Carbon Skeleton Synthesis

CARBON SKELETON SYNTHESIS

In general, it is most efficient to construct the carbon skeleton first and then adjust the functionality to give the target. Thus in retrosynthetic analysis we most often move backward from the target to compounds which contain functional groups important in carbon–carbon bond-making reactions. As a consequence carbonyl groups play a very important role in retrosynthetic analysis. They are very useful sources of both electrophilic and nucleophilic carbon that can be used in making carbon–carbon bonds. Based on our earlier discussions, a carbonyl group can be seen to influence the polarity of nearby carbons as shown.

By bond polarity and resonance, the carbonyl carbon and a carbon β to the carbonyl carbon can be utilized as electrophilic centers—the carbonyl group by direct nucleophilic addition and the β carbon by Michael addition to an α,β-unsaturated ketone. By resonance interaction, the α position in carbonyl compounds and γ positions in α,β-unsaturated carbonyl compounds can be con-verted to nucleophilic centers by proton removal. These “normal” polarities are used frequently in retrosynthetic planning as points of disconnection to establish potential bond-forming steps using carbonyl groups.

As an initial exercise consider the synthesis of C from cyclohexyl bromide. First one has to note relevant facts about the target, such as (a) it contains an ester and (b) it has two more carbons than the starting material so a two-carbon fragment will have to be attached by a new carbon–carbon bond.

Looking at the starting material, it is noted that it contains an electrophilic carbon; thus the needed carbon–carbon bond could be formed by reaction of a two-carbon nucleophile with the electrophilic center of bromocyclohexane. Noting that the ester group needed in the product acidifies the α position, it could be used to make the nucleophilic carbon required for carbon–carbon bond formation. Retrosynthetically this can be written as

So the synthesis could be done in one step by making the anion of methyl acetate and reacting it with bromocyclohexane. The polarities of the reaction partners match nicely, but the problem is that alkylations of secondary bromides with enolates often give poor yields. The enolate is a strong base, which pro-motes elimination in the secondary bromide rather than giving the substitution product needed in the synthesis. Thus elimination from cyclohexyl bromide to cyclohexene would be a major process if the reaction were attempted. While the retrosynthetic step seems reasonable, the synthetic step has known difficulties. It is important to work backward in the retrosynthetic analysis and then check each forward step for validity.

What is needed in the synthesis of C is a two-carbon nucleophile (or its equivalent) which is less basic than an enolate so elimination is not competitive. If product C is recognized as an acetic acid derivative, then the following analysis can be made. A malonate ion used as the carbon nucleophile is much less basic than a simple ester enolate and hence undergoes substitution readily but does not promote elimination effectively, particularly in secondary systems.

Now clearly there will have to be some functional group adjustment in the synthesis because a hydrolysis of the alkylated malonate must be carried out in order to give decarboxylation to the acetic ester derivative. Either an unsymmetric malonate must be used that can be differentially hydrolyzed or both ester functions of the malonate could be hydrolyzed and after decarboxylation the acid could be reesterified. The actual synthesis could be planned as shown in (10.1) or (10.2):

Synthesis (10.2) contains an extra step but uses very cheap and available starting materials. Synthesis (10.1) is shorter and goes in high yield but requires anhydrous conditions for the alkylation and a more expensive malonate starting material.

The use of Pd(0) to form carbon–carbon bonds was shown to be very effective in many cases. Could one of the coupling methods catalyzed by Pd(0) be used to attach the two-carbon fragment needed to construct the skeleton of C? Since we are restricted to cyclohexyl bromide as the starting material, direct reaction with Pd(0) is not feasible because Pd(0) does not give oxidative addition with saturated bromides. Moreover saturated bromides do not undergo transmetallation with Pd(0), so it could not serve as the second component in a Pd(0)-catalyzed coupling. Thus the reactivity requirements of Pd(0)-catalyzed coupling reactions are incompatible with the starting material and thus are not usable for the present construction.

Next consider the synthesis of M from “readily available” starting materials. The relevant facts about M are that (a) it contains an aromatic ring, an acetate ester, and a vinyl group; (b) it has a straight chain attached to the ring; and (c) all the functional groups are isolated.

To begin the retrosynthetic analysis, note that the acetate ester is easily pro-duced from the corresponding alcohol A. Therefore conversion of A to M using acetic anhydride/pyridine could be used in the synthetic step. (Remember: For each retrosynthetic step, a reaction must be available to accomplish the syn-thetic step.)

Now the alcohol functional group in A is a natural point for bond discon-nection to take place since alcohols are the products of carbon nucleophiles and carbonyl groups. If we consider bond a in our retrosynthetic analysis, then the next retrosynthetic step would be

Since the anion N is a nonstabilized carbanion, an organometallic nucleophile such as an organolithium or a Grignard reagent could be prepared from the corresponding bromide.

The bromide could be prepared from 3-phenyl-1-propanol ($59/kg). The unsaturated aldehyde O can be made by oxidation (CrO₃.py) of 4-penten-1-ol ($41.80/10 g). A cheaper way is to make ethyl 4-pentenoate from ethyl acetate ($15/gal) and allyl bromide ($19/100 g) and reduce it to the aldehyde O with DIBAH ($19/ 0.1 mol).

Thus a synthesis of M based on this retrosynthetic analysis would start with ethyl acetate, allyl bromide, and 3-phenyl-1-propanol.

Now we go back to A and consider disconnection at a different bond. Suppose we recognize that alcohol A could easily come from reduction of ketone K.

Now considering the polarities possible, a great number of disconnections can be envisioned. Choosing bond b means that polarity (with respect to the carbonyl group) would be

and thus an enolate reacting with a carbon electrophile would be appropriate. A valid retrosynthetic step would be

Because the tosylate is primary, substitution should be the major pathway (although in this case elimination could be problematic because of conjugation with the phenyl ring). We note, however, that the enolate needed is the kinetic enolate of 5-hexen-2-one. This poses a regiochemical control problem which can be solved by making the N ,N -dimethylhydrazone of the ketone. The ketone 5-hexene-2-one is available ($46.20/25 g) or can be made by allylation of acetone.

Thus a synthesis based on this retrosynthetic analysis starts with β-phenylethanol ($35.60/kg), acetone, and allyl bromide. This route is comparable to the first in both number of steps and cost. It differs in that regiochemical control of enolate formation is a crucial feature. Several other syntheses of K can be devised by other disconnections suggested by the natural polarities engendered by the ketone group.

Next consider compound R. When the relevant facts are considered, we see that R is merely an olefin with a saturated ring present. Because of the five-membered ring and because the target has 12 carbon atoms, it is unlikely that compounds with the carbon skeleton of R will be available commercially; hence carbon–carbon bond-forming reactions will be needed to assemble the carbon skeleton.

Moreover the carbon–carbon double bond is a natural starting point for bond disconnection. A logical retrosynthetic step would be disconnection to a ketone because a Wittig reaction could be used to convert the ketone to the ethylidene product. (Note that dehydration of an alcohol to the olefin is not a viable synthetic step because dehydration would lead to a mixture of trisubstituted olefins.)

Once the ketone is recognized as a useful intermediate, normal polarities can be used to disconnect it retrosynthetically. For example, a good disconnection could be as shown, where Michael addition to ethyl vinyl ketone by a cyclopentyl anion would give the needed ketone.

The cyclopentyl nucleophile, which should be an organocuprate to ensure Michael addition, could be produced from cyclopentyl bromide. The synthetic sequence consistent with the retrosynthetic analysis turns out to be a rather simple synthesis of what at first sight is a more difficult molecule.

Now there are a variety of other ways to disconnect R in the retrosynthetic analysis. As long as each synthetic step is valid and the target can be produced by the proposed synthetic route, then it is a correct solution. There can be many correct synthetic solutions for a given target and the “best” one may depend on factors other than those related strictly to the synthetic viability. Availability of starting materials, disposal of reaction by-products, number of steps, reagent sen-sitivity, expected yields, number of purifications, and the stereochemistry (among others) all contribute to the evaluation of a synthetic route.