Synthesis of boron-cluster containing amino acids and preparation of their polymers

(1)

Synthesis of boron-cluster containing amino acids and

preparation of their polymers

Vom Fachbereich Biologie/Chemie der

Universität Bremen genehmigte

DISSERTATION

zur Erlangung des Grades eines

Doktors der Naturwissenschaften

-Dr. rer. nat. -

von

Irina N. Slepukhina

aus

Ekaterinburg (Russland)

Bremen 2006

(2)

Tag des öffentlichen Kolloquiums:

26.04.2006

Gutachter der Dissertation:

1. Gutachter: Prof. Dr. Detlef Gabel

2. Gutachter: Prof. Dr. Dieter Wöhrle

(3)

I would like to express my sincere gratitude to the following persons:

• My supervisor, Prof. Dr. Detlef Gabel, for accepting me as a PhD-student, for the interesting theme of my PhD-work, for help, good advice and a lot of patience.

• Prof. Dr. Dieter Wöhrle for being my co referent.

• Prof. Dr. Gerd-Volker Röschenthaler for being my examinator.

• Dr. Thomas Dülcks for help in recording and analyzing the mass spectra, for introducing me to the theory of mass spectrometry.

• Dr. Klaus Rischka for help in recording and analyzing the MALDI-TOF spectra. • Renate Alberts, for support and helpfulness.

• Barbara Hoffmann-Gabel, for support and advice in my oral presentations.

• Members of working group of Prof. Detlef Gabel for nice atmosphere during the work and many interesting discussions and advice.

• The Ernst A. C. Lange Foundation for support. • My parents for their continuous encouragement.

(4)

1. Introduction

1.1. What is BNCT?

Boron neutron capture therapy (BNCT) is an investigational form of a two-part radiation therapy which has the potential ability to selectively kill tumor cells embedded within normal tissue. The concept of BNCT was introduced in 1936, four years after the discovery of neutrons. The idea of the therapy is simple and elegant. A tumor-seeking compound containing the stable isotope 10B is introduced into blood and given time to be accumulated in the tumor. The tumor is then irradiated with epithermal neutrons, which are captured by the 10B isotope. Capturing neutrons causes the boron nuclei to break apart, resulting in the emission of α-radiation and recoiling 7_Li

nuclei. For every reaction, there is a release of 2.79 MeV of energy (Figure1).

B 1n 5 10 5 [ B]11 0 He Li He Li 2 4 2 4 7 7 3 3 γ0.48 MeV 2.79 MeV (6%) 2.31 MeV (94%) Figure 1. Mechanism of BNCT (www.osaka-med.ac.jp).

The ion products 4He and 7Li are strongly cell toxic and have short ranges, about 9 and 5 µm in tissue respectively. A chemical compound containing atoms of the element "boron" is infused intravenously into the subject’s body and concentrates more in tumor cells (e.g., glioblastoma or melanoma) than in normal cells. After the tumor cells have been selectively loaded with boron in this way, the tumor site is irradiated with neutrons from a small research nuclear reactor. Neutrons are atomic radiation particles which in the absence of boron have a less harmful effect on tissue. However, the absorption of these neutrons by the boron atoms in the tumor cells causes the boron atoms to emit alpha particles. Alpha particles are also atomic radiation

(8)

particles, but the distance they travel is about the diameter of a tumor cell, so any surrounding normal cells would be much less affected by the alpha radiation.

1.2. Criteria for BNCT agents

From previous experiences, some criteria for an “ideal” BNCT agent have been selected (Panza 2003). In order to be therapeutically useful, an ideal boronated candidate should have:

• High tumor targeting selectivity: BNCT agents should accumulate preferentially into tumour tissues. They must also clear the blood rapidly to avoid inducing necrosis in the vasculature. The optimal tumor: blood ratio is around 5:1.

• Low toxicity: this is a challenging aspect when one considers the amount of agent that must be administered to achieve the requisite levels of boron in the tumor.

• Proper water solubility: required for intravenously administration of the BNCT agent. • High uptake by cancer cells: the therapeutically useful amount of boron in tumor cells

usually accepted is between 10 and 35 µg of 10_{B per gram of tumor. This amount is}

substantially reduced if the boron is concentrated in or close to the cell nucleus.

1.3. Classes of third generation boron compounds

The structures that may be classified as “third generation” compounds have, in general, a more biochemical and physiological basis for achieving tumor cell selectivity that the structures that have been previously designed and synthesized (Panza 2003). Those compounds can be divided in to different categories as shown in the following list:

1. Cellular Building Blocks

a. Boron-Containing Nucleic Acid Precursors

b. Boron-Containing Amino Acids and Related Peptides c. Lipids and Phospholipids

d. Carbohydrates 2. Lipoproteins

3. Liposomes

4. Porphyrins and Phthalocyanines 5. DNA Binders

a. Alkylating Agents b. Intercalators c. Groove Binders d. Polyamines

(9)

e. Di- and Oligonucleotides: Antisense Agents 6. Receptor/Antigen Binders a. Antibodies Monoclonal/Bispecific b. Growth Factors c. Hormones 7. Other Compounds a. Radiation Sensitizers b. Miscellaneous

1.4. Preparative methods for synthesis of α-amino acids

This thesis is concerned with the synthesis of α-amino acids. Therefore, their methods of preparation are reviewed.

1.4.1. Direct ammonolysis of α-halo acids

An α-halogen-containing acids is treated with a large excess of concentrated aqueous ammonia. For example:

CH3CH2COOH

Br2, P _CH

3CHCOOH NH3 (excess) CH3CHCOO

Br NH3

Propionic acid

α-Bromopropionic acid

Alanine 70%

Scheme 1. Direct ammonolysis of an α-halo acids.

This method is used for synthesis of amino acids with the amino group situated at any distance, but appropriate β-, ω-halogen substituted acids are less available than α- halogen substituted ones.

1.4.2. From potassium phthalimide

This method is a modification of the Gabriel synthesis of amines. The yields are usually high and the products are easily purified.

(10)

N-K+ O O + ClCH2CO2C2H5 N O O CH2CO2C2H5 1)KOH/H2O 2)HCl 97% CH2CO2 NH3 + CO2H CO2H C2H5OH + Potassium

phthalimide Ethyl chloroacetate

Glycine

85% Phthalic acid

Scheme 2. Synthesis of amino acids from potassium phthalimide.

A variation of this procedure uses potassium phthalimide and diethyl α-bromomalonate to prepare an imido malonic ester.

1.4.3. The Strecker synthesis

Treating an aldehyde with ammonia and hydrogen cyanide produces an α-amino nitrile. Hydrolysis of the nitrile group of the α-amino nitrile converts the latter to an α-amino acid. This synthesis is called the Strecker synthesis.

CH3 O H NH4CN CH3 H C NH2 CN -H2O H2O(H+) CH3 H C C NH3 O O Aminonitrile Alanine

Scheme 3. The Strecker synthesis.

1.4.4. Hydrolysis of hydantoins

When hydantoins are heated for relatively long periods of time with a large excess of barium hydroxide in aqueous solution the α-amino acids are obtained. This property of hydantoins has been of very great value in the synthesis of α-amino acids which are difficult to obtain by other methods. The hydantoin synthesis of α-amino acids was first suggested as a general method by Wheeler and Hoffman, and was used by them in the synthesis of phenylalanine and of tyrosine.

(11)

Ba(OH)2 _HOC 6H4CH2CHCOO NH3 CO2 H2O NH3 HN N H CHCH2C6H4OH O O

Scheme 4. Hydrolysis of hydantoins.

1.4.5. Malonic ester synthesis

A number of syntheses were realized through malonic ester by the following ways:

1.4.5.1. First, an alkylmalonic ester is synthesized from Na malonic ester. The halfhydrazide – halfester is received by treatment with hydrazine. The α-amino acid is obtained after treatment with nitrous acid and Curtius rearrangement.

COOC2H5 CH COOC2H5 Na RCl COOC2H5 HC R COOC2H5 NH2NH2 COOC2H5 HC R C NHNH2 O HNO2 COOC2H5 HC R C N=N=N O H2O(H+) R H C C NH3 O O OH

-Scheme 5. Malonic ester synthesis (way 1).

1.4.5.2. Another way of synthesis through malonic ester includes its nitrosation and following reduction of nitrosomalonic ester to aminomalonate. Then the amino malonic ester is acetylated and alkylated via its sodium derivative.

COOC2H5 CH2 COOC2H5 HNO2 COOC2H5 C COOC2H5 HON H2/Ni COOC2H5 CH COOC2H5 H2N (CH3CO)2O CH3 C O H N COOC2H5 CH COOC2H5 Na, than RCl CH3 C O H N COOC2H5 C COOC2H5 R H2O(H+) R C(COOH)2 NH2 R HC NH3 C O O

(12)

1.4.5.3. The third way includes previous described methods of synthesis of amino acids. COOC2H5 CH COOC2H5 Na C6H5CH2Cl COOC2H5 HC CH2C6H5 COOC2H5 KOH HCl COOH HC CH2C6H5 COOH Br2, ether, reflux COOH C CH2C6H5 COOH Br heat C6H5CH2CHCOOH Br NH3 (excess) C6H5CH2CHCOO NH3 Phenylalanine 35% overall yeld

Scheme 7. Malonic ester synthesis (way 3).

1.4.6. Reduction of oximes, hydrazones and phenylhydrazones of

α-ketoacids

CH3 C C O OH O CH3 C NOH C OH O CH3 C C N NH2 OH O CH3 C C N NHC6H5 OH O Acetoacetic acid H+ CH3 H C NH3 C O O

Scheme 8. Reduction of oximes, hydrazones and phenylhydrazones of α-ketoacids

.

These synthetic amino acids obtained by the methods described so far are, of course, optically inactive, and must be resolved if the active materials are desired for comparison with the naturally occurring acids or for peptide synthesis.

1.4.7. Resolution of DL-amino acids

One of the methods for resolving amino acids is based on the use of enzymes called deacylases. These enzymes catalyze the hydrolysis of N-acylamino acids in living organisms. Since the active site of the enzyme is chiral, it hydrolyzes only N-acylamino acids of the L configuration.

(13)

When it is exposed to a racemic modification of N-acylamino acids, only the derivative of the L-amino acid is affected and the products, as result, are separated easily.

DL-RCHCO2 NH3 racemic form (CH3CO)2O DL-RCHCO2H CH3CONH deacylase CH3CO2H H3N H CO2 R H CO2 R CH3CONH

L-Amino acid D-N-Acylamino acid

Easily separated

Scheme 9. Resolution of DL-amino acids

.

1.4.8. Stereoselective synthesis of amino acids

This method developed by Bonish is based on a rhodium complex with (R)-1,2-bis(diphenyl- phosphino)propan, a compound that is called “(R)-prophos”. When a rhodium complex of norbornadiene (NBD) is treated with (R)-prophos, the (R)-prophos replaces one of the molecules of norbornadiene surrounding the rhodium atom to produce a chiral rhodium complex.

C CH2 H

(C6H5)2P P(C6H5)2

(R)-prophos

[Rh(NBD)2]ClO4 (R)-prophos [Rh((R)-prophos)(NBD)]ClO4 NBD

Chiral rhodium complex H3C

Scheme 10. Chiral rhodium complex.

Treating this rhodium complex with hydrogen in a solvent such as ethanol yields a solution containing an active chiral hydrogenation catalyst, which probably has the composition [Rh((R)-prophos)(H)2(EtOH)2]+. This can convert 2-acetylaminopropenoic acids stereoselectively to

(14)

C C R H CO2H NHCOCH3 Z-3-Substituted 2-acetylaminopropenoic acid 1) [Rh((R)-prophos)(H)2(solvent)2]+, H2 2) OH-_{, H} 2O, heat then H3O+ C H R CO2 H3N L-Amino acid

Scheme 11. Stereoselective synthesis of amino acids.

1.5. The development of boron-containing amino acids and related

peptides

The interest in the development of boron-containing amino acids and related peptides is due certainly in part to the fact that 4-dihydroxyborylphenylalanine (BPA) is one of the two clinically used BNCT agents. The view is that such cellular building blocks may be required to a greater extend in more rapidly proliferating cells (i.e., tumor cells) that in their normal counterparts. This requirement may lead to the synthesis of peptides and proteins derived from these amino acids and thereby their retention in the cell’s matrixes. This, once again, would meet the objective for both tumor targeting and retention.

In addition to the para isomer of BPA (Snyder 1958; Malan 1996; Kirihata 1996; Samsel 1992), the ortho and meta positional isomers have been synthesized and evaluated (Yoshino 1993; Yoshino 1996a; Yoshino 1996b; Kuszewski 1968). The ortho isomer exists as a cyclic internal anhydride (Kuszewski 1968). In Figure 2, dihydroxyboryl analogues of phenylalanine that have been prepared are shown.

B HO OH NH2 OH O O B HO OH NH2 OH B HO OH

para-BPA meta-BPA ortho-BPA

NH2 OH O

Figure 2. Dihydroxyboryl analogues of phenylalanine.

Only the para isomer of BPA has been subjected to extensive biological evaluation. Early on, it was recognized that the compound’s low aqueous solubility was a hindrance to its intravenous administration. To overcome this problem, researchers initially used the ability of the boronic acid moiety to complex with carbohydrates as a means of increasing the compound’s aqueous

(15)

solubility (Kinoshita 1992; Honda 1992; Yoshino 1989). This has been achieved by using a fructose complex, which has been used in clinical studies. Others have attempted to increase water solubility significantly by the incorporation of nonionic, hydrophilic groups into BPA (Nemoto 1993; Nemoto 1995; Takagaki 1996). The latter would obviously change the biochemical nature of the amino acid. In Figure 3, structures are presented in which the carboxyl functional been converted to mono-, di-, and tetrahydroxy amide derivatives.

B OH OH H2N H N O O O OH OH OH OH B OH OH H2N H N O B OH OH H2N NH O B OH OH H2N OH O OH OH OH 18_F

Figure 3. Hydrophilic derivatives of BPA, and a positron-emitting analogue of BPA (18F-BPA).

One limitation in evaluating the pharmacodynamics of any of these boron compounds is the inability to study their real-time in vivo kinetics. There are no useful boron radionuclides, and for this reason 18F-BPA was synthesized in which the radiolabel was inserted into the 2 position of the aromatic ring (Ishiwata 1992; Ishiwata 1991). The utilization of this structure is based on the fluoro analogue of BPA being completely analogous to BPA itself in terms of its uptake and retention by tumors. Positron emission tomography of 18F-BPA in patients with malignant brain tumors has demonstrated the targeting ability of BPA (Ueda 1996; Kabalka 1997a; Imahori 1997).

An important question about BPA was which of the para isomers, the D and L analogues, was more useful in tumor targeting. The view was that the isomer analogous to the naturally occurring amino acid (L) should be more effective. In vivo experiments in tumor-bearing animals and in vitro cell culture studies have shown that this is the case, supporting the assumption the amino acid transport system may be operative for L-BPA in achieving elevated

(16)

tumor concentrations (Papaspyro 1994; Belkhou 1995; Coderre 1987). A number of different dihydroxyboryl-containing amino acid has been synthesized. These include those related to aspartic acid and cysteine (Matteson 1964) but little has been reported on their potential as BNCT agents. Other research has focused on the replacement of carboxyl group in the amino acid by a dihydroxyboryl group (Coutts 1996; Kettner 1984). The rationale was that such compounds may act as transition-state inhibitors of enzymes but little has been done with respect to their possible use as BNCT compounds.

Not all of the syntheses of boron-containing amino acid have focused on those containing the boronic acid moiety. Others have incorporated the borane zwitterionic group. One example is ammoniacarboxyborane, H3NBH2COOH, which may be viewed as an isomer of glycine

(Spielvogel 1989). This development has led to synthesis of a number of borane-containing amino acids shown in Figure 4, and distribution studies in tumor-bearing animals are described. Understandably, there has been interest in making analogues that contain boron clusters, and it was for this reason that one of the earlier compounds described was o-carboranylalanine (o-Car).

N H2 B O (H3C)3N O OCH3 B H3C NH2 O NHEt OH H2 B O H3N

Figure 4. Boron-containing amino acid derivatives.

It was first synthesized independently by Zakharkin et al. (1970) and Brattsev et al. (1969). The compound was prepared as the racemic mixture. Subsequently, other methods for its preparation in higher yield have been described (Wyzlic 1992; Wyzlic 1993; Wyzlic 1996a), as well as the stereoselective synthesis of the L-isomer ((S)-configuration) initially by Schwyzer et al. (1979). More recently, Kahl (1993; 1994; 1996a), Sjöberg (1992; 1993) and Moroder (1995) have independently developed more useful stereoselective syntheses and received both the L and D forms (Figure 5).

(17)

C C COOH NH2 H C C COOH H H2N D-configuration (R)-o-carboranylalanine L-configuration (S)-o-carboranylalanine C C Br N O O O Ph TMGA, CH2Cl2, -23oC TMGA=tetramethylguanidinium azide C C N O O O Ph N3 Ti(Oi-Pr)4, BzOH, 70oC C C OBz O N3 H2, Pd/C, EtOH R or S isomers 89% 89% N N H3C Boc O CH2 C CH B10H10 N N H3C Boc O C H2 C C 1) CF3CO2H 2)HClaq,acidic cation exchange resin H2 C C C C HO2C H H2N R isomer (R)-o-carboranylalanine

Figure 5. The L- and D-enantiomers of o-carboranylalanine.

Since the carboranyl group approximates the phenyl group in its three-dimensional sweep, it was envisaged that this amino acid may simulate phenylalanine in its biochemical properties. Of special interest was a biological assessment of o-Car and its in vivo comparison with the clinically useful BPA. Improved synthetic procedures for o-Car as well as greater understanding of its physicochemical properties were necessary. These provided the basis for the studies with tumor-bearing mice, demonstrating that o-Car attained higher blood concentrations and lower tumor levels than did BPA at comparable time intervals and that the former showed no evidence of tumor selectivity (Yong 1995a; Yong 1995b; Pettersson 1993a; Pettersson 1993b; Pettersson 1994). The fact that the carboranyl moiety is highly lipophilic and 112 times more hydrophobic than the indol side chain of tryptophan may account for its persistence in blood through noncovalent association with blood lipids (Soloway 1998). These results are supported by previous studies in which phenylalanine (Phe) and tyrosine (Tyr) residues in various bioactive peptides and polypeptides are replaced with the o-Car moiety (Schwyzer 1979; Fischi 1977; Leukart 1979). Such analogues were also biologically active and, in some instances, showed a prolongation of activity by comparison with their Phe counterparts. These results demonstrate

(18)

the need to reduce the lipophilicity of the carboranyl group by the incorporation of functionalities that will balance the compound’s lipophilic properties.

At this time, a number of different carborane-containing amino acids have been synthesized and evaluated (Jacobs 1976; Malmquist 1996a; Wyzlic 1992; Wyzlic 1993; Wyzlic 1996a; Wyzlic 1996b; Malmquist 1996b; Olsson 1996; Varadarajan 1991; Prashar 1993a; Nishiaki 1997). Not all of these are classical α-amino acids (Beletskaya 2004) (Figure 7). A very interesting amino acid structure has been prepared by Kahl and Kasar (Kahl 1996b) involving o-, m-, and p-carboranes in which the amino carboxyl functions are located on the different carbon atoms of the carborane nucleus. Another category is described by Kabalka et al. (Kabalka 1997b; Kabalka 1997c) in which either a m-carborane or a nido-carborane nucleus is attached to 1-aminocyclobutanecarboxylic acid. The low aqueous solubility of these o-carborane-containing amino acids is a key disadvantage once again in their biological evaluation.

Recent positron emission tomography (PET) investigations carried out at the University of Tennessee on BNCT patients using fluorine-18 labeled BPA and carbon-11 labeled 1-aminocyclobutanecarboxylic acid revealed that cyclic amino acids localize in Glioblastoma multiforme (GBM) and malignant melanoma (MM) tumors more avidly than BPA (Hübner 1998). For this reason several amino acids with quaternary α-carbon atom were synthesized (Figure 6) (Kabalka 1999; Kabalka 2001; Zaidlewicz 2004).

(HO)2B (CH2)n NH2 CO2H n=0; 2; 3; 7 H2N CO2H n B OH OH n=1; 3 (HO)2B (HO)2B NH2 CO2H CO2H NH2

Figure 6. Amino acids with quaternary α-carbon atom.

Hawthorne et al. (Varadarajan 1991) used the degradation of the o-carborane moiety by base to produce the negatively charged nido-carborane species. Such structures show improved water solubility. An alternate approach considered by him and others is to attach polyol structures to such compounds (Hawthorne 1991). It has not been determined what is the appropriate balance between the lipophilic/hydrophilic properties that the amino acids must possess.

(19)

Another very interesting carborane-containing amino acid was generated by the insertion of the carborane group into 3,4-dihydroxyphenylalanine (DOPA) (Prashar 1993b). This compound, in contrast to BPA, would not have to lose the boron moiety in order to generate the indolequinones, the precursors of melanin that occur in significantly elevated amounts in melanomas. The structure is shown in Figure 7.

C C C C CC C C C C C C H _CH 3 COOH H2N NH2 O OH NH2 COOH HO OH H2N _COOH O H2N COOH NH2 COOH

Figure 7. Carborane-containing amino acid analogues.

In addition to carborane-containing amino acids, a polyhedral borane anion analogue of methionine has been described (Soloway 1998) (Figure 8).

S S S CH3 H2N O HO CH3 H3C S H N O COOH H2N O N H COOEt

2-Figure 8. Boron-containing analogue of methionine,

DL-S-(10-dimethylsulfidooctahydrodecaboranyl)methionine, and BSH-glutatione disulfide derivative.

(20)

The next examples of amino acids derived from the closo-dodecaborate and the cobalt bis(1,2-dicarbollide) anion were prepared by the group of Bregadze (Sivaev 2000; Sivaev 2002) by the ring–opening reaction with acetamidodiethylmalonate (glycine anion equivalent for amino acid synthesis) in acetonitrile in the presence of potassium carbonate, followed by acidic hydrolysis and decarboxylation (Figure 9).

C C C Co C (OCH2CH2)2CHCOOH NH3 O(CH2CH2)2CHCOOH NH2

2-Figure 9. Amino acids, synthesized by ring-opening reaction.

The synthesis of these boron-containing amino acids stimulated interest in the development of boron-containing peptides. The rationale was based on the projected increased need by tumor cells for the protein precursors and the feasibility that small peptides may cross cellular membranes and be utilized by tumor cells. Some boron-containing di- and tripeptides were derived from the zwitterionic borane-containing amino acid analogues and their coupling to various amino acids (Sood 1990). Examples of these are shown in Figure 10. Interest in such compounds primarily arose from their hypocholesterolemic and triglyceridemic activity and not solely from their BNCT potential.

B H2 H3N N H O NH2 O OH H N R O B R' OH OH B H2 (H3C)3N N H O O O CH3 B H2 (H3C)3N N H O H N O OEt O

Figure 10. Boron-containing peptides.

A second effort in the development of boron-containing peptides focused on the use of carboranylalanine by Schwyzer et al. (Fischli 1977; Leukart 1979; Escher 1981; Schwyzer 1981). Work by Hawthorne et al., using o-carborane-containing amino acids of both the closo

(21)

and nido analogues, was done in conjunction with the development of boron-containing antibodies (Paxton 1992; Kane 1993; Varadarajan 1991). Others related to their use as tumor targeting agents and involved racemic mixtures as well as enantiomeric structures. Representative compounds are shown in Figure 11.

C C C C C C H2N O N H O HO TBDMS H N O But O O NH2 H _CH 3 H N O HO S O O N CH3 CH3 2

Figure 11. Carborane-containing peptides.

There has been little advantage in developing peptides that could be formed from BPA together with other amino acids since these would possess a reduced boron percentage in comparison to BPA itself. However, they might provide very useful information as to the types of dipeptides that achieve tumor cell uptake and persistence. Nevertheless, the major emphasis in developing peptides has been with boron cluster, the carboranes, as well as the polyhedral boran anions. An example of the latter is the glutathione analogue, shown in Figure 8, whose synthesis (Soloway 1998) was driven by the improved tissue distribution characteristics of BSH (B12H11SH2-) and

BSSB (B24H22S24-) when used in concert with glutathione.

The preparation and evaluation of small boron-containing peptides is in an early stage of development by comparison with BPA and other boron-containing amino acids. It remains to be determined whether the rationale for their development, namely, the improved tumor to normal tissue and blood ratios, as well as enhanced tumor concentration compared to the amino acids from which they are derived, can be achieved. In this area as well, developing more hydrophilic analogues of peptides and proteins has been viewed as an important objective (Yanagie 1997; Nagasawa 1990).

(22)

2. Aim of the work

As mentioned above (paragraph 1.2), successful cancer therapy requires the selective accumulation of ~5-30 ppm 10B in tumor. A potential method for the delivery of boron to tumors is an antibody-based approach which requires each immunoprotein to deliver ~103 boron atoms in order to achieve the required 10B concentrations. This requirement presents a chemical issue which has led us to aim of chemical synthesis of water-soluble 10B-rich “macro” molecules.

2.1. Synthesis of [B

12

H

12

]

2-

containing amino acids

The aim of this part of work is the synthesis of water-soluble, [B12H12]2- containing amino acids

as boron carrier for BNCT. Water-soluble functionalized derivatives of the dodecahydro-closo-dodecaborate anion [B12H12]2- are promising candidates for boron neutron capture therapy

(BNCT) for cancer. Such boron containing structures allow obtaining molecules that bring a large number of boron atoms per molecule. For the anion [B12H12]2- the amount of boron per

weight of ion is about 92%.

The traditional approach for the functionalization of the [B12H12]2- anion consists of the

introduction of a reactive centre such as an ammonio- (Sivaev 1999; Peymann 1997; Grüner 1997), mercapto- (Nagasawa 1990; Gabel 1993), or hydroxy- (Peymann 1996; Semioshkin 1996) group into the boron cage, followed by attachment of the side chain containing a functional groups. As a side chain we should use such kind of functional group which can give easily an amino acid structure. From preparative methods of synthesis of α-amino acids described above we decided to use, first, the modified method from malonic ester and, second, the hydrolysis of hydantoins.

Acetamidodiethylmalonate is a very convenient precursor for synthesis of amino acids. After hydrolysis it gives amino acids as a racemic mixture.

Synthesis of amino acids from hydrolysis of hydantoins seems for us an interesting possibility for synthesis of α-alkyl amino acids. Such kind of amino acids already have pharmaceutical application (Fedel 1987). It will be also interesting to use them for the second aim of this work – synthesis of peptides.

2.2. Polymerization of synthesized amino acids to peptides

The second goal of this work is the polymerization of newly synthesized [B12H12]2- containing

(23)

anhydride (NCA or Leuchs’ anhydride) was chosen. Polymerization of Leuchs’ anhydride can be started with, e.g., a fluorescent amine, which would allow detection and quantification. The advantage of this idea is to produce polymers which carry boron in each monomer unit. There is only a single amino group and a single carboxyl group at each end of polymer.

2.3. Characterization

The obtained products should be identified by 1H-, 13C-, 11B-NMR, mass spectrometry, IR spectra.

2.4. Toxicity test

After synthesis of water-soluble amino acids, we cannot neglect their biological test as a BNCT agent. Toxicity tests, by using V 79 Chinese hamster cells, will give to us the first necessary information on toxicity, required for future application of these compounds.

(24)

3. Results and discussion

3.1. Synthesis of B

12

H

122-

containing amino acids

For achievement of the task of this thesis three objects of investigation were chosen: mercapto-undecahydro-closo-dodecaborate (BSH) 2, ammonio-mercapto-undecahydro-closo-dodecaborate (BNH3)

3, hydroxo-undecahydro-closo-dodecaborate (BOH) 4 (Scheme 12). Methods of their preparation are known and described in the literature (see paragraph 2.1).

2-SH NH3 OH 2- - 2-1 2 3 4

Scheme 12 Objects of investigation.

Our first trial was connected with the cyanoethyl derivative from BSH (CE-BSH). Gabel and co-workers (Gabel 1993) investigated the alkylation reaction of BSH and halides. They showed that cyanoethyl group could be used as a convenient protective group for the sulfur of BSH.

This is illustrated again (Scheme 13) by reaction of 5 with 4-bromobutylacetamidodiethylmalonate to form 6, and then further to form the mono-substituted derivative 7 with a strong base, such as tetramethylammonium hydroxide. It is necessary to notice that in the case of 4-chlorobutylacetamidodiethylmalonate this reaction failed.

(25)

S (CH2)4 CO2C2H5 CO2C2H5 NHCOCH3 NC -TMA+ S NC 2-2TMA+ 5 6 (CH2)4 CO2C2H5 CO2C2H5 NHCOCH3 X X=Cl, Br in case of X=Br 2TMA+ N(CH3)4OH S (CH2)4 CO2C2H5 CO2C2H5 NHCOCH3 2-7

Scheme 13. Synthesis of S-(5-bis(ethoxycarbonyl)-5-acetamidopentyl)-thio-undecahydro-closo-dodecaborate(2-) bis-tetramethylammonium salt.

Hydrolysis of derivative 7 was carried out in 6M NaOH for 3 days of reflux. Since the tetramethylammonium salt of this amino acid dissolves in water, we isolated 9 as tetraphenylphosphonium salt, which is not water-soluble (Scheme 14).

2TMA+ S (CH₂)₄ CO₂C₂H₅ CO₂C₂H₅ NHCOCH3 2-S (CH2)4 _COOH NH2 2Ph₄P+ 2-NaOH 2TMA+ S (CH₂)₄ CO₂H NH2 2-7 8 9 Ph₄PBr

Scheme 14. Hydrolysis of S-(5-bis(ethoxycarbonyl)-5-acetamidopentyl)-thio-undecahydro-closo-dodecaborate(2-) bis-tetramethylammonium salt 7.

Since our task was to synthesize water-soluble amino acid, we should change the cation of compound 9. Tetraphenylphosphonium salts are very well soluble in polar organic solvents such as methanol, acetonitrile. To receive water-soluble amino acids it is possible to exchange Ph4P+

to Cs+ with CsF in methanol.

Alkylation of 5 with chloroacetone gives the sulfonium salt 10. Such kind of ketone can easily form hydantoins in a Bucherer-Berg reaction (Ware 1950). We decided to carry out this reaction from the cyanoethyl derivative 10 and the deprotected ketone 11 (Scheme 15). In the first case, the cyanoethyl group was removed during the reaction. The general procedure is to warm the ketone derivative with 2 moles of sodium cyanide and 5 moles of ammonium carbonate in 50 per

(26)

cent ethanol for 4 hours. The preparation of the hydantoin from the sulfonium salt 10 to 12 directly gives only 30% of yield, whereas the overall yield of the second way (from 10 via 11 to 12) is 60-65%. Probably, the basicity of the bases NaCN and (NH4)2CO3 is not enough for

elimination of acrylonitrile from 10. Prolonged alkaline hydrolysis gives amino acid 13 with a quaternary α-carbon atom.

S -NC O CH3 TMA+ S 2-2TMA+ O N(CH3)4OH _CH 3 10 11 S N H NH O O 2-2TMA+ 1) NaOH, HCl 2) Ph4PBr S 2-2Ph4P+ COOH NH2 NaCN, (NH4)2CO3 12 13 NaCN, (NH4)2CO3

Scheme 15. Two pathways of synthesis of hydantoin 12.

For the preparation of oxygen derivatives we used an analogous synthetic way (Scheme 16). The essential difference between the schemes lies in the boron-carring agents. Alkylation of the hydroxyl group attached to the boron cluster 4 could be achieved in acetone using K2CO3 as a

base. Similar to Peymann (Peymann 1996) we showed that alkylation of BOH afforded monoalkylated derivatives.

(27)

2-2) CsF, methanol R-X: Br(CH2)4C(COOEt)2(NHCOCH3) ClCH2COCH3 O 2-2Cs+ NaCN, (NH4)2CO3 O (CH2)4 COOH NH2 R 1) NaOH, HCl 2) Ph4PBr 2-2Ph4P+ O 2-2Cs+ NH HN O O 1) NaOH, HCl 2) Ph4PBr O 2-2Ph4P+ COOH NH2 16 17 18 OH 2But4N+ 1) K2CO3 acetone R-X 4 14: R=-(CH2)4C(COOEt)2(NHCOCH3) 15: R=-CH2COCH3 in case of 14 in case of 15

Scheme 16. Synthetic ways of preparation of amino acids 16 and 18.

The next step of synthesis was connected with alkylation of ammonio-undecahydro-closo-dodecaborate. Various N-substituted [B12H12]2- derivatives had already been described (Peymann

1977; Justus, unpublished results). It was shown that the use of short-chain alkylating agents results in the formation of trialkyl derivatives [B12H11NR3]- (R=CH3, C2H5). An increase of the

substituent size leads to the appearance of sterical hindrance and formation of dialkyl derivatives (Justus, unpublished results). The sterical structure of the acetamidodiethylmalonate fragment gave to us hope to receive mono-substituted compounds. This reaction was carried out in acetonitrile in the presence of KOH. The presence of the base is necessary to deprotonate the ammonium nitrogen atom. But in contrary to our expectations we obtained a mixture of mono- and dialkyl derivatives 20 and 22 (Scheme 17).

(28)

NR (CH2)4 CO2C2H5 CO2C2H5 NHCOCH3 (CH2)4 -TMA+ NH2R -Br NHCOCH3 CO2C2H5 CO2C2H5 TMA+ C2H5O2C C2H5O2C NHCOCH3 RHN (CH2)4 CO2C2H5 CO2C2H5 NHCOCH3 -TMA+ KOH 3 R=H 19 R=CH3 20 R=H 21 R=CH3 22 R=H 23 R=CH3

Scheme 17. Alkylation of ammonio- (3) and methylammonio- (19) undecahydro-closo-dodecaborate.

Our trials to separate them by TLC chromatography (diethylether:acetonitrile, different mixtures) did not bring us any results. The same reaction was repeated with

methylammonio-undecahydro-closo-dodecaborate, in which one atom of hydrogen in the ammonium group is changed to a

methyl group. Also with this compound the same problems were obtained. Synthesis of the monoalkylated derivatives is problematic.

Recently, Sivaev and co-workers (Sivaev 1999) described another possibility for preparation of mono-substituted derivatives of BNH3. They reported about synthesis of Schiff base, derived

from the reaction of BNH3 and different aldehydes (HC(O)R, where is R=C6H5, 4-C4H4Cl,

CH=CHCH3, CH=CHC6H5, etc). After reduction of the Schiff base with NaBH4 they received

the appropriates amines.

With this approach we decided to make an analogous reaction with aldehyde 24 (Scheme 18). The reaction was carried out in methanol with a catalytic amount of NaOH. First, BNH3 was

dissolved in methanol and reacted with NaOH. After 30 minutes the aldehyde was added. After 12 hours CsF in methanol was added. The obtained compound was analyzed. Mass spectra showed only BNH3 as a cesium salt. The starting material was obtained in quantitative amount.

(29)

NH3 -TMA+ H O NHCOCH3 CO2CH3 5% NaOH 3 24 methanol HN -TMA+ NHCOCH3 CO2CH3

Scheme 18. Trial to receive Schiff base.

The reason we see in aldol condensation although the amount of NaOH is very small. In the first step, the base (hydroxide ion) abstracts a proton from the α carbon of one molecule of acetaldehyde to give a resonance-stabilized enolate ion.

H HC CH R O HO CH O HC CH O HC Enolate ion R R HOH R=CH(CO2CH3)(NHCOCH3)

Scheme 19. Mechanism of Aldol addition (step 1).

In the second step the enolate ion acts as a nucleophile – as a carbanion – and attacks the carbonyl carbon atom of a second molecule of the aldehyde. This step gives an alkoxide ion.

R H2 C CH O CH O HC R R H2 C C H O H C CH O An alkoxide ion R CH O HC R

In the third step, the alkoxide ion abstracts a proton from water to form an aldol. This step takes place because the alkoxide ion is a stronger base than the hydroxide ion.

(30)

R H2 C C H O H C CH O R R H2 C C H OH H C CH O R HOH _OH

Stronger base Weaker_base

A reaction of formation of the Schiff base between BNH3 and an aliphatic aldehyde has never

been reported. Only the reaction between the BNH3 and the aromatic aldehydes is known

(Sivaev 1999).

Alkyl sulfonates are frequently used as substrates for nucleophilic substitution reactions because sulfonate ions are excellent leaving groups. With this approach we decided to use tosylated serine (Theodoropoulos 1967) for reaction with CE-BSH (Scheme 22). As was shown before, tosylates undergo nucleophilic reaction such as SN2 to give product 26. After removing the

cyanoethyl group and the protection groups on the carboxylic and the amino groups we received boronated analogue of cysteine 28.

S NC -2TMA+ TMA+ S 2-CN _O N O O O O S O O N O O O O S 2TMA+ N O O O O 2-S 2Ph4P+ NH2 OH O 2-1) 6M HCl, NaOH 2) Ph4PBr N(CH3)4OH 5 ₂₅ 26 27 28

Scheme 22. Synthesis of boronated analogue of cysteine.

The solubility of the newly synthesized amino acids depends of the cation. The tetraphenylphosphonium cation makes the amino acids soluble in methanol and acetonitrile, the tetrabutylammonium cation – in dichloromethane and water; tetramethylammonium salts of amino acids dissolve in water. It is interesting to notice that other derivatives of [B12H12]2- with

(31)

3.2. Synthesis of polymers by ring-opening polymerization of α-amino

acid N-carboxy-anhydrides

In any consideration of methods that may be employed for the union of two amino acids via a peptide linkage, the direct condensation of the amino group of the amino acid and the α-carboxyl group of the other, with the elimination of a single molecule of water, is most readily and simply visualized. Such an idealized synthesis may be represented by:

NH2CHRCO2H NH2CHR'CO2H NH2CHRCO NHCHR'CO2H H2O

Achievement of peptide bond synthesis through direct condensation, as formulated above, is difficult. However, this is difficult not only because of dipolar nature of amino acids, but also because of energy considerations as well.

In 1906, Hermann Leuchs, a student of Emil Fisher, discovered the class of N-carboxy-amino acid-anhydrides, also known as Leuchs’ anhydrides, or abbreviated NCAs. These compounds are, even 100 years after their discovery, valuable intermediates in organic synthesis due to their reactivity: NCAs polymerize with the elimination of carbon dioxide to yield polypeptides, model compounds for proteins. They can be used as starting material for a variety of pharmaceutical interesting products and for synthetic polymers (fibers and films).

Two basically different approaches exist for the synthesis of N-unsubstituted α-amino acid NCAs (Kricheldorf 1987):

1. Cyclization of N-alkoxycarbonyl-amino acid halogenides, the so-called “Leuchs Method”.

2. Phosgenation of free α-amino acids or suitable derivatives, the so-called “Fuchs-Farthing Method”.

The most widely used method for the preparation of NCAs is the phosgenation of free amino acids. This method was first applied by Fuchs and was later elaborated by Farthing for a broad variety of NCAs. The first step of the phosgenation seems to be the formation of an N-chloroformyl-amino acid (Scheme 23), because addition of aniline yields 5-phenyl-hydantoic acids.

(32)

OH O NH2 R H Cl O Cl O N O O H R H OH O NHCOCl R H -HCl -HCl +Aniline OH O NHCONHC6H5 R H

Scheme 23. Synthesis of N-carboxy-anhydrides with phosgene.

α-Amino acid NCAs have four reactive sites – two electrophilic groups, CO-2 and CO-5, and two nucleophilic groups, NH and C-H after deprotonation.

The initiating of ring-opening polymerization can occur from action of different nucleophiles, which can be subdivided into two classes: protic nucleophiles and bases on the one hand and aprotic nucleophiles or bases on the other hand. Furthermore, it is useful to classify all initiators according to the following criteria (Kricheldorf 1990):

1. The reaction site of the NCA attacked by the initiator.

2. Reactivity of the initiator relative to that of the active chain end.

3. Incorporation of initiators into the peptide chain forming a dead chain end via a covalent bond.

Here, the two most important mechanisms will be presented. The detailed description of mechanisms with different initiators is found in (Kricheldorf 1990).

Initiation with primary amines

Primary amines are the most widely used initiators of all protic nucleophiles. They exclusively attack the carbonyl group C-5 of the NCAs. Because aliphatic primary amines are basic enough to stabilize the initially formed carbamic acid by deprotonation the chain growth can proceed in two ways (Figure 24). First, the active chain end is an amino group: a mechanistic pathway which is called “amine mechanism” or “normal propagation”. Second, the active chain end can be a carbamate ion: the “carbamate mechanism”. Because both protonation of carbamate ions and decarboxylation of carbamic acids are reversible reactions, it will be largely depend on the reaction conditions, in particular on temperature, solvent, CO2 pressure. Experimental

(33)

O N O O R R'-NH2 nucleophilic attack H N HO O N H O R' R N O O O R H H _H N O O N H R' O R N O O O R H H2N R N H O R' N H O O O O H N O N H O R' R R +H+ -CO2 N H HO O H N O N H O R' R R -H+ -CO2 H2N N H O R' R + n-2 NCA - n-1 CO2 n "Carbamate-Mechanism" "Amine-Mechanism"

Scheme 24. Mechanism of polymerization of N-carboxyanhydrid by initiation with primary amines.

Primary alkylamines are more nucleophilic than active chain ends, so, initiation is more rapid than propagation. Therefore, polymers with high molecular weights (average degree of polymerization DP>150) are in general not possible.

Initiation with tertiary amines

Tertiary amines develop a new mechanistic concept – the activated monomer mechanism AMM: instead of nucleophilic attack the deprotonation of unsubstituted NH group takes place. This means that the initiator reacts exclusively as base and not as nucleophile (Figure 25). The formed

(34)

anion then attacks the next N-carboxyanhydride at CO-5 and forms a dimer. This dimer can be attacked by the next deprotonated N-carboxyanhydride, and so on. As a result of these condensations, a polymer with high molecular weight can be form. The initiator is not incorporated to the polymer chain, but acts only as a catalyst in protonation/deprotonation equilibrium. Therefore, the control of molecular weight or molecular weight distribution is not possible. R'3N O N O O R H O N O O R _R' 3NH+ O N O O R H O N O O R O NH O O R R'3NH+ +NCA H+ Transfer -CO2 O N O O R _R' 3NH+ O N O O R O NH2 R N-Aminoacyl-NCA + n-1 NCA - n-1 CO2 O N O O R _O H N O NH2 R R n

Scheme 25. Mechanism of polymerization of N-carboxyanhydrid by initiation with tertiary amines.

Both described mechanisms are always in competition to each other and give in each case side reactions. Exchanging between amine- and AM- mechanisms during polymerization reaction is also possible.

For synthesis of boron containing polymers

S-(5-amino-5-carboxylpentyl)-thio-undecahydro-closo-dodecaborate(2-) bis-tetraphenylphosphonium salt 9 was chosen. To receive Leuchs’

anhydride we used “Fuchs-Farthing Method” with phosgene. We showed that gaseous phosgene can be used as well as phosgene dissolved in toluene as 20% solution (Scheme 26). Both ways gave the product 29 with high yield (90-95%).

(35)

S (CH2)4 _COOH NH2 2Ph4P+ 2-O Cl Cl 20% in toluene S (CH2)4 2Ph4P+ 2-HN O O O O Cl Cl gas tetrahydrofuran acetonitrile 9 29

Scheme 26. Synthesis of Leuchs’ anhydride 29.

As initiators for polymerization n-butylamine (the more frequently used for polymerization of natural amino acids) and the fluorescent dansyl amine (see paragraph 2.2) were chosen. The polymerization reaction was carried out with n-butylamine in dry acetonitrile. The molar ratio 10:1 (NCAs:n-butylamine) was used. The reaction mixture was stirred at room temperature for 4 days. After elimination of the solvent in vacuum, the residue was analyzed.

In the positive mode of elecrospray ionization mass spectrum we observed only one peak, which corresponds to the tetraphenylphosphonium cation. In the negative mode – a series of peaks, but no one belongs to the structure of the desired polymer (Scheme 27). The negative mode of MALDI-TOF showed the peak of highest mass at m/z 1860 amu, but it was difficult to determine whether this peak has a boron isotopic pattern. Since each monomer of the synthesized polymer contains a anion with charge “2-”, in the case of successful polymerisation we should receive a very polar compound. In the literature we did not find any examples of analysis of such kind of systems with ESI or MALDI mass spectrometry.

HN N H O S B12H11 2-2Ph4P+ 29 H2N H n n=10

Scheme 27. Polymerization with n-butylamine.

At the same time polymerization with the fluorescent dansyl amine as initiator was made. This time polymerization was carried out with the simplest combination of molar ratio 2:1 (NCA:initiator) (Figure 28).

(36)

H2N H N S O O N H2N N H O H N O S S N H S O O B12H11 2-B12H11 2-N 2Ph4P+ 2Ph4P+ 30 29

Scheme 28. Polymerization with dansyl amine.

From the mass spectrum of synthesized material we did not determine any peaks for the desired dimer, but comparing the negative spectra of ESI with the previous one we found similarities of most peaks present. This fact gave to us the idea that polymerization, probably, goes without interaction of primary amines and the initiator “plays role” only as catalyst. Then polymerization goes via the AM mechanism. The reason of this could be the sterical hindrance: the dodecaborate cluster with two tetraphenylphosphonium cations around prevents nucleophilic attack of the primary amine and, therefore, the last one deprotonate unsubstituted NH group of the anhydride 29.

Hydrogen

Boron Sulfur

Carbon Oxygen Nitrogen

Figure 12. 3D structure of the anhydride 29 (structure shown without tetraphenylphosphonium cation).

(37)

In the Table 1 we present common peaks in the negative MS of synthesized materials. ESI-MS spectrum and the calculated isotopic pattern of probable anions are shown in Appendix 2.

Common peaks of negative ESI-MS (m/z)

Probable structure of the anion

R=-(CH2)4-S-B12H11 2-100 m/2 ? 141 m/1 [B12H11] -(see paragraph 3.3) 157 m/1 [B12H11]--H++OH- 180 m/2 ? 420 m/2 ? 668 m/1 HN O O O R Ph4P+ -684 m/1 O N R O O O R NH2 3*Na+ -HN N R O O COOH R 3*Na+ -699 m/1 O N R O O O R NH2 2*Na+ -K+ HN N R O O COOH R 2*Na+ -K+

(38)

1181 m/1 O N R O O O R N H O R NH2 -H+ 4*K+ 2*Cu+

-Table 1. Common peaks in ESI-MS of synthesized polymers.

The presumable structures given in Table 1 are based on the known capacity of amino acids to form metal complexes (e.g. biuret reaction).

IR spectra of the synthesized materials did not show signals in the area of 1650-1690 cm-1_{, which}

correspond to the peptide bond. Comparison of IR spectra of synthesized materials and Leuchs’ anhydride (Figure 13) made it clear that signals at 1846 and 1780 cm-1 (2-CO and 5-CO of 2,5-oxazolidinedione cycle) did not disappear. These facts confirm the presence of an anhydride group in the synthesized products.

wavenumber (cm-1) 500 1000 1500 2000 2500 3000 3500 4000 tr ans miss ion ( % ) Amino acid 9 Leuchs' anhydride 29

Polymerization with n-butylamine Polymerization with dansyl amine

(39)

Gel filtration chromatography is a separation based on size. In gel filtration chromatography, the stationary phase consists of porous beads with a well-defined range of pore sizes. The stationary phase for gel filtration is said to have a fractionation range, meaning that molecules within that molecular weight range can be separated. Molecules that are small enough can fit inside all the pores in the beads and are said to be included. These small molecules have access to the mobile phase inside the beads as well as the mobile phase between beads and elute last in a gel filtration separation. Molecules that are too large to fit inside any of the pores are said to be excluded. They have access only to the mobile phase between the beads and, therefore, elute first. Molecules of intermediate size are partially included - meaning they can fit inside some but not all of the pores in the beads. These molecules will then elute between the large ("excluded") and small ("totally included") molecules.

For the gel filtration experiment the sample of polymerization with dansyl amine was used. The sample was dissolved in acetonitrile and injected to the Phenomenex Phenogel 5µ 10*3A column. Since dansyl amine was used as the initiator of polymerization, detection of UV chromatogram signals at 337 nm was done. In UV chromatogram (Figure 14, curve B) several peaks are present. The first peak –peak of the biggest size molecule in mixture comes at 10.0-12.0 minutes and the last one – peak of the smallest size molecule (probably the starters) comes at 14.5 minutes. 2712DG02.D: TIC ±All 2712DG02.D: UV Chromatogram, 337 nm 2712DG02.D: EIC 1180 ±All 2712DG02.D: EIC 1090 ±All 0.5 1.0 5 x10 Intens. 0 20 40 60 0 2000 4000 6000 8000 0 500 1000 1500 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 Time [min] A B C D

Figure 14. UV chromatogram (B) of gel filtration, total ion current chromatogram (A) and mass traces of peaks m/z 1180 (C) and m/z 1090 (D).

Mass spectrum on Figure 15 shows only a fragment peak at m/z 141, which belongs to the [B12H11]- ion and a peak at m/z 1181. The peak at m/z 1181 could belong to the structure

(40)

determined above (see Table 1), or to a fragment peak of polymer. But the absence of peptide bond signal in IR spectra does not confirm the suggestion of presence big molecules (polypeptides) in the synthesized mixture.

Gel chromatography was done also with the same sample with detection at 220 nm. This experiment did not clarify the way of polymerization and information about synthesized products. The conditions of gel chromatography and MS spectrum of measurements at 220 nm are present in Appendix 2.

In conclusion of this chapter we can say that dodecaborate-containing amino acids are able to form N-carboxy anhydrides, but polymerization with primary amines goes via the AM mechanism. Since the control of molecular weight or molecular weight distribution is not possible, we tried to determine this with available technique (ESI-MS, MALDI-TOF, gel chromatography). It is difficult to make final conclusion since presumable anions from ESI-MS spectrum could be also a fragment ions of big size molecule.

141.1 317.0 699.4 839.5 1084.8 1180.8 1366.5 1509.3 1799.0 All, 10.3-11.6min (#388-#439) 141.1 299.7 479.9 575.5 940.6 1181.01280.9 All, 8.2-8.9min (#308-#336) 141.1 252.1 435.6 629.1 778.4 898.6 1090.6 1328.9 1459.5 _{1740.6 1848.1} All, 12.0-12.5min (#452-#471) 140.1 195.1 435.1514.2594.4 689.5 954.4 1089.6 _1361.5 _1508.6 1931.1 All, 12.9-13.4min (#485-#507) 0 200 400 600 800 Intens. 0 500 1000 1500 0 100 200 300 0 100 200 250 500 750 1000 1250 1500 1750 2000 m/z

Figure 15. Mass spectra (ESI-MS, negative mode) of separated by gel filtration products (time of separation is indicated in the right corner of the spectra).

(41)

3.3. Mass spectrometric behavior of B

12

-cluster-containing compounds

All synthesized compounds were characterized by electrospray mass spectrometry. In this chapter we report about a new synthesis (Scheme 29) of a polyhedral borane carboxylic acid 32 and its mass spectrometric behavior. We found that this kind of behavior is characteristic for all B12H11R cage-containing compounds.

Boron-containing carboxylic acids are interesting objects for BNCT - they can form stable bonds with amino groups in the target molecules. The synthesis of 32 proceeded as expected from similar compounds prepared in the past (Gabel 1994).

In the course of the characterization of B12-cluster-containing compounds, we found in

electrospray ionization mass spectrometry (ESI-MS) signals which did not correspond to the signals expected from the presumed product, but which were first interpreted as signs of an incomplete reaction, or a degradation of the compound upon storage. We investigated this phenomenon and found that this was not the case, but that ESI MS induced fragmentation of the compound. The origin of the fragments and fragmentation pathways could be clarified.

2 -2 -(CH3)4NOH _S _O OH CN S Br OH O CN S O OH 31 32 5 2TMA+ TMA+ 2TMA+

Scheme 29. Synthesis of (3-carboxyethyl)-thio-undecahydro-closo-dodecaborate(2-).

Electrospray MS

To induce fragmentation of the sample ions, collision-induced dissociation (CID) in the nozzle-skimmer region ("nozzle-skimmer-CID" or sCID) as well as tandem-MS was used (Figure 16).

(42)

Nozzle ~150 V

Skimmer ~100 V

Ion-trap MS ~1 mbar ~10-3 _mbar _~10-5 _mbar

Metal capillary, ID 100-200 µm ±2-5 kV ~0 V Skimmer-CID region MS/MS region Sample solution Nozzle ~150 V Skimmer ~100 V Ion-trap MS ~1 mbar ~10-3 _mbar _~10-5 _mbar

Metal capillary, ID 100-200 µm ±2-5 kV ~0 V Skimmer-CID region MS/MS region Sample solution

Figure 16. Schematic diagram of an ESI mass spectrometer. Fragmentation can occur in the skimmer-CID region, or in the MS/MS region.

In sCID, fragmentation is induced by colliding the sample ions with the background gas in the intermediate-pressure region (the so-called "nozzle-skimmer region", gas pressure about 10-1 mbar) of the ESI interface. Increasing the voltage difference between the nozzle and the skimmer (VsCID) raises the kinetic energy of the ions passing through this region, hereby increasing the

collisional energy of the ions with the background gas (consisting mainly of nitrogen and of some residual solvent vapor) and eventually causing fragmentation. The VsCID applied in this

investigation were between 75 and 100 V and are indicated in the mass spectra shown in Figure 17. 87.1 123.0 141.1 0 2 4 6 8 4 x10 Intens. 60 80 100 120 140 160 180 m/z V(sCID)= 96 V V(sCID)= 76 V V(sCID)= 85 V

Figure 17. ESI-MS of 32, using increasing sCID voltage.

In contrast, tandem-MS (or MS/MS) takes places in the mass analyzer, i.e. in the high-vacuum region of the MS (pressure about 10-5 mbar). In MS/MS, the ion of interest is isolated within the

(43)

ion trap by ejecting all other ions out of the trap. Then this ion is accelerated (by applying a suitable high-frequency AC voltage) and collides with a collision gas (usually He) present at a pressure of 5⋅10-6_{mbar. The fragment ions thus generated are then detected by a normal mass}

scan. The degree of fragmentation depends on the AC voltage amplitude; in this investigation, this amplitude was adjusted empirically so that a low intensity signal of the parent ion remained visible in the fragment ion spectrum. Alternatively, the residence time of the ion was increased.

Generally, ESI-MS is considered to be the softest of all known MS ionization methods; consequently, fragmentation of sample ions is usually not observed unless it is deliberately induced by either sCID or MS/MS. Hence, the negative-ion ESI mass spectrum of 32 was expected to show only the signal of the doubly charged molecular anion at m/z 123 (2-). However, the spectrum (Figure 18) shows two additional peaks: a doubly charged ion at m/z 87 (2-), which was attributed to [B12H11SH]2-, and a singly charged ion at m/z 141 (1-) whose mass

and isotopic pattern suggest the molecular formula [B12H11]-.

87.1 123.1 141.2 0.0 0.5 1.0 1.5 2.0 4 x10 Intens. 60 80 100 120 140 160 180 m/z [B12H11] -[B12H11S(CH2)2COOH] 2-[B12H11SH] 2-Figure 18. ESI-MS of 32.

Thus, the mass spectrum seemed to indicate a mixture of compounds rather than a pure compound, which, however, was not in accordance with results from HPLC, NMR (Figure 19) and IR. In order to investigate whether the additional ion signals originated from the MS measurement, a series of sCID experiments were carried out. The results are shown in Figure 17.

(44)

HPLC chromatogram of 3 (pp m)1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 (pp m)1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 (pp m)1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 (pp m)1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 2.0-(-0.4) (11H, B₁₂H₁₁) 3.1 (24H, N+_(CH 3)4) 2.36 (2H, -CH₂-) 2.07 (2H, -CH₂-)

Figure 19. NMR spectra of acid 32 (solvent DMSO-d6).

It is seen that with increasing VsCID the relative intensity of the ion signal at m/z 123 (2-)

decreases while the other two ion signals increase (since all spectra are normalized to the highest intensity peak, which is m/z 87 (2-) in all cases, the relative intensity increase of this ion is not directly visible). Since generally upon increasing the fragmentation energy the intensity of a parent ion is expected to decrease while that of the fragments should increase, this finding clearly shows that the ion signals at m/z 87 (2-) and 141 (1-) must be fragments of the molecular ion at m/z 123 (2-).

We have observed the signal at m/z 141 (1-) rather frequently in ESI mass spectra of substituted B12-clusters; its origin had not yet been investigated. In the case of the molecule considered here,

the mass spectrum shows two fragment peaks, which raises the question of the fragmentation pathway. With sCID this cannot be clarified because all ions present in the nozzle-skimmer region at a given time might be fragmented. Hence it cannot be distinguished whether the primary fragmentation product of the molecular ion is m/z 87 (2-) or whether m/z 141 (1-) arises independently. We therefore performed an MS/MS experiment to obtain further information on this issue. In contrast to sCID, MS/MS is selective because the ion to be investigated is isolated before fragmentation. The negative-ion ESI-MS/MS mass spectrum of m/z 123 (2-) is shown in Figure 20.

(45)

Figure 20. MS/MS of peak 123.

The spectrum shows that the main fragmentation product is m/z 87 (2-), i.e. the [B12H11SH]2- ion.

This fragmentation reaction can be easily explained by abstraction of acrylic acid (Scheme 30). Upon protonation, the resulting [B12H11SH]2- ion can lose H2S, to give m/z 141 (1-).

S OH O SH 2 -m/z=123.1 z=-2 m/z=87.1 z=-2 m/z=141 z=-1 + OH + O H2S H+

Scheme 30. Generation of fragment ion of 32 in ESI-MS.

This is also in accordance with the empirical rule that mass spectrometric fragmentation of even-electron ions occurs preferentially by abstraction of small neutral molecules; frequently, the driving force for the reaction is the stability of the neutral molecule. In this case, both the resulting fragment ion and the neutral acrylic acid are stable particles. For [B12H11SH]2-, H2S as

the stable neutral fragment is generated by protonation of the sulfur.

The obvious conclusion from this result is that the ion at m/z 141 (1-) must be a secondary fragmentation product of m/z 123 (2-), i.e. it must originate from the [B12H11SH]2- ion (m/z 87

(2-)). To test this, we subjected the ion at m/z 87 (2-) from the MS/MS spectrum shown in Figure 20 again to MS/MS conditions, i.e., we performed an MS3 experiment. The resulting spectrum showed a weak signal at m/z 141 (1-), which was found to increase with increasing residence time of the m/z 87 (2-) ion in the ion trap. Thus, the m/z 141 (1-) signal does not only stem from

(46)

the sCID region, but can also be formed under MS/MS conditions. The same signal was observed with an MS/MS spectrum of [B12H11SH]2- (Figure 21).

86.9 141.0 0 200 400 600 800 1000 Intens. 60 80 100 120 140 160 180m/z 86.9 141.0 0 200 400 600 800 1000 Intens. 60 80 100 120 140 160 180m/z Figure 21. MS/MS of [B12H11SH]2-.

Obviously, the formation of m/z 141 (1-) requires the presence of the background gas which is present in the nozzle-skimmer region, but whose concentration is low in the high vacuum of the ion trap. In the skimmer region, this background gas, although it consists mainly of nitrogen, also contains residual solvent vapor as source for protons; we can tentatively explain this fragmentation reaction to occur via a gas-phase proton attachment to the doubly charged BSH anion, followed by abstraction of H2S (see Scheme 30). The same explanation might hold for the

high vacuum conditions in the ion trap; here, due to the low background (and hence also solvent) gas pressure, longer residence times of the precursor ions are needed to form detectable amounts of the product ion.

From the protonated form of B12H12(2-) ion, H2 abstraction has been found to be energetically

possible and leads to a rather stable B12H11(1-) anion. The energy required has been calculated

from only 3 kcal/mol (Mebel 1989) to 7.5-11 kcal/mol (Mebel 1999) (Scheme 31).

B12H11 -B12H13+

+3-11 kcal/mol -H2

Scheme 31.

The MS/MS spectrum of [B12H11]- showed no fragment ions, but instead a series of peaks at

higher masses with mass differences of 16 amu. To clarify the processes occurring here we chose to repeat the MS/MS experiment with monoisotopic exitation, i. e. only the highest intensity peak of the B12H11 isotopic distribution was selected and fragmented. The resulting spectrum is

(47)

shown in Figure 22. Two series of peaks can be distinguished, which can be attributed to two different processes:

• The first process (red line) starts from [B12H11]- at m/z 141 to give the peak m/z 157. The

difference in 16 amu gives us the idea about attachment of H2O and abstraction of H2.

Thus, the peak at m/z 157 should correspond to the structure [B12H10OH]-.

Hydroxylation can be observed until 7 hydroxyl substituted ions (Figure 22) and finishes with [B12H4(OH)7]- (peak m/z 253).

H O 141.2 157.2 173.2 189.2 205.2 221.2 255.2 271.1 287.2 303.2 237.2 0 200 400 600 800 Intens. 100 120 140 160 180 200 220 240 260 280 300m/z 253.2 +H2O 141 157 [B12H11]- [B12H10OH] --H2 m=16 +H2O -H2 m=16 [B12H9(OH)2] -173 ...

Figure 22. MS/MS of 141. Process 1 (hydroxylation).

In Figure 23 we present the presumable mechanism of hydroxylation with abstraction of one molecule of hydrogen. B B B H H H O H H B B B OH H -H₂ H

(48)

In collaboration with Dr. Matthias Hofmann from Anorganisch-Chemisches Institut Heidelberg, theoretical calculations were done. The calculations showed that hydroxylation of [B12H11]- unit

occurs with substitution of hydrogen preferentially in position 7, 8 and 10 (Figure 24).

Figure 24. Hydroxylation of [B12H11]-.

• The second process (Figure 25, blue line) is first observed with the fourfold hydroxyl-substituted ion [B12H9(OH)4]- (peak m/z 207). The origin of this peak can be explained

by attachment of H2 (probably from the previous process) to the [B12H7(OH)4]- ion.

Further hydroxylation proceeds analogously to the first process, finally leading to [B12H3(OH)10]- (peak m/z 303). 141.2 157.2 173.2 189.2 205.2 239.2 255.2 271.1 287.2 303.2 223.2 207.2 0 200 400 600 800 Intens. 100 120 140 160 180 200 220 240 260 280 300m/z [B₁₂H₇(OH)₄] -m=2 +H2O -H2 m=16 [B12H9(OH)4]- ... +H2 205 207 [B₁₂H₈(OH)₅] -223 +H2O -H2 m=16

Figure 25. MS/MS of 141. Process 2 (attachment of H2 and subsequent hydroxylation). We suppose that the presumable mechanism of the second process can occur with forming two

Synthesis of boron-cluster containing amino acids and preparation of their polymers