Synthesis of a Novel Ureidopeptide HIV-1 Protease Inhibitor
Synthesis of a Novel Ureidopeptide HIV-1 Protease Inhibitor
In preliminary studies, it was shown that peptides with the urea backbone modification are more stable toward proteolysis. A novel hexapeptide analogue (1) was designed, based on two known HIV-1 protease inhibitors.
AIDS is a disease caused by the HIV virus that has greatly affected the world, thus a great deal of chemistry has been focused on the goal of developing a method of treatment for this illness. HIV comes from a single long peptide chain, which is broken into three functional enzymes in its active form: protease, integrase, and reverse transcriptase. The protease portion of the active virus acts as a set of “molecular scissors” which cuts the large chain into the functional portions by amide bond cleavage. Integrase integrates the virus into a host cell genome, reverse transcriptase converts viral RNA to DNA. Many well-known inhibitor strategies have chosen to focus on the reverse transcriptase to develop treatments such as AZT. These inhibitors alone have proved insufficient since the virus has been able to mutate to other, more stable, forms. This caused a shift in strategies to protease inhibitors with the theory that being able to stop the virus before it can be activated would be the best method. Other work has focused on synergistic inhibition by drug “cocktails.
To better understand the idea behind the design of this inhibition strategy, one must know a little more about the target enzyme. HIV-1 protease exists as a dimer, with each monomer containing 99 amino acids. This is an aspartyl protease since the active site is located at the junction of Asp-25 and Asp-25’. These residues are indicated in bold in Figure 1. In many typical inhibitors, a hydroxyl group has been used to mimic the tetrahedral intermediate in amide bond hydrolysis. It has also been shown that cyclic ureas have experienced success as inhibitors for this target. It is also of note that there are some “greasy” pockets of hydrophobicity caused by glycine residues at positions 26, 47, and 48. In the design of our ureidopeptide, we have combined the best of both worlds. Ureas have similar interactions in an active site to hydroxyl groups, while being more stable toward proteolysis. However, some of the cyclic ureas tend to not be as flexible as peptides. By combining the qualities present in the peptidic inhibitor (2) of Dreyer (Biochemistry 1992, 31, 6646-6659) as pictured at the left, with those of the cyclic urea (3) designed by Lam (Science 1994, 263(5145), 380-4) , we hope to have both the fit of a peptide while having the stability present in a urea. Based on this collective background knowledge, we believed that our hexapeptide mimetic would show improved inhibition.
Results and Discussion
The synthesis of 1 is performed by first synthesizing each tripeptide on either side of the urea then joining via a Hoffmann rearrangement. Synthesis of each half follows standard methodology for peptide synthesis, progressing from C- to N-terminus to avoid racemization of the amino acids. Though the project was started by a previous undergraduate, I have now performed all steps, and the yields and experimental details are from my experiences.
Scheme 1 details the synthesis of the left half of the molecule. I have performed reaction (a) two separate times, each with relative success. The first time had a 48% yield while the second had a quantitative yield. Formation of product was confirmed by NMR. When performing this for the second time, according to the same conditions, I found that the mixture was no longer stirring. Seeing this brought to my attention that solubility was a problem. After discussing this issue (also seen in other parts of the reaction) with my advisor, we observed that, when the phenylalanine residue was present, solubility was poor, which affected yields. To remedy this, a trace amount of DMF was added to the reaction mixture to improve solubility, which caused the solid to go in solution. This allowed the reaction to complete better, and give a quantitative yield.
The next step is removal of the Z protecting group, which is pictured to the right. This can be accomplished via hydrogenation with a palladium catalyst. This protecting group is rather facile to remove, but works efficiently to block unwanted combinations of amino acids, causing only one N-terminus to be present. The TLC and NMR indicated purity with the quantitative yield of greasy yellow product. No solubility issues were faced with this step of the synthesis.
The final step for Scheme 1 was the synthesis of Z-Ala-Ala-Phe-NH2. Poor solubility was noted and more THF was added to get reaction mixture better into solution, then the reaction was left overnight again. This time TLC showed completion. Product showed to be pure by TLC, giving a 55% yield. The purity of the product was confirmed by NMR. The yield and timing of this reaction would most likely have been helped if DMF had been used, so the reaction conditions (c) as described in Scheme 1 reflect the newest procedure.
next portion of the experiment is described in Scheme 2, listed at the right. For this portion of the experiment, I was able to use some of the product of a previous undergraduate. However, I am reporting only my own yields and experimental data. TLC showed purity of the product, as well as NMR. This reaction had an 86% yield.
I have run two different deprotections of Z-Val-Val-OMe. In the first attempt, I obtained a 96% yield, while the second was quantitative (actually 100.5%). NMR was used to confirm purity.
Reaction (c) of Scheme 2 was the formation of Z-Phe-Val-Val-OMe. This reaction was also performed twice. I first obtained a 63% yield, using NMR to show its purity. In the second attempt, I had a 75% yield. No NMR was taken since TLC exactly matched the earlier attempt, which was not good lab practice.
The next reaction is the deprotection of Z-Phe-Val-Val-OMe. Two attempts were also made on this. The first had a 45 % yield confirmed by NMR. The second attempt gave a 64% yield, confirmed by the same conditions.
The final stage of the synthesis is the urea coupling. Scheme 3 describes this set of reactions. Of note in this synthesis is the fact that a Hoffmann rearrangement is performed on 4 to yield an isocyanate intermediate, which can then be reacted with the free amine of piece 5 to form the urea linkage. In my first attempt at this reaction, my notes were incomplete, and I did not know to add the pyridine. It did not work. After having referred to two other documents, I discovered that the pyridine was needed. I added it into the reaction mixture already in the (b) phase of the scheme shown above. This did not produce the desired results. NMR showed a complex mixture, and TLC showed three separate spots, which didn’t correlate well with anything known. After discussion with my advisor and other group members, we decided it would be best to leave this product to be worked up later, if it was deemed to be relevant after a successful run of the reaction. From this experience and discussion with Dr. Jennifer Kowalski (a previous worker in ureidopeptides) I was able to determine the precise conditions needed for the reaction, and made a second attempt that gave a 23% crude yield. I had to dry load the column for flash column chromatography to purify, giving a 9% yield of pure product. This was not good at all. We decided to wait until I had more 1 to carry the reaction any further. In my third attempt at this reaction, there was a bit more success. To aid solubility, DMF was added to the reaction this time. The third attempt gave a 28% crude yield. We decided to try purification by HPLC using acetonitrile and water. The gradient with the best separation was 25-70% acetonitrile/water. In the first attempt at separation in prep scale, someone had switched the solvent bins to have acetone instead of acetonitrile. This ruined this run, which contained more than half of my product. While storing these fractions, we ran a good run on a portion of the product. Product was in fraction 4, but I only had 1.6 mg. I did get NMR and the sample was shown to be pure (a copy of the spectra is attached to the back of this report). 79.8 mg of product was recovered from the acetone/water fractions, meaning that I had a minimum of 25% recovery in the HPLC run. I attempted several different solvent systems for a flash column in order to improve solubility, while maintaining the separation of the 10% methanol/dichloromethane. After trying combinations including acetonitrile, dichloromethane, petroleum ether, and methanol, I didn’t find a better solvent system. Based on this data and discussion with my advisor, I used part of the crude product recovered from the bad HPLC run to dry load the column, still running with 5% methanol/dichloromethane. Early in purification, I accidentally added 5% acetonitrile/dichloromethane. Once back to the correct solvent system, TLC was used to find the product, but not much showed up. Nothing appeared by TLC in the specifically combined fractions or acetonitrile fractions, but the other fractions of the run contained product. Since I did not know from which fraction(s) the product had come, I flushed the column with 10% methanol/dichloromethane but found nothing, even after concentrating samples and checking by TLC. From this purification, 4.7 mg of Z-Ala-Ala-Phe-[u]-Phe-Val-Val-OMe (1) was obtained, which was a recovery of 12.8%. This low value can partially be attributed to the acetonitrile mistake.
Several conclusions can already be drawn based upon the work completed. There are two current problem areas of this synthesis that still need to be addressed. The first is the issue of purification. As is common knowledge, reverse-phase HPLC often is associated with more product loss than usual. The question that must now be evaluated is the difference between recovery on good runs of both a flash column and an HPLC. When used properly, HPLC provides more accuracy by using UV-Vis detection, while the last column attempt showed poor ability to find the product by TLC. However, if the flash column had been run properly (not using acetonitrile), the recovery may have been substantially better. The second problem is efficiency of the urea coupling reaction (Scheme 3). Currently my yields are lower that Dr. Kowalski ever showed with the different target molecules in the research she conducted. Upon discussion with my advisor, we have decided that some product loss may arise from workup of the urea coupling reaction. For future systems, we will concentrate the reaction to dryness, and apply the whole mixture to a flash column in an effort to obtain better yields. In order to master the reaction, a model system should be used to better master this reaction. These two areas need to be addressed in order for the synthesis to be completed for there to be enough Z-Ala-Ala-Phe-[u]-Phe-Val-Val-OMe (1) to biologically assay the Z-protected form and the free amine form.
To complete this project, the direct comparison between HPLC and flash column chromatography must first be completed. Then, the final product must be produced. Lastly, enough of both the protected and unprotected ureidopeptide must be produced to perform a more complete characterization and biological assays. After these steps are completed, this project can be brought to a close.
Scheme 1, reaction (a) was performed twice. The first time, in a dry round-bottom flask, equipped with a stir bar and under nitrogen, I combined 503 mg (1 eq) Z-Ala-OH, 496 mg (1.1 eq) H2N-Phe-NH2, 334 mg (1.1 eq) HOBT, and 0.87 mL (2.2 eq) DIEA in 25 mL dry THF and let stir in an ice bath for about fifteen minutes. Then I added 477 mg (1.1 eq) EDCI. The reaction mixture was creamy-white in color. After 24 hours the appearance was about the same, and TLC (10% methanol/dichloromethane) showed the reaction to be as complete as in previous attempts (but not all starting material was consumed). I concentrated the sample to dryness, then dissolved in ethyl acetate. I washed the product three times each with 10% Citric Acid, saturated NaHCO3, and brine, then dried the organic layer with anhydrous sodium sulfate. After concentrating to dryness, I loaded and ran a flash column, collecting 40 fractions, with product showing in 16-34. There was a good deal of trouble in dissolving the product. I ended with 402 mg product, which was a 48% yield. Product formation was confirmed by NMR. When performing this for the second time, according to the same conditions, I found that the mixture was no longer stirring. To alleviate solubility problems a trace amount of DMF was added to the reaction mixture to improve solubility, which caused the solid to go in solution. This allowed the reaction to complete better, and give a quantitative yield.
Scheme 1, reaction (b) showed good results. I transferred 311 mg of the Z-Ala Phe-NH2 to a round bottom flask with methanol and a small amount of Pd(OH)2/C. I equipped the flask with a stir bar and hydrogenation balloon apparatus. I flushed the flask with one balloon of argon, then with hydrogen. I added more hydrogen and began stirring the reaction. After the reaction was complete by TLC, I filtered through a Celite patty then concentrated to dryness. TLC and NMR were taken to determine purity. I had a quantitative yield of greasy yellow product.
Scheme 1, reaction (c) was also rather facile. In a dry round bottom flask equipped with stir bar and nitrogen line I combined 308 mg (1 eq) H2N-Ala-Phe-NH2, 261 mg (1.1 eq) Z-Ala-OH, 161 mg (1.1 eq) HOBT, and 0.410 mL (2.2 eq) DIEA in 15 mL dry THF and let stir for fifteen minutes in an ice bath, then added 226 mg (1.1 eq) EDCI and let react overnight. When checking for completion by TLC, the reaction was not yet complete. Poor solubility was noted and more THF was added to get reaction mixture better into solution, then the reaction was left overnight again. This time TLC showed completion. Workup went the same as in reaction (a) of Scheme 1, giving a 55% yield. The purity of the product was confirmed by NMR. In the future, DMF should be used to help solubility and timing.
Scheme 2, reaction (a), I combined 501 mg (1 eq) Z-Val-OMe, 369 mg (1.1 eq) H2N-Val-OMe (the HCl salt), 300 mg (1.1 eq) HOBT, and 0.76 mL (2.2 eq) DIEA in 20 mL THF in a dry round bottom flask with a stir bar and nitrogen line. After stirring for 15 minutes in an ice bath I added 420 mg (1.1 eq) EDCI and let react for 24 hours. When reaction was complete by TLC, I concentrated the sample to dryness then dissolved in ethyl acetate. After washing three times each with 10% citric acid, saturated NaHCO3, and brine, I dried the organic layer with anhydrous sodium sulfate and TLC showed purity of product. I concentrated again to dryness and had 625 mg, an 86% yield. NMR also showed purity.
Scheme 2, reaction (b) was performed twice. In the first attempt, I used 404 mg Z-Val-Val-OMe in methanol with a small amount of Pd(OH)2/C, using a round bottom flask with stir bar and balloon hydrogenation apparatus. After flushing with one balloon of argon, I flushed with hydrogen, then refilled and let react for a total of 6 hours, when TLC showed completion of the reaction. After filtering through Celite and concentrating to dryness, I had 245.4 g, a 96% yield. NMR showed that the product was pure. In my second attempt, I used 624.7 mg of product under the same conditions, but showed completion by TLC in around 3 hours. After filtration through Celite and concentration to dryness, I had a yield of 396.9 mg. This is quantitative (actually 100.5%). NMR was again used to confirm purity.
Scheme 2, reaction (c) was run twice. In a dry round bottom flask with stir bar and nitrogen line, I combined 150mg (1 eq) H2N-Val-Val-OMe, 214.4 mg (1.1 eq) Z-Phe-OH, 96.7 mg (1.1 eq) HOBT, and 0.25 mL (2.2 eq) DIEA in THF with trace DMF. After stirring in an ice bath for 15 minutes, I added 137.3 mg (1.1 eq) EDCI and let react for four hours. At this time, TLC showed disappearance of starting material. I concentrated the reaction to dryness, dissolved in ethyl acetate, and washed three times each with 10% citric acid, saturated NaHCO3, and brine. After drying the organic layer with anhydrous sodium sulfate, I concentrated to dryness. I had 209.1 mg product, which was a 63% yield. NMR showed purity. In the second attempt, 102 mg (1 eq) H2N-Val-Val-OMe, 146 mg (1.1 eq) Z-Phe-OH, 65.8 mg (1.1 eq) HOBT, and 0.1 mL (2.2 eq) DIEA were combined in 6 mL THF with 0.5 mL DMF in a dry round bottom flask with stir bar and nitrogen line. After reacting four hours, TLC showed completion of the reaction. The exact same workup from the first attempt was performed. The reaction yielded 171 mg product, which translates to a 75% yield. No NMR was taken since TLC exactly matched the earlier attempt.
Scheme 2, reaction (d) had two attempts made on it. Basic reaction conditions used the same assembly of a round bottom flask, stir bar, and hydrogenation balloon apparatus, with methanol as the solvent and a small amount of Pd(OH)2/C. After reacting, each reaction mixture was filtered through Celite. In the first attempt, I started with 211.2 mg of the protected peptide, and yielded 70.5 mg (45%). NMR was used to determine purity. For the second attempt, I started with 203.1 mg Z-Phe-Val-Val-OMe and obtained 140.4 mg product, which was a 94% yield. TLC and NMR showed purity.
Scheme 3, reactions (a) and (b) did not work the first time since pyridine was added much too late. In my second attempt, I combined 48.6 mg (1 eq) Z-Ala-Ala-Phe-NH2 (4), 62 mg (1.5 eq) PIFA ([bis(trifluoroacetoxy)iodo]benzene), and 0.015 mL (2 eq) pyridine in >6 mL THF in a dry round bottom flask with stir bar and nitrogen line. reaction progress was monitored by TLC to see formation of the isocyanate intermediate. Between 2.25 hours and 3 hours, very little change was noted between TLC plates, so I added the 35.9 mg (1 eq) H2N-Phe-Val-Val-OMe (5) and 0.037 mL (2.2 eq) DIEA, then continued monitoring by TLC. After about 4.25 hours the starting material has appeared to disappear, so I continued with workup. I concentrated the sample to dryness, then dissolved in ethyl acetate. After washing three times each with 10% citric acid and brine, I dried the organic layer with anhydrous sodium sulfate. TLC appeared relatively pure, and crude weight (obtained after concentrating to dryness again) was 18 mg, a 23% crude yield. To load the flash column for purification, I dissolved silica in methanol to deactivate, then added my product and concentrated to dryness. This was loaded on top of the column, which was run with 10% Methanol/dichloromethane. Product only appeared in fractions 2-7 (out of approximately 20). I combined these together and concentrated to dryness, giving 7 mg of pure Z-Ala-Ala-Phe-[u]-Phe-Val-Val-OMe (1) by TLC, which was a 9% yield. This was not good at all. We decided to wait until I had more 1 to carry the reaction any further. In my third attempt at this reaction, there was a bit more success. To aid solubility, DMF was added to the reaction this time. In a round bottom flask with stir bar and nitrogen line 197.5 mg (1.1 eq) Z-Ala-Ala Phe-NH2 (4), 241.3 mg (1.5 eq) PIFA, 0.061 mL (2 eq) pyridine were combined in THF with trace DMF. At one hour (after TLC monitoring and staining with both CAM and Ninhydrin) I added 141.2 mg (1 eq) H2N-Phe-Val-Val-OMe (5) and 0.144 mL (2.2 eq) DIEA. After 4.5 more hours, TLC showed completion of the reaction. The reaction was concentrated to dryness, dissolved in ethyl acetate, and washed three times each with 10% citric acid and brine. The organic layer was dried with anhydrous sodium sulfate and concentrated to dryness. Crude yield was 86.3 mg (28% yield). Purification has been performed using HPLC and flash column, and is still occurring. NMR has been taken and has shown purity (spectrum is attached at the back of this report).