The first solid phase experiment and resulting optimisation of cleavage conditions. of cleavage conditions

In order to assess the suitability of 11 for incorporation into polyamide structures through the application of solid phase synthesis, a simple target compound was envisioned. This consisted of the alkylation subunit of duocarmycin sandwiched between two alanine residues (32, scheme 3.2). Standard conditions used routinely in our lab for peptide synthesis were planned to prepare it (scheme 3.2). This preliminary work was undertaking prior to completion of the scale up process with racemic 11 accessed from the pilot synthesis.

The synthesis began from a commercially available preloaded alanine Wang resin. As such, the manufacturer’s specification provided a substitution level with respect to the Scheme 3.2 First solid phase experiment: synthesis of 32. Red structure = structure of the Wang

linker.

116 quantity of alanine loaded. This allowed for fine control of the stoichiometric ratios of reagents used for the following couplings.

The term Wang, refers to the structure of the linker, which anchors the alanine residue to the polystyrene based resin.¹⁵¹ This is a very common linker for Fmoc solid phase peptide synthesis. Its structure is depicted in red in scheme 3.2. The amino acid is attached via a benzyl ester. However, unlike that used in the Merrifield approach, which requires treatment with hydrofluoric acid to affect cleavage, the structure of the Wang linker, allows cleavage by treatment with TFA. The increased acid sensitivity of the benzyl ester of the Wang linker is the result of the para benzyl ether. This provides mesomeric electron donation which stabilises the carbocation resulting from cleavage.

When conducting any solid phase synthesis. It is imperative that the resin is appropriately swelled to allow efficient access to the individual polymer chains. This was achieved by solvating the resin in DCM for 30 mins, followed by DMF for a further 30 mins.

As discussed one of the advantages of solid phase synthesis is the removal of by- products and excess reagents by washing of the resin. This occurs between every step. In the proceeding discussions, this process will not be highlighted, and unless otherwise stated the resin was washed between treatments with an appropriate volume of DMF at least 6 times.

The commercially available resin was supplied with Fmoc-protection of the loaded alanine. This was removed by treatment with 40 % piperidine in DMF for 10 min, followed by 20 % piperidine in DMF for 5 min twice. Separate treatments help to ensure full deprotection, but has not been proven to be necessary. The mechanism of Fmoc- deprotection by piperidine has already been discussed in section 3.1.

Next would follow the first coupling of 11, and to the best of our knowledge the first solid phase amide coupling of a duocarmycin structure. Amide bond formation is typically promoted by formation of an activated ester of the carboxyl group. Many coupling reagents have been developed. Uronium salts are arguably amongst the most popular choice for solid phase couplings, of which HBTU is a typical example. It is one of the older uronium salts, and more efficient alternatives exist. However, it is relatively cheap when compared to the newer generations of uronium salts, and thus is still commonly used as cost effective option. Indeed, HBTU was the standard coupling reagent used in our lab for solid phase peptide synthesis, and as such was considered first for the application of 11 to the solid phase methodology.

117 The term uronium salt is misleading. This refers to the structure proposed when the reagent was first introduced (33, scheme 3.3).¹⁵² However, more recent evidence suggests it is more commonly found as the guanidinium isomer (34, scheme 3.3).¹⁵³ The likely mechanism of amide bond formation using HBTU is depicted in scheme 3.3. A tertiary amine base, such as DIPEA is used to deprotonate the carboxylic acid. The resulting carboxylate anion attacks the electrophilic sp² hybridised carbon of the guanidinium structure. Subsequent collapse of the tetrahedral intermediate, results in the formation of the isouronium cation, and the oxybenzotriazole anion. These two species then react to form oxybenzotriazole ester, with the release of tetramethylurea. Attack of this activated ester by the amine gives the desired amide bond after deprotonation.

Uronium salts are used in an equimolar quantity to the carboxylic acid, and are typically premixed before addition to the amine. This is to reduce the occurrence of competing guanylation of the amine, which can lead to capping of the resin (scheme 3.4).¹⁴⁵ When using HBTU, it is also common to also use HOBt as an additive. This increases the

Scheme 3.3 Mechanism of amide bond formation using HBTU. Also shown are the different isomers of HBTU 33 and 34.

118 efficiency of the oxybenzotriazole ester formation. When coupling amino acids, this is also said to act to supress potential racemization. As the isouronium ester, N-protected amino acids can be particularly prone to base-induced racemization through either enolization or oxazolone formation (Scheme 3.5).¹⁵⁴ This is due to the increased acidity of the alpha proton.

For coupling of 11, this is not a specific concern, but there is still the potential to benefit from faster formation of the oxybenzotriazole ester with the use of HOBt as an additive.

Solid phase couplings are usually performed with a large access of the active ester, in order to drive reactions to completion. Near quantitative conversions are important to prevent the accumulation of resin bound deletion products. However, given the precious nature of 11, it was decided that couplings should be first attempted with the smallest excess possible.

Therefore, 1.1 equiv. of 11 was treated with an equimolar quantity of HBTU, and a twofold excess of both HOBt and DIPEA, in DMF. After 30 seconds the mixture was added to the resin. The resin was shaken for 2 hours. At this point a sample of the resin was taken, and subjected to Kaiser testing.

Scheme 3.4 Potential capping of resin by guanylation.

119 The Kaiser test is a qualitative test used to detect the presence of primary amines.¹⁵⁵ A small sample of the washed resin is treated with a drop of three solutions. These are, 5%

w/v ninhydrin in ethanol, 80 % w/v phenol in ethanol, and a 2:98 mixture of aqueous 0.001 M KCN and neat pyridine. The sample is heated to 120 ^oC for 5 mins. Formation of a deep blue colour indicates the presence of primary amines.

The Kaiser test is based on the reaction between ninhydrin and primary amines, which results in the formation of the chromophoric compound ‘Ruhemann's Purple’. The mechanism of this reaction has been the subject of some controversy. However it is now

Scheme 3.5 Potential base catalysed racemization, of the isouronium ester form of Fmoc-protected amino acids, by enolization or oxazolone formation during couplings.

120 generally accepted that the most likely mechanism (scheme 3.6),¹⁵⁶ involves the dehydration of ninhydrin to give 1,2,3-indantrione. This condenses with the amine to give the Schiff base. Decomposition gives an intermediary amine derivative of ninhydrin, which condenses with another molecule of 1,2,3-indantrione to give ‘Ruhemann's Purple’.

Unexpectedly, the Kaiser test performed after 2 hours, was negative, suggesting complete coupling of 11, despite the modest excess of the reagent used. This was a surprise, as it was anticipated that 1.1 equiv. of 11, would not be sufficient to affect a quantitative coupling. The plan in the event of a positive Kaiser test would have been to perform additional 2 hour couplings with 0.5 equiv. of 11 until a negative Kaiser test was observed.

This was deemed to be a good reagent conserving strategy for driving couplings to completion. Fortunately this did not appear to be necessary.

Scheme 3.6 Mechanism of the Kaiser test.

121 Subsequently the indoline nitrogen was Fmoc-deprotected with piperidine as before, and coupled to a final alanine residue. This was achieved using Fmoc-Ala-OH (5 equiv.), HBTU (5 equiv.), and DIPEA (10 equiv.), in DMF. As before the reagents were premixed for 30 secs prior to addition to the resin. Since a large excess of amino acid was used a more standard 45 min reaction time was employed. Kasier testing was not employed here, as the formation of ‘Ruhemann's Purple’, does not occur between the reaction of ninhydrin and secondary amines. Other resin tests exist which are capable of detecting free secondary amines. These are typically used to monitor couplings of proline. A good example is the chloranil test.¹⁵⁷ However this was not employed here due to the lack of reagent availability. Therefore, it was decided to assume that the single coupling with a large excess of amino acid was likely sufficient.

After removal of the N-terminal Fmoc group of the alanine residue, the resin was prepared for cleavage by extensive washing with DMF, followed by DCM, and drying under a stream of nitrogen. Cleavage was affected by treatment of the resin with a solution of 95

% TFA, 2.5 % TIPS, and 2.5 % water. After shaking for 2 hours, the cleavage mixture was filtered and concentrated by rotary evaporation, followed by precipitation with the addition of cold Et₂O.

The crude product was analysed by reverse phase analytical HPLC. Initial results were disappointing. Although an obvious product dominated the HPLC trace, with a strong peak at 9.0 min, several significant side products were also observed (figure 3.1a). It seemed unlikely that these could represent deletion products. Firstly there were more side products than could be predicted to have resulted from incomplete couplings.

Furthermore, Kaiser testing had suggested that the first coupling was quantitative.

Incomplete Fmoc-deprotection also has the potential to result in deletion products, however considering the reliability of this step, it too seemed unlikely to explain the observed impurities.

HPLC analysis was monitored by UV absorbance at the 254 nm, and 214 nm wavelengths. The side products showed greater intensity at 254 nm, and this strongly suggested that they contained aromatic character derived from the duocarmycin residue.

It was possible that the large excess of activated alanine was able to react, not only with the indoline nitrogen, but also lead to acylation of nucleophilic sites on the indole scaffold of the duocarmycin residue. However, this seemed unlikely with activation of the alanine by HBTU. Even if an acid chloride of alanine had been used, such a reaction would be likely to require Friedel-Craft like conditions.¹⁵⁸ Furthermore, the indole side chain of tryptophan is not reported to be susceptible to acylation during peptide couplings.

122 As this synthesis had been conducted with racemic 11, it was briefly considered that perhaps the impurities were the result of separation of diastereomeric deletion products.

The pattern of impurities certainly could be described as two pairs either side of the main product. However, the main product would also be a mixture of diastereoismers, and no separation of the peak at 9 min had been observed.

These explanations seemed unlikely, and therefore attention turned to the cleavage conditions. The cleavage cocktail contained 95 % neat TFA, and it was possible that the duocarmycin residue was not stable under these conditions. Therefore, small sample of Figure 3.1 HPLC analysis of crude 32 after cleavage under varying conditions. a) 95 % TFA, 2.5 %

TIPS, 2.5 % H2O. b) 50 % TFA, 50 % DCM. c) 95 % TFA, 5 % DCM. d) 47.5 % TFA, 47.5 % DCM, 2.5 % TIPS, 2.5 % H2O. 10 mg of dried resin was cleaved under either conditions a, b ,c, or d, with

5 mL of the respective cleavage cocktail for 2 hours. The cleavage mixture was filtered and evaporated to dryness. The crude was dissolved in 1 mL of MeOH and analysed by HPLC at 254

nm. Agilent Eclipse XDB-C18 column, 4.8 x 150 mm, 5 µm. Solvent A: [Water and 0.05 % TFA], Solvent B: [ACN and 0.05 % TFA]. Gradient: 0% [B] to 95 % [B], from 0 min to 15 mins, 95 % [B] to 0 % [B] from 15 to 20 mins. Monitored UV 254 nm. Flow rate 1 mL/min. Column temperature 40 ^oC.

123 11 was treated with neat TFA. Monitoring by TLC showed decomposition of this reagent.

This suggested, that the unexpected impurities were the result of side reactions during scavengers (TIPS and H2O), saw a dramatic improvement (figure 3.1b). The side products that eluted before the main product at 9 min, which had been the most intense impurities under the original conditions, were no longer present. However the two impurities at 9.5 and 9.7 min appeared to have grown in intensity when compared to the main product at 9 min. It seemed that the inclusion of scavengers protected against the formation of these two impurities; but the question arose, was it the reduction in TFA concentration, or the omission of scavengers which had protected against formation of the other side products?

It was at least conceivable, that the combination of TFA as proton donor, and the trialkylsilane, as potential hydride donor, could lead to reduction of the indole scaffold to an indoline.¹⁵⁹ Indeed, the use of TES as a scavenger in the cleavage of tryptophan containing peptides is known to cause reduction of the indole side chain.¹⁶⁰ However, TIPS is cited as a good alternative when this problem is encountered, as it is significantly less prone to producing this side reaction.

In order to ascertain whether reducing the TFA concentration had derived a benefit. The next sample of resin was subjected to cleavage with 95 % TFA and 5 % DCM (figure 3.1c). Here, it can clearly been seen that the high concentration of TFA has had a detrimental effect. The main product peak previously observed at 9 mins, is no longer the majority product, in fact it is arguably no longer observed at all. Furthermore, the two impurities at 9.5 and 9.7 min now dominate the HPLC trace.

The logical extension of these findings was to combine a reduced concentration of TFA with the inclusion of the scavengers. Thus, the next sample of resin was cleaved using 47.5 % TFA, 47.5 % DCM, 2.5 % TIPS, and 2.5 % H₂O. This combination appeared to produce a synergistic benefit (figure 3.1d). Under these conditions the main product at 9 min, was the only discernible peak. This was confirmed as the desired product by accurate mass analysis.

Scavengers, are included in cleavage cocktails to protect potentially nucleophilic residues from alkylation by cationic species derived from cleavage. In the case of peptides containing trifunctional amino acids, this can include the tert-butyl cations derived from

124 simultaneous side chain deprotection. This is of course not applicable to the alanine residues present in 32. However, another source of carbocations is the linker itself.

Scheme 3.7 Acidic cleavage of 32 from Wang resin. Production of p-quinone methide, and resin bound cations. Potential alkylation of indole scaffold leading to soluble and permanently resin

bound impurities.

125 Scheme 3.7 depicts the acidic cleavage of 32 from Wang resin. The Wang linker is attached to the polystyrene resins via a benzyl ether, and is linked to the product by a benzyl ester. As discussed previously, it is the electron-donating effect of the para-benzyl ether, which increases the acid sensitively of the benzyl ester, and allows for complete cleavage by treatment with TFA. However, an undesired consequence is the potential for subsequent cleavage of the linker remnant itself. This liberates the highly reactive p-quinone methide species, which is then available to alkylate sensitive residues;¹⁶¹ in this case, the indole scaffold of the duocarmycin structure, by electrophilic aromatic substitution. This would result in soluble side products, and is likely responsible for the impurities seen at 9.5 and 9.7 min, which are reduced by the addition of scavengers.

It is likely that the side product at 9.5 min represents alkylation at the most nucleophilic 3 position of the indole, depicted in scheme 3.7, and the side product at 9.7 mins, represents the additional alkylation of this product at the 6 position (standard indole numbering) or at the indole nitrogen itself. Attack of the aromatic ring of the benzyl ether protecting the phenol group seems less likely. A second route to this impurity can also be envisioned, when p-quinone methide in not released. Here, the resin bound carbocation generated reacts first with the indole scaffold, and this product is then cleaved from the resin at the benzyl ether.

It can be seen in figure 3.1, that addition of TIPS and H₂O protects against the formation of these side products, as the intensity of their peaks grows with the omission of these scavengers (figure 3.1a vs figure 3.1b, and figure 3.1c). Here H2O acts as a completing nucleophile, and TIPS as a hydride donor. The synergy in protection observed between reduction in TFA concentration and the addition of scavengers (figure 3.1a vs figure 3.1d), most likely results from the reduced liberation of p-quinone methide. Thus, the smaller quantity of this soluble reactive species, is more effectively quenched by the concentration of scavengers. The resin bound-carbocation resulting from cleavage of the benzyl ester still exists in the same quantity. However, this species can also be quenched by the polystyrene resin itself, leading to increased crosslinking of the resin. This effect has been used to explain the reduced swelling capacity of Wang resins post TFA cleavage.¹⁶¹ As can be seen in scheme 3.7, liberation of p-quinone methide also produces a second carbocation on the polystyrene resin itself. This too can be envisaged to react with the sensitive indole scaffold. The consequence here would be a so called ‘back alkylation’, which permanently sequesters the product to the resin, reducing yields. Closer inspection of figure 3.1 provides some tentative evidence for this effect. The test cleavages were identical except for the composition of the cleavage cocktail, and were analysed by HPLC

126 after dissolution of the crude in the same volume of MeOH. Thus the intensity of the HPLC peaks gives a qualitative indication of recovery. Under the most destructive cleavage conditions of no scavengers and high TFA concentration, where back alkylation would be greatest (figure 3.1c), the highest intensity peaks are those of the soluble alkylated side products, at around 25 mAU. At reduced TFA concentration and with the presence of scavengers (figure 3.1d) the only discernible peak is the desired product with much greater intensity of around 140 mAU. This suggests a reduction in recovery of the product, which is consistent with loss due to back alkylation. Furthermore, comparison of figure 3.1a, and figure 3.1c, demonstrates that at high TFA concentration but with the presences of scavengers, which would protect against back alkylation, the desired product still dominates with an intensity of around 60 mAU.

Further support for these suspected side reactions during cleavage, can be found in the literature, concerning the cleavage of tryptophan containing peptides from Wang resin.

Side protects resulting from the alkylation of the indole side chain of tryptophan by liberated p-quinone methide, have been isolated and characterised.^{162, 163} Furthermore, Wang resin is reported to produce low yields of tryptophan containing peptides, even after soluble side products have been taken into account, and this has been attributed to losses incurred by back alkylation to the resin.^{162, 164} Considering that it is the more nucleophilic 3- position of the pyrrole ring of the duocarmycin indole which is unsubstituted, compared to the 2-position in the tryptophan indole, syntheses incorporating 11, may be particularly susceptible to such side reactions. Furthermore, the 6-position of the duocarmycin indole is further activated over the same position in the tryptophan indole, by the ortho-benzyl ether. Although this position is sterically shielded by the benzyl group.

It is reported that for tryptophan containing peptides such side reactions can be reduced by the use of Boc side chain protection.^{165, 166} This reduces the nucleophilicity of the indole. It is suggested that removal of this group during cleavage by TFA is retarded at the carbamic acid intermediate,^{166, 167} which is lost during work up, offering protection

It is reported that for tryptophan containing peptides such side reactions can be reduced by the use of Boc side chain protection.^{165, 166} This reduces the nucleophilicity of the indole. It is suggested that removal of this group during cleavage by TFA is retarded at the carbamic acid intermediate,^{166, 167} which is lost during work up, offering protection