corresponding iV-formylmelphalan has been reported in the literature (47), but due to the method of prepa­

ration the product was apparently partly racemized. Turning to the carboxyl end of the molecule, one finds that the ethyl and isopropyl esters of merphalan or sarcolysine have been prepared by the Russian workers (46), and biological tests showed that there was very little difference between the esters and the original amino acid as far as antitumor ac­

tivity (rat sarcoma 45) was concerned. The corresponding ethyl ester of melphalan has now been prepared (47) and also shown to be similar in activity to melphalan itself, although when used in a regional perfusion of a cancerous limb it proved to be more vigorous in its side effects than the free amino acid (48). The same group (47) in England have prepared melphalan amide hydrochloride [ ( X V I I I ) , R = N H², R' = H ] as well

CHCH2COOH

I

NH2

(XVI)

170 J . Μ . J O H N S O N A N D F . BERGEL

as iV-formylmelphalan amide [ ( X V I I I ) , R = N H²; R' = CHO] and the ethyl amide [ ( X V I I I ) , R = N H E t , R' = C H O ] .

/ ^ V- C H j C H C O O H Μ — — C H j j C H C O R

NHR NHR'

(xvii) (xvra)

From their results one can pick out two points of interest, viz., (a) the general toxicity, hemotoxicity, and antitumor activity are greatest in compounds carrying a free amino group and (b) acylation of the free amino group reduces the biological activity although, in most cases, the ratio between the toxic and effective antitumor dose (the therapeutic index) remains roughly the same.

Focusing attention on some peptides containing merphalan (sarcoly­

sine) or melphalan, two separate investigations may be considered. First, there are the contributions by the Russian group with the general tend­

ency towards the use of sarcolysine (merphalan) and other D L - or racemic amino acids (46, 49, SO). All of their peptides of sarcolysine (merphalan) to date¹ are iV-formyl derivatives, for instance, iV-formylsarcolysylphenyl- alanine ethyl ester ( X I X ) . Other peptides recorded are those where the amino acid residue is valine, glycine, tryptophan, S-chloroethylcysteine, methionine, S-benzylhomocysteine, and leucine. It was claimed that all of these peptides have a favorable therapeutic index with low hemotoxicity.

It is interesting to note here that some peptides of chlorambucil of general formula ( X X ) have been prepared (50, 51), and biological tests showed that when R was represented by a natural amino acid the com­

pounds possessed strong carcinostatic activity, whereas when R was an unnatural amino acid activity was low or absent.

C O O C²H

M — / V- C HaC H C O N H C H C H r \ 7 Μ — /

V-

(XIX) (XX) Turning now to the peptide work carried out in England (47), with

the tendency towards the use of melphalan and other optically active amino acids, a series of dipeptides of melphalan was prepared of general

1 Note added in proof: The preparation of sarcolysine peptides (with valine) in which the amino group is free has recently been recorded. [I. L. Knunyants, Κ. I. Karpavichus, and Ο. V. Kil'disheva, Izv. Akad. Nauk. S.S.S.R. Otd. Khim.

Nauk. p. 1024 (1962).]

2 3 . B I O L O G I C A L A L K Y L A T I N G A G E N T S 171 formula ( X X I ) , where R was such that the amino acid residue was L-valine, L-leucine, L- and D-alanine, and L- and D-phenylalanine.

Μ—({ \ — C H²C H C O O E t

ι / / \

NHCOCHCH2 V x>

NHCOCH3

(XXIV)

Biological testing of these compounds has shown that in general they have no great advantage over melphalan, as their activity is approxi­

mately equivalent to that of the melphalan they contain. However, so far with the L-leucyl dipeptide it appears that the compound is effective against a wider range of experimental tumors than the parent substance.

Whether this proves to be the case under clinical conditions, particularly in view of the usefulness of melphalan in regional perfusion (52) of patients with malignant melanoma, remains to be seen. On the other hand, in parallel with the Russian claim that their compounds containing formyl- or acetylsarcolysine (merphalan) possessed a better therapeutic In addition to this, three tripeptides have been prepared, viz., L-valylgly- cylmelphalan ethyl ester dihydroehloride, L-alanyl-L-leueylmelphalan ethyl ester dihydroehloride, and the interesting L-valylmelphalanylgly- cine ethyl ester dihydroehloride ( X X I I ) , where the cytotoxic amino acid is in the center of the molecule. One tetrapeptide, viz., L-valylglycyl- glycylmelphalan ethyl ester dihydroehloride ( X X I I I ) has also been prepared.

Γ\_

- / ~ V M

Μ — ^ y— CH2CHCOOEt C H3 N^ > = /

NHCOCHR CHCHCONHCHCONHCHjjCOOCaHs

N H2 CHg N H2- 2 H C 1

( X X I ) ( Χ Χ Π )

COOC²H⁵

CHCHCONHCH²CONHCI^CONHCHCH²—(* ^NY- M

C H / N H2- 2 H C 1

( Χ Χ Π Ι )

1 7 2 J . Μ . J O H N S O N A N D F . BERGEL

index, the English workers (47) found in preliminary tests with repre­

sentative dipeptides of melphalan (similar to those above) in which the terminal amino group was blocked by an acetyl residue that, for instance, iV-acetylphenylalanylmelphalan ethyl ester ( X X I V ) had a substantially more favorable therapeutic index than melphalan itself. One may perhaps ask why all this tedious work on modifications of the original phenyl­

alanine mustard has been undertaken. The reason is obviously that all the groups engaged in this work tried to produce a compound possessing the very desirable antitumor properties of melphalan, but with a very much lower toxicity and hemotoxicity.

Before concluding this section on amino acid derivatives as carriers of the nitrogen mustard group, two further sets of compounds could be mentioned briefly: first, those where the mustard moiety replaces the α-amino group of the amino acid, as reported by Ishidate and his co­

workers (58, 54) and more recently by Nyhan and Busch (55), e.g., the mustard derivative of alanine ( X X V ) which was found to be active on the Yoshida sarcoma; and second, those in which the nitrogen mustard moiety replaces the ω-amino group in diamino acids ( X X V I ) , produced by Japanese (56) and Russian (57) teams.

C H 3 C H C O O H Μ (CHS) nCHCOOH

I I

Μ N H²

(XXV) (XXVI)

4 . C O M P O U N D S W I T H L A T E N T A C T I V I T Y

Another line of approach in the nitrogen mustard field concerns the concept of "latent activity," which means the preparation of compounds possessing low chemical and biological activities, but which would undergo degradation under physiological conditions to form a moiety with high alkylating power and consequently pronounced antitumor effects. With this end in view a series of practically inert azo mustards was syn­

thesized by Ross and Warwick (58, 59), who put forward the suggestion that these compounds might become activated by a process of in vivo re­

duction of the azo link by enzymes, and that certain substituents on the aromatic ring might facilitate this reduction. This was demonstrated in vitro with purified xanthine oxidase and most readily with compound C B 1 4 1 4 ( X X V I I ) , which also proved to be the most active member of this series against the Walker rat carcinoma 2 5 6 . The effective molecular species was considered to be ( X X V I I I ) , the end product of the complete reductive fission.

23. BIOLOGICAL ALKYLATING AGENTS 173

COOH

fxxvn) (χχνιπ)

A different group of latent compounds was synthesized and examined by Ross et al. (40, 42) and Connors (41), viz., mustard derivatives of aryl- and arylalkylhydantoins. The two compounds in this series which were most active are represented by formulas ( X X I X and X X X ) , the former being the precursor of the me ία-mustard of phenylglycine already mentioned [ ( X I I I ) , η = 0 ] .

C O — N H

(XXIX) (XXX) The next group with amino acids as carrier includes O-carbamoyl

derivatives of serine and threonine (60). Here the mustard radical is attached in the form of an amide ( X X X I ) , and the proximity of an electron-attracting carbonyl group weakens the reactivity of the chlo­

rine atoms. It was rather surprising to find that the threonine deriva­

tive [ ( X X X I ) , R = C H³, R' = C H ( N H²) C O O H ] showed no biological activity, in contrast to the serine compound referred to as mercasin or

R

I

M . C O . O . C H . R ' (XXXI)

CB 3159 [ ( X X X I ) , R = H, R' = C H ( N H²) C O O H ] , which with low toxicity exerted similar antitumor effects to those of the urethane deriva­

tives [ ( X X X I ) , R = H, R' = C H³] (61), as studied by Bushby (62).

When using phosphoric acid with = P = O a s a carrier of the mustard group, another interesting compound endoxan, Cytoxan, or cyclophospha­

mide ( X X X I I ) (68a) was developed from original work by Friedman and Seligman (63b). One of the reasons for the preparation of this drug

174 J. Μ. JOHNSON AND F. BERGEL was the possibility that it might be split by phosphoramidases of the organism to release the unsubstituted nitrogen mustard [ ( I ) , R = H ] . Indeed, its inactivity in tissue culture and its activity against various

^ C H j — H N ^

^ C H — 0 ^ [ » (ΧΧΧΠ)

animal tumors strongly supports the working hypothesis just mentioned.

Although, in the space of this article, one cannot enumerate all possible variations of carriers of the mustard group, brief mention should be made of heterocyclic systems. A large number of compounds of this type have been prepared and tested, one of which, called dopan ( X X X I I I ) , has been found by the Russians (64) to be worthy of clinical use.

OH

(XXXIII)

The corresponding uracil mustard was prepared and studied in America (65). Other heterocyclic agents belonging to the class of antimalarial structures have been mentioned in previous reviews and papers (66-68), and other types with references can be found in a list (69) published by the Cancer Chemotherapy National Service Center, Bethesda.

B. Dimethanesulfonates

Some years ago a series of methanesulfonic acid esters of the general formula ( X X X I V ) was developed by Timmis (70), two members of which have found clinical use: the C⁴-compound called Myleran [ ( X X X I V ) , η = 4] and the C⁹-compound [ ( X X X I V ) , η = 9 ] , known as nonane

(70, 71). Myleran was shown to have selective action on the circulating

CH3 CH3

I I

CH3S020(CH2)„OS02CH3 CH3S020CHCH2CH2CHOS02CH3 (XXXIV) (XXXV)

23. BIOLOGICAL ALKYLATING AGENTS 175 blood elements of rats. In contrast to the mustards, which depress the lymphocytes rather than the neutrophiles, Myleran has the reverse effect (70). Since then, it has become the drug of choice for the treatment of chronic myelogenous leukemia. Dimethyl Myleran ( X X X V ) , which turned out to be more water soluble than the parent compound, was prepared and tested (72). Its neutrophile-depressing activity was about three times greater than that of Myleran itself (72), and as it also showed a quicker response it was tried in cases of acute leukemia (73).

In view of the biological properties of a bischloroethylamino derivative of dideoxy-D-mannitol, degranol (74), the preparation of a series of dimethanesulfonyl esters of sugar alcohols was undertaken (75, 76). The compound of formula ( X X X V I ) , also a dideoxy-D-mannitol derivative, known as mannitol Myleran, was found to be an effective tumor inhibitor with a wide animal tumor spectrum (75) but had relatively little effect on bone marrow. Just as among the amino acid mustards, steric conditions in this series of polyols plays an important part, e.g., the L-mannitol diesters proved to be ineffective.

C H2O S 02C H3

H O — C — H

I

H O — C — H

I

H — C — O H

I

H — C — O H

I

C H J J O S O J J C ^

( X X X V I )

For those readers who are especially interested in the methanesulfo- nates, a list has been compiled (77) of all the methanesulfonic acid esters which have been prepared and tested against experimental tumors.

C . Ethylenimines

Three of the earlier compounds in this series (13) were triethylene- melamine ( T E M ) ( X X X V I I ) , triethylenephosphoramide ( T E P A )

( X X X V I I I ) , and triethylenethiophosphoramide (TESPA) ( X X X I X ) , T E M is effective against chronic leukemia, but appears to be more toxic than the nitrogen mustard H N 2 [ ( I ) , R = C H³] , whereas TESPA is the most widely used, being more stable than T E P A and also effective on

176 J. Μ. JOHNSON AND F. BERGEL CH,—CH,,

Ν

I

: H,

I ||

- H

/ \

CH² Ο CH² CH² S CH²

Λ ιι / i < ιι /

Ν — Ρ — Ν Ν — Ρ — Ν I \

1 /

CH² Ν CH² CH² Ν CH²

/ \ / \

C H²— C H² CH² CH²

(XXXVIII) (XXXIX) some solid tumors. A compound which is structurally related to TESPA

has now been developed (78) and is known as OPSPA ( X L ) . Biological

CH²CH2 S CH² C H X H j Ο X H²

/ 2 X II / \ 2 / 2 X II / 2

Ο Ν—Ρ—Ν Ο Ν — Ρ — Ν

\ / ι \ Ι \ / ι \

CH²CH² ^ Ν ^ CH² CH²CH2 ^ Ν ^ CH²

CH² Cf^ C H²— C I ^ (XL) (XLI) tests have shown this compound to be active against the Flexner-Jobling

carcinoma in rats, a result which is also produced by its major sulfur-free metabolite Μ Ε Ρ Α ( X L I ) . This activity can be enhanced by simultaneous administration of azaserine, aminonucleosides, or 6-mercaptopurine. Very recently, two new compounds have been reported which are structurally similar to ΜΕΡΑ. They were considered to be "duel antagonists," incor­

porating in one molecule two different chemotherapeutic moieties, viz., the bis(ethylenimine) phosphoro and carbamic acid residues. The first of these [ ( X L I I ) , R = — C²H⁶] (79) has shown significant activity against various animal tumors (80-82). The second compound [ ( X L I I ) , R = C H²C^eH⁵] has been found to possess a more pronounced antitumor activity both in animals and man (81). In fact, it was reported (79, 80)

23. BIOLOGICAL ALKYLATING AGENTS 177 CH²

| \ Ο

Ν—Ρ—NHC-OR

II

CH

1 /

I II

Ν

ο

C H 2 C H 2

(XLII)

that the benzyl analogue [ ( X L I I ) , R = C H²C⁶H⁵] produced complete cures in some D B A / 2 mice inoculated with leukemia L 1210, a result which has never before been obtained by the same workers with other agents. A further report has also appeared (83) of encouraging results ob­

tained with this drug in the treatment of bronchogenic carcinoma. Work on ethylenimino derivatives of quinones (84-86) has led to two interesting compounds, one designated Bayer Ε 39 [ ( X L I I I ) , R = — C H²C H²C H³] and the other Bayer A 139 [ ( X L I I I ) , R = — C H²O C²H⁵] . These gave some interesting results in biological and biochemical investigations and are now undergoing extensive clinical trials. The same applies to a third and later addition to this series, represented by formula (XLIV) and known as Bayer 3231 (87-89).

N o t all ethylenimine derivatives are predominantly antitumor agents.

It will be remembered that among a series of mono- and polyfunctional ethylenimines developed by Walpole et al. (90), it was demonstrated that the mono functional derivatives ( X L V ) , in contrast to the polyfunctional ones (e.g., X L V I and X L V I I ) , possessed no antitumor activity, but were carcinogenic (91, 92). The conclusion was reached that neither poly func­

tionality, nor cross-linking, nor the formation of a polyreactive polymer or micelle was necessary for the carcinogenic action of ethylenimines.

That the various biological effects of alkylating agents, viz., antitumor activity, carcinogenicity, mutagenic action, and chromosome damage do not always run parallel is shown by the differences in antifertility activity of some of their representatives, including ethylenimines (93). It was

(XLIII) (XLIV)

178 J. Μ. JOHNSON AND F. BERGEL found that Myleran [ ( X X X I V ) , η = 4] interfered with spermatogenesis in rats for several weeks, while melphalan (XI) was ineffective in this respect; T E M ( X X X V I I ) also rapidly induced infertility. This infertility was also produced by other ethylenimine derivatives, being greatest in trifunctional and least in monofunctional compounds. While the alkylat­

ing agents tested were carcinogenic, other chemical types of carcinogenic agents did not produce sterility. A recently published review (94) lists the synthetic substances which are known to interfere with animal repro­

duction.

CH2

RCON C H2

CH2 CH2

NCOHN(CH2)wNHCON

C l £ ^ C H ,

(XLV) (XLVI)

CH,

^ N O C ( C H A C O N

CH2 CH2

(XLVn) D. Epoxides a n d Others

During recent years activity in this field has not been very great, but one might mention the work of two different laboratories. Firstly, Walpole (92) et al. found that on testing various epoxides, for example ( X L V I I I - L ) , certain monofunctional members ( X L I X ) were carcino­

genic. This phenomenon is similar to that observed with the ethylenimines previously mentioned. Secondly, Gerzon et al. (95) established that the compound (LI) possessed high chemotherapeutic activity when tested against the mouse leukemia Ρ 1534, and later that compound (LII) was even more active. Both these compounds are now undergoing clinical trials.

CH2—CH—R—CH—CH,

ν ^V

^A

CH—CH

A

R—Ν

\

Ο

(XLVHI) (XLIX) (L)

23. BIOLOGICAL ALKYLATING AGENTS 179

Cfk-yCH—CHjf-ri Ν—CH2— C H - C H2

Ο Ο (LI)

C H2 — C H — C H2 — ^ ^N— C H2— CH-^CH2

(LII)

A completely different type of alkylating agent in the form of dimethyl- nitrosamine, D M N (LIII), was recently reported by Magee and Barnes

(96), who found that rats, when fed on a diet containing 50 ppm of D M N , developed primary hepatic tumors and that protein synthesis was in­

hibited. Further work by Farber and Magee (97) using D M N - C^{1 4} led to the belief that rat liver R N A was methylated, probably by diazomethane

CH3

\

N—NO / CH3

(LIII)

as an intermediate, to produce chemically abnormal products. These results support the idea (98) that alkylation of cell constituents (in the first instance D N A ) is an important factor in carcinogenesis by alkylating agents, but there exists also some evidence for the role of altered R N A in the early stages of cancer production (99, 100).

III. MECHANISM OF ACTION

The preceding sentences connect the foregoing section on newer repre­

sentatives of the family of alkylating agents with the important problem of how such agents may bring about their in vivo effects. It is safe to assume, as mentioned in the introduction, that these chemically reactive compounds interact with nucleophilic centers of receptor molecules of the cell. But first, ignoring tissue culture or ascites tumor cells in vitro, how

180 J. Μ. JOHNSON AND F. BERGEL do the drugs arrive at the receptor sites in the whole animal or man without getting lost on the way? The ideal conditions are obviously that their transport should be as smooth as possible and that few side reactions should take place, particularly with water or any irrelevant tissue con­

stituents (competition factors). Then there should be a satisfactory concentration of the agents in the target area, which depends to a large extent on the degree of its vascularization. But granted that such blood supply would serve tumors at least as well as the corresponding healthy host tissue, it is still highly desirable to have a better drug permeability of cancer cells, a lower degradation rate, and a greater number of receptor molecules than with normal cells. However, it is only too well known that by what ever detailed mechanisms these compounds achieve their effects

(in the end irreversible cell damage in the case of antitumor drugs), the ideal conditions described just now, never exist all at the same time.

Focusing attention on the "receptor molecules," the two alternatives are first, proteins, including enzymes, and second, the nucleic acids, espe­

cially D N A [a number of observations mainly by Ambrose et al. (101, 102) extend these possibilities to the lipoproteins of the cell surface].

Within these macromolecules the most favored centers for reaction under specified ionic conditions [cf. Ross (2)] are —COOH, — N H², —SH,

— P 0³H and, as will be seen below, tertiary nitrogen atoms of heterocyclic systems. It was thought that the outcome of such reactions was fatal alterations of either (a) enzymic activities, (b) reduplication processes, (c) surface properties of the cell, or any combination of these effects. On the other hand, in view of the mutagenic activity of some of the alkylating agents, and also considering the role of D N A as a genetic controlling factor (see transforming principles), the most widely accepted idea was that the difunctional alkylating agents interacted with certain groups within the D N A molecule and caused cross-linking and a change in steric configuration; yet there was no absolute certainty as to the true connec­

tion between this mechanism of action and the biological effects.

However, during the last few years, some very interesting results have been obtained giving rise to three well-supported hypotheses. One school believed, at least at the beginning, that the important reaction was the esterification of phosphate groups in nucleoprotein [Alexander et al.

(103)]. Another school came to the conclusion that the major interaction was that with the nitrogen atoms of purine or pyrimidine bases in nucleic acids [Skipper; Lawley and Wallick; Butler and Press (104-108)], while the third school defended the hypothesis of an interaction with sulfhydryl groups of essential proteins [Roberts and Warwick (109-112)]. Con­

sidering first the evidence for the interaction with nitrogen atoms of the

23. BIOLOGICAL ALKYLATING AGENTS 181 bases, Lawley and Wallick (105) showed that the action of the nitrogen mustard H N 2 on D N A at pH 7 and at a temperature of 37° changed the shape of the D N A absorption curve, indicating alkylation of purine or pyrimidine rings. Similar reactions with nucleotides suggested that the guanine residue was most susceptible to alkylation, which was also con- firmed by acid hydrolysis of the alkylated D N A . A new component was found, the spectrum of which was compared with those of alkylguanines and found to resemble that of a 7-alkylguanine. This led the authors to believe that the intermediate of the alkylation was a quaternary salt. Sup- port for this was obtained by comparing the alkylation of 1,7-dimethyl- guanine (LIV) by methyl sulfate, as a model agent, with that of guanylic acid (LV).

HO (LV)

The outcome of this study was that the products obtained from guanylic acid were 7-methylguanosinium-9-ribotides [ ( L V I ) , R = ribose phos- phate] .

(LVI)

In alkaline solution at 23°, the quaternary salt underwent a reaction leading to products with different spectra, which were thought to be ring fission products in which the ribose residue was still retained. It is interesting to note here that when Hems (118) subjected a solution of

182 J. Μ. JOHNSON AND F. BERGEL guanylic acid to ionizing radiation by X-rays the main product obtained was ( L V H ) , from which it was inferred that splitting of the imidazole ring had taken place. This establishes a link between the effects of X-rays

and of alkylating agents which have been termed "radiomimetic com­

pounds" in the past.

In continuation of his work, Lawley (106) estimated the relative re­

activities of deoxyribonucleotides towards methyl sulfate at pH 7 and at a temperature of 37°. He showed that guanine was more readily attacked than cytosine, which behaved similarly to adenine, and that thymine was the most resistant base. The same set of experiments also brought to light the fact that alkylation of sugar phosphate residues only took place to a small extent (but see below) and that the main reaction was directly with the bases. The same author also established that these bases still possessed the same order of reactivity when present as con­

stituents of D N A , and that attack by alkylating agents could be expected to be greater on nucleic acids containing a preponderance of guanine and cytosine residues rather than of adenine and thymidine units. In addition, when deoxyguanylic acid was methylated with methyl sulfate at pH 7.2 and at a temperature of 37° (107), 7-methylguanine was again obtained by hydrolysis, but under much milder conditions than for the guanylic acid product, degradation being appreciable at pH 7. Unless a rearrange­

ment accompanied its hydrolysis, the intermediate product would be of the same type as formula (LVI), but with R = deoxyribose phosphate.

All of this demonstrated that alkylation of the guanine residues of D N A could lead to considerable changes in the characteristics of the macro- molecule due to the quaternization of nitrogen groups, as well as to cleavage of the sugar phosphate residue from the alkylated guanine moieties. This does not mean that some adenine units were not attacked at the same time. Brookes and Lawley (114) showed that methylation of adenosine or adenylic acid under neutral conditions gave mainly the 1-methyl derivative together with 3-methyl and probably 1,3-dimethyl derivatives. In this respect, alkylating agents differ from ionizing radia­

tion in that the latter, according to Hems (113), causes a ring-opening OH

(LVH)