Designed, synthetically accessible bryostatin analogues potently induce activation of latent HIV reservoirs in vitro.

Bryostatin is a unique lead in the development of potentially transformative therapies for cancer, Alzheimer's disease and the eradication of HIV/AIDS. However, the clinical use of bryostatin has been hampered by its limited supply, difficulties in accessing clinically relevant derivatives, and side effects. Here, we address these problems through the step-economical syntheses of seven members of a new family of designed bryostatin analogues using a highly convergent Prins-macrocyclization strategy. We also demonstrate for the first time that such analogues effectively induce latent HIV activation in vitro with potencies similar to or better than bryostatin. Significantly, these analogues are up to 1,000-fold more potent in inducing latent HIV expression than prostratin, the current clinical candidate for latent virus induction. This study provides the first demonstration that designed, synthetically accessible bryostatin analogues could serve as superior candidates for the eradication of HIV/AIDS through induction of latent viral reservoirs in conjunction with current antiretroviral therapy.

HIV/AIDS is a global pandemic[1]. The Joint United Nations Programme on HIV/AIDS estimated that the number of people living with HIV, the virus that causes AIDS, totaled 33.3 million in 2009 (see http://unaids.org/globalreport/Global_report.htm). In the same year, there were 1.8 million AIDS-related deaths.

The leading treatment for HIV is highly-active antiretroviral therapy (HAART), a combination of drugs that halts viral proliferation by inhibiting several stages of the viral life cycle. For many patients, this treatment strategy has transformed HIV into a manageable, chronic disease by reducing their plasma viral loads often to undetectable levels. However, HAART is not curative, as it does not address genomically-integrated latent viral reservoirs that slowly resupply replication-competent active virus. As such, discontinuation of HAART results in viral rebound and disease progression[2]. In addition, HAART is costly, has associated side effects[3], and requires strict adherence to treatment regimens[4] to avoid the emergence of viral resistance[5].

The most formidable obstacle to the eradication of HIV is the persistence of various latent proviral reservoirs[6-8]. These are believed to be primarily established following integration of the HIV genome into that of activated CD4+ T-cells and other cell types. In rare cases, these infected cells transition to quiescent memory cells in a process that reversibly inhibits expression of the integrated HIV provirus, rendering it unsusceptible to HAART[9]. As a consequence, eradication of HIV in HAART-suppressed patients would require elimination or inactivation of these proviral reservoirs.

As HAART targets only actively replicating virus, it has little influence on latent viral reservoirs. It is estimated that decades of HAART treatment would be required for depletion of the reservoir source[10]. Therefore, agents that can controllably facilitate purging of the latent virus from these reservoirs could reduce the time for depletion and provide a strategy to eradicate HIV when used in combination with HAART. In this approach, which is now being actively pursued[11], the deliberate induction of viral replication from its latent state is proposed to eliminate HIV-harboring cells either by direct viral cytopathic effects or by rendering those cells susceptible to immune system regulation. Concomitant HAART would prevent infection of healthy cells from the released virions. This combined approach would thus eliminate both the latent and active viral pools.

One experimental strategy for latent viral reactivation is direct immunological modulation of resting memory T-cells; however, administration of cytokines and/or antibodies has thus far lacked clinical efficacy[12,13]. Another approach employs pharmacological modulation of signaling pathways associated with viral reactivation. Toward this end, Protein Kinase C (PKC) has emerged as an important target. Prostratin (Fig. 1), a non tumor-promoting 12-deoxy phorbol ester that binds to and activates PKC, has demonstrated promising in vitro and ex vivo activities in reactivating latent HIV[14,15] and is currently being advanced as a clinical candidate for latent viral reservoir clearance[16]. A source of synthetic prostratin has recently been reported that also provides access to even more effective analogues[17].

Figure 1

Bryostatin, Prostratin, and synthetic analogues

Like prostratin, bryostatin 1 also targets PKC but it is significantly more potent (Ki= 1.35 nM vs. 50 nM) and thus represents a unique lead for HIV reservoir clearance. Bryostatin 1, the lead member of the bryostatin family, was reported in 1982 by Pettit and Clardy[18], and has subsequently been shown to possess promising in vitro, in vivo, and human clinical[19] activities (see http://clinicaltrials.gov) for a diverse array of indications[20] including cancer and Alzheimer’s disease[21]. Early reports also suggested that bryostatin might induce HIV reservoir clearance[22,23], and this has been further supported in more recent studies[24]. An attractive aspect of bryostatin is that it has already been tested in human trials, albeit for other cancer-related indications. However, further underscoring the need for a reliable supply of a more effective agent, patient accrualin a recent bryostatin clinical trial was terminated by the National Cancer Institute and the clinical investigators “given the more potent bryostatin analogs in development”, a focus of this current study[25].

Unlike prostratin and its analogs for which a practical five-step synthesis has been reported[17], bryostatin has not been reliably available in quantities needed for research and sustained clinical studies. Isolated yields from natural sources are low (10−3 to 10−8%) and variable; the GMP production required 14 tons of the marine bryozoan Bugula neritina to provide just 18 g of bryostatin 1[26]. Although this supply was sufficient to initiate preclinical and clinical research on bryostatin, economic and environmental factors have severely limited further development of this source and related aquaculture production. Engineered biosynthesis, while promising, is still in development and has yet to impact clinical supply[27]. While impressive progress has also recently been made in the realm of bryostatintotal synthesis[28-34], approaches to the potent bryostatins require at best approximately 40 total steps. Importantly, clinical studies have uncovered off-target toxicities with bryostatin. In principle, issues of both supply and undesired effects could be addressed by the design of simplified, clinically superior, functional analogues that can be produced in a more step-economical fashion.

To address these supply and performance issues, our group proposed in 1988 that the activities of bryostatin could arise from only a subset of its functionality[35]. As such, bryostatin-like activity could be achieved with simplified and thus more synthetically accessible designed analogues dubbed “bryologs” that incorporate the key activity-determining functionalities[36,37]. Using this function-oriented synthesis[38] approach, we reported in 1998 the first simplified bryostatin analogues[39] and more recently similar bryologs that are even more potent than the natural product[40,41,42]. Significantly, these analogues, several of which were prepared in under 30 total steps, are currently the most synthetically accessible agents with bryostatin-like activity, PKC translocation selectivity[43], and potency. Moreover, a related bryostatin analogue has been shown to be well tolerated and efficacious in an in vivo mouse cancer model[44]. To date, there has been no reported study of simplified bryostatin analogues for the activation of latent HIV reservoirs as required for disease eradication. Herein, we report the convergent syntheses of seven members (1–7) of a new family of bryostatin analogues incorporating a tetrahydropyranyl B-ring formed through a versatile Prins macrocyclization. We also disclose for the first time that such designed, synthetically-accessible analogues can induce activation of latent HIV expression in a cellular latency model with potencies similar to or better than bryostatin 1. We also demonstrate that these simplified bryostatin analogues share the functional activity of the clinical candidate prostratin in this assay but are up to four orders of magnitude more potent than prostratin. These findings establish that bryologs, in addition to their therapeutic potential for cancer and Alzheimer’s disease[45], could serve as synthetically accessible and potentially superior clinical candidates as adjuvants with HAART for the eradication of HIV infection.

Results

The assembly of the new bryostatin analogues 1–7 was proposed to entail a highly convergent strategy in which a spacer domain (Fig. 2: 8, 9, or 10) would be conjoined with a recognition domain (11) through a two-step esterification/Prins-driven macrocyclization[46] sequence. This process would form a readily diversifiable B-ring pyran and set C15 stereochemistry to produce the key intermediates 20–22. Support for this strategy comes from our prior work on the syntheses of simplified bryostatin analogues[41] as well as bryostatin 9[33]. As the name implies, the recognition domain is proposed to directly contact PKC, thereby influencing analogue-PKC affinity, while the spacer domain is proposed to control bryolog conformation and to influence translocation of the bryolog/PKC complex and its insertion into the cellular membrane. The preparation of analogues 1–7 was proposed to start from a common A-ring intermediate 12, which was prepared and elaborated as described previously[42] into hydroxyesters 13 and 14. Hydroxyester 13 would be used to synthesize the C7-deoxy analogues 1 and 5, while hydroxyester 14 would be used to prepare the C7-oxy analogues 2, 3, and 6. Analogues 4 and 7, which possess bryostatin 1-like A-ring functionalization, would be prepared from spacer domain 10[33].

Figure 2

Synthesis of analogues 1–4 via Prins-driven macrocyclization a

a Reagents and Conditions: When X = H: (a) TESCl, imidazole, CH2Cl2, 95%; (b) CeCl3·2LiCl, TMSCH2MgCl, THF; (c) Silica gel, CH2Cl2, 85% over 2 steps; (d) lithium naphthalenide, THF, 84%; (e) TPAP (10 mol %), NMO, 4Å MS, CH2Cl2; (f) NaClO2, NaH2PO4, 2-methyl-2-butene, 2:1 t-BuOH:H2O, 91% over 2 steps; When X = OTBS: Conditions as before: (a) >99%; (b, c) 82% over 2 steps; (d) 91%; (e, f) 97% over 2 steps. (g) 2,4,6-trichlorobenzoyl chloride, Et3N, then DMAP, recognition domain 11, PhCH3, rt. (h) From and PPTS, EtOH, rt; From i. PPTS, MeOH, rt, ii. TBSCl, imidazole, CH2Cl2. (i) From and HF·pyridine, THF, rt; From i. PPTS, MeOH, rt, ii. HF·pyridine, THF, rt, iii. PPTS, 4:1 THF:H2O; (j) i. TESCl, imidazole, CH2Cl2, ii. Ac2O, DMAP, pyridine, CH2Cl2; (k) HF·pyridine, THF.

Our studies started with previously prepared hydroxyesters 13 and 14[42], which were converted into allylsilanes 15 and 16, respectively, in 3 steps involving C11 silylation, CeCl3·2LiCl-mediated[47] double nucleophilic addition of TMSCH2MgCl, and Peterson olefination of the carbinol product. Knochel’s CeCl3·2LiCl salt provided optimal yields for the nucleophilic addition step[33]. Intermediates 15 and 16 were then separately debenzylated with lithium naphthalenide (84–91% yield), and the resultant C1 alcohols oxidized with TPAP/NMO followed by NaClO2 to provide the spacer domains 8 and 9 (91–97% yield over 2 steps).

Spacer domains 8, 9, and 10 were then each esterified with recognition domain 11 using Yamaguchi’s esterification to provide the Prins macrocyclization precursors 17, 18, and 19, respectively. Significantly, we found that C11 deprotection and Prins-driven macrocyclization of 17, 18, and 19 can be promoted in one operation and in excellent yields (up to 90%) using mild conditions (catalytic PPTS in alcohol solvent) providing macrocycles 20, 21, and 22, respectively. Importantly, only a single diastereomer is obtained in each case. The Prins-driven macrocyclizations of 17 and 18 proceeded rapidly with PPTS/EtOH without C26 desilylation. Macrocyclization of 19 with PPTS/MeOH occurred less rapidly giving macrocycle 22 and someC26-desilylated product; this crude mixture of macrocycles was either further deprotected to directly provide analogue 4 (51% over 3 steps) or reprotected to provide intermediate 22 (73% overall yield) en route to analogue 7. Analogues 1 and 2 were prepared by deprotection of macrocycles 20 and 21, respectively. The C7-OH analogue 2 was converted into the C7-OAc analogue 3 by C26 silylation followed by C7 acylation and desilylation.

Stoichiometric ozonolysis of the C13-methylidene subunits of bryopyrans 20, 21, and 22 followed by Horner-Wadsworth-Emmons olefination afforded intermediates 26, 27, and 28, respectively (Fig. 3). While we have shown that reagent-controlled olefination can afford greater selectivity for the Z-enoate[33], preparation of both E- and Z-enoate isomers was required for ongoing comparative activity studies. In this study, the activities of only the Z-enoates are reported (vide infra). Deprotection of intermediates 26–28 followed by reverse-phase HPLC purification afforded enoate analogues 5 and 7 and intermediate 29 en route to analogue 6. The C7-OH intermediate 29 could be converted to the C7-OAc analogue 6 using the protocol described above involving C26 silylation followed by C7 acylation and desilylation.

Figure 3

Synthesis of analogues 5–7 via ozonolysis followed by Horner-Wadsworth-Emmons olefination a

a Reagents and Conditions: (a) O3, CH2Cl2, −78 °C, then thiourea, MeOH:CH2Cl2; (b) trimethyl phosphonoacetate, NaHMDS, THF, 0 → 4 °C; (c) From 26 or 27: HF·pyridine, THF, rt; From 28: i. HF·pyridine, THF, rt, then ii. PPTS, 4:1 THF:H2O, rt; (d) i. TESCl, imidazole, CH2Cl2, ii. Ac2O, DMAP, pyridine, CH2Cl2; (e) HF·pyridine, THF.

Analogues 1–7 were found to exhibit excellent affinities for a rat brain mixture of PKC isoforms with single-digit nanomolar or subnanomolar Ki values (Table 1). This PKC mixture was selected because it allows for calibration of the potency of the new bryostatin analogues 1–7 with data obtained previously for other bryostatin analogues[39-42]. Importantly, this represents the first quantitative comparison of the relative affinities of bryostatin 1, prostratin, and synthetic bryostatin analogues using a common PKC mixture. Significantly, all of the new bryostatin analogues (1–7) were more potent in this assay than the clinical candidate prostratin. Similar to previous observations on simplified substrates[41], Z-enoate analogues 5, 6, and 7 were found to be slightly more potent than their des-enoate counterparts 1, 3, and 4. As we had observed for the B-ring dioxane scaffold[42], and similar to the potency difference between bryostatin 1 (C7-OAc, lit. Ki = 1.4 nM35) and bryostatin 2 (C7-OH, lit. Ki = 5.9 nM35), the C7-OH analogue 2, while highly potent (Ki = 3.4 nM), was less so than the C7-OAc analogue 3 (Ki = 0.42 nM) or the C7-deoxy analogue 1 (Ki = 0.58 nM).

Table 1

Analogue PKC affinity and activity in the J-Lat cell line model of HIV latency

Analogue	X (C7)	Y (C9)	B-Ring	PKC Ki (nM)a	J-Lat EC50 (nM)b
Bryostatin 1	OAc	OH	Z-Enoate	0.28 (0.18 – 0.44)	1.61 (0.92 – 2.84)
Prostratin	-	-	-	6.6 (4.1 – 10.6)	>1000
1	H	H	Pyran	0.58 (0.41 – 0.81)	37.4 (21.6 – 64.9)
2	OH	H	Pyran	3.4 (1.7 – 6.6)	15.2 (7.5 – 30.8)
3	OAc	H	Pyran	0.42 (0.22 – 0.77)	32.0 (16.3 – 62.8)
4	OAc	OH	Pyran	0.95 (0.67 – 1.4)	0.46 (0.31 – 0.69)
5	H	H	Z-Enoate	0.46 (0.28 – 1.1)	1.9 (0.91 – 3.95)
6	OAc	H	Z-Enoate	0.32 (0.17–0.60)	1.15 (0.42 – 3.18)
7	OAc	OH	Z-Enoate	0.79 (0.58 – 1.1)	0.38 (0.21 – 0.69)

Determined in a rat brain isoform mixture; results from single experiments are presented; error ranges indicate 95% confidence intervals from nonlinear regression analysis.

EC50 for induction of GFP transcription in the J-Lat cell line. GFP transcription indicates transcription of the HIV-Long Terminal Repeat and correlates with viral reactivation from latency; error ranges indicate 95% confidence intervals.

After establishing that the novel bryostatin analogues bind effectively to PKC, we tested their HIV latency induction activity. Several cellular models have been developed to assess the ability of particular agents to activate HIV from post-integration latency[48]. These models typically employ stable cell lines harboring an integrated, yet transcriptionally-silent, partial HIV genome. Upon treatment with an appropriate agent, viral reactivation is assessed by observation of either the production of natural viral proteins or other reporter proteins whose expression is coupled to that of the virus. J-Lat cell lines harbor near full-length latent HIV proviruses that express green fluorescent protein (GFP) upon stimulation of virus expression[49]. Treatment of J-Lat cells with certain PKC activators such as prostratin[50] or bryostatin 1[24] results in the induction of HIV expression[51]. Alongside these benchmark agents, we examined the ability of our novel bryostatin analogues to induce HIV expression in J-Lat cells (clone 10.6) at varying concentrations. In this assay, bryostatin induced expression of HIV at concentrations as low as 1 nM (Table 1). Significantly, several of the bryostatin analogues displayed comparable or better potency, when compared to bryostatin, in both EC50 and the percentage of cells induced to express latent HIV. For example, analogues 4 and 7 both had EC50 values below 1 nM and induced HIV expression in a higher percentage of cells than bryostatin at all concentrations tested between 0.1 nM and 1 μM (See Supplementary Figure S1). Similar to the results obtained in the PKC binding affinity assay, Z-enoate analogues 5, 6, and 7 were found to be more potent than their des-enoate counterparts 1, 3, and 4 in their ability to activate latent HIV expression in the J-Lat cell line. Moreover, all of the bryostatin analogues evaluated in this assay were at least 25-fold more potent than prostratin in inducing HIV from latency, with analogues 4 and 7 being over 1000-fold more potent.

Discussion

Bryostatin 1 shows great promise as a lead candidate in the search for transformative therapies aimed at cancer and neurodegenerative disorders such as Alzheimer’s disease. More recently, Bryostatin 1 was shown to induce the expression of HIV from latent viral reservoirs, an activity of potential importance for eradication of HIV infection. However, a sustainable, cost effective supply of bryostatin 1 has not been established. More importantly, an agent that does not exhibit the side effects of bryostatin would be desired. To address these issues of both supply and therapeutic performance, we have designed a series of bryostatin analogues (1–7) that can be synthesized on scale and can be tuned for optimal clinical performance. Analogues 1–7 were found to exhibit excellent affinities for PKC with single-digit nanomolar or subnanomolar Ki values, and all were more potent in this assay than the preclinical candidate prostratin. Significantly, these simplified bryologs were shown to induce expression of latent HIV in vitro with potencies similar to or better than the natural product, bryostatin 1, and at doses up to 1000-fold lower than that of prostratin.

The study presented herein demonstrates that bryostatin analogues potently activate latent HIV reservoirs in vitro. Notably, none of the compounds evaluated in this assay showed overt toxicity at concentrations up to 1 micromolar based on flow cytometry forward and side scatter profiles. While in vivo studies of these bryostatin analogues 1–7 in an animal viral induction model are in progress, a related analogue has been shown to be well tolerated and efficacious in a mouse cancer model, even at doses of in excess of that reported herein (up to 1 mg/kg)[44]. Moreover, the convergent nature of this strategy allows for additional tuning if needed to improve therapeutic performance. Further evaluation of these compounds in additional in vitro, ex vivo, and in vivo latency inductions assays is in progress.

This study is the first demonstration that designed, simplified analogues of bryostatin can serve as therapeutic leads for the eradication of HIV/AIDS. Given the problems associated with the supply of bryostatin 1 and with its side effects, this study provides a reliable synthetic source of agents that are comparable or superior to bryostatin in activity and can be tuned to accommodate clinical needs. Viral reactivation with these agents, performed in combination with HAART, would purge latent viral reservoirs while simultaneously depleting active virus. Coupling this approach with a targeted therapeutic such as an immunotoxin to more rapidly kill any resulting virus-expressing cell represents an additional therapeutic opportunity[52]. This type of approach, now enabled by the availability of highly potent and tunable analogues, could be used to clear all replication-competent HIV from infected individuals, thereby providing a strategy for disease eradication.

Methods

Experimental details for the synthesis of all new compounds, including experimental procedures, characterization and spectral data are provided in the Supplementary Information. Assay protocols for the PKC competitive binding assay and HIV latency activation assay are also provided in the Supplementary Information.