α-Amino bisphosphonate triazoles serve as GGDPS inhibitors

Abstract

Depletion of cellular levels of geranylgeranyl diphosphate by inhibition of the enzyme geranylgeranyl diphosphate synthase (GGDPS) is a potential strategy for disruption of protein transport by limiting the geranylgeranylation of the Rab proteins that regulate intracellular trafficking. As such, there is interest in the development of GGDPS inhibitors for the treatment of malignancies characterized by abnormal protein production, including multiple myeloma. Our previous work has explored the structure-function relationship of a series of isoprenoid triazole bisphosphonate-based GGDPS inhibitors, with modifications having impact on enzymatic, cellular and in vivo activities. We have synthesized a new series of α-amino bisphosphonates to understand the impact of modifying the alpha position with a moiety that is potentially linkable to other agents. Bioassays evaluating the enzymatic and cellular activities of these compounds demonstrate that incorporation of the α-amino group affords compounds with GGDPS inhibitory activity which is modulated by isoprenoid tail chain length and olefin stereochemistry. These studies provide further insight into the complexity of the structure-function relationship and will enable future efforts focused on tumor-specific drug delivery.

Graphical Abstract

Protein prenylation is a form of post-translational processing that introduces a hydrophobic moiety at a cysteine residue near the C-terminal of proteins in the Ras superfamily, thus facilitating association with a cellular membrane. Both of the linear isoprenoids farnesyl diphosphate (FPP, 1) and geranylgeranyl diphosphate (GGPP, 2) are substrates for prenyl transferases with the group transferred dependent upon the peptide sequence at the C-terminus of the protein substrate. Among the proteins modified by GGPP are the Rho proteins involved in cancer cell migration¹ and the Rab proteins essential for intracellular trafficking.² Given their important roles in key cellular processes, it is not surprising that inhibitors of protein prenylation have been studied extensively as potential anti-proliferative agents and for treatment of other diseases.³ The first farnesyltransferase inhibitor in clinical use, lonafarnib, was approved by the FDA for treatment of progeria in 2020.^4–6

While many have sought inhibitors of the prenyl transferases, our approach has been to focus on inhibitors of the downstream enzyme geranylgeranyl diphosphate synthase (GGDPS), the enzyme that appends an additional isoprene unit to FPP to afford GGPP. An effective inhibitor of this enzyme limits cellular levels of GGPP, ultimately depriving geranylgeranyl transferases of the necessary isoprenoid substrate, and thus disrupting protein gernaylgeranylation.^7–8 Our focus has been on development of GGDPS inhibitors as potential anti-myeloma agents. Disruption of Rab geranylgeranylation results in impairment in monoclonal protein trafficking within myeloma cells, leading to ER stress and cell death.^9–11

We have reported a number of compounds that are potent inhibitors of GGDPS. The first was digeranyl bisphosphonate (3, DGBP),¹² which showed an IC₅₀ of ~260 nM for GGDPS and good selectivity over the related enzyme farnesyl diphosphate synthase,¹³ but cellular activity was observed only at high concentrations (~10 μM). Independent crystallographic analysis indicated that DGBP’s V-shaped structure occupied the enzyme’s active site, with the phosphonates complexed to resident magnesium ions and the isoprenoid chains occupying the FPP and GGPP channels.¹⁴ More recently we have focused on monoalkyl compounds assembled through click chemistry and incorporating a triazole ring system as a substitute for one isoprene unit. These efforts first yielded the inhibitor 4, which showed enzyme activity at 2.2 μM (IC₅₀) and cellular activity ~1 μM.¹⁵ Surprisingly, as a mixture of olefin isomers the homologue 5 (VSW1198) showed an IC₅₀ of 45 nM against GGDPS, high specificity for GGDPS over FDPS, and cellular activity at concentrations as low as 30 nM in myeloma cells.^16–17 After preparation of the individual isomers and methylation of the alpha position, bioassays revealed these new compounds 6 (RAM2093) and 7 (RAM2061) had IC₅₀ values of 125 and 86 nM respectively, with cellular activity at concentrations as low as 20 and 25 nM levels.¹⁷ While these levels of potency are certainly encouraging, as is the evidence of in vivo anti-myeloma activity of 7,¹⁸ more recent efforts have focused on understanding the impact of other modifications at the alpha position on inhibitory activity.¹⁹ In particular, we have been interested in the potential for alpha modifications that could allow subsequent conjugation to another agent for more targeted delivery to the tumor cell. In this context, we describe here the preparation of a family of α-amino bisphosphonates based on our lead isoprenoid triazoles.

Past experience with click chemistry^{15, 20} indicated that disconnection to geranyl (8) or neryl (9) azide and propargyl alcohol (10) would rapidly provide the triazole alcohol 11 (Scheme 1). In fact, the click reaction proceeded smoothly giving the alcohol 11 as a ~3:1 mixture of olefin isomers due to the unavoidable [3,3] sigmatropic rearrangement of the allylic azide 8/9.²¹ Because such mixtures of olefin isomers have demonstrated biological activity,⁸ compound 11 was carried further in the reaction sequence as a mixture. Treatment of alcohol 11 with PBr₃ gave the corresponding bromide 12 which was converted to the nitriles 13 without isolation. When the mixture 13 was treated with diethyl phosphite and zinc chloride²² the expected amino bisphosphonate ester 14 was obtained, and subsequent hydrolysis gave the desired amino bisphosphonates 15. As explained in more detail below, the in vitro activity observed with this mixture encouraged preparation of the pure olefin isomers as well as the C₁₁ analogues.

Scheme 1.

Synthesis of amino bisphosphonate triazoles with C₁₀ isoprenoid chains. Structures for the mixed olefin isomers are shown in black, the pure trans olefins are shown in red, and the pure cis olefins are shown in blue.

Separation of the individual olefin isomers of the amino bisphosphonates 14 or 15 was not at all straightforward. As observed with other bisphosphonates, the polarity of the phosphonate groups simply overwhelms the chromatographic impact of the olefin stereochemistry. However, the nitrile 13 was the least polar compound encountered during these studies and very careful chromatography was successful in separating the major (presumably E isomer 16) and the minor (presumably Z isomer 18) olefins. Each of these isomers was treated separately with diethyl phosphite and zinc chloride to obtain the amino bisphosphonate esters 17 and 20. For the trans olefin isomer 17, the methyl carbon of the first isoprene unit has a relatively upfield shift due to the gamma effect of the first methylene carbon and is found ~17 ppm, while the C-4 methylene resonance is found at 39.7 ppm.²³ In the Z olefin isomer 20 these steric effects are reversed (Figure 2) which shifts the methylene resonance upfield to ~32 ppm, while the methyl resonance is found relatively downfield at ~23 ppm. These shifts allow assignment of the two olefin isomers. After the olefin stereochemistry was established, hydrolysis of the individual isomers gave the geranyl (18) and neryl (21) triazole amino bisphosphonates respectively.

Figure 2.

Aliphatic resonances in the ¹³C NMR spectra of the bisphosphonate esters 17 (trans) and 20 (cis).

To complete this set of bisphosphonates, it was necessary to prepare the homogeranyl and homoneryl analogues. Each could be prepared by a reaction sequence parallel to that used to obtain compounds 18 and 21 but starting with the homologated alcohols. Preparation of homogeraniol (22) from geraniol can be accomplished by any one of several short sequences.^24–26 After conversion of this alcohol to the corresponding bromide (23), reaction with sodium azide provided the desired azide 24. A click reaction with propargyl alcohol (10) gave the triazole alcohol 25. The corresponding bromide 26 was generated by reaction with PBr₃ and immediately converted to the desired nitrile 27 after minimal purification. Conversion to the bisphosphonate ester 28 was accomplished as described for the geranyl compound and final ester hydrolysis gave the targeted bisphosphonic acid 29.

The isomeric Z-olefin was prepared through a parallel reaction sequence that started with homonerol (Scheme 3, 30), itself prepared^24–26 via a short sequence from the Z-olefin nerol. A seven-step sequence parallel to that employed for preparation of the E-olefin 29 then was used to elaborate the homoallylic alcohol 30 to the key triazole 33, and then to the final target 37. The five new amino bisphosphonates, compounds 15, 18, 21, 29, and 37, then were tested for their activity in both enzyme and cell assays.

Scheme 3.

Synthesis of the Z-olefin 37.

Two methods were employed to evaluate effects on protein geranylgeranylation: 1) immunoblot analysis of unprenylated Rap1a (a substrate of geranylgeranyl transferase I) and, 2) measurement of intracellular lambda light chain levels via ELISA as a marker for disruption of Rab geranylgeranylation.⁹ Lovastatin, an inhibitor of HMG-CoA reductase, was used as a positive control as it disrupts all protein prenylation. As shown in Figure 3A-B, the mixture of C₁₀ isomers 15 disrupts protein geranylgeranylation in RPMI-8226 cells at concentrations as low as 0.5 μM. However, evaluation of the pure isomers revealed a striking impact of the olefin stereochemistry on activity, with the trans olefin 18 displaying no activity (including at concentrations as high as 10 μM, data not shown) and the isomeric cis olefin 21 showing activity at concentrations as low as 0.25 μM. Consistent with this finding, the homonerol-derived 37 showed superior cellular activity compared to the homogeraniol-derived 29 (Figure 3C-D). Interestingly, while the C₁₁ trans olefin 29 was more active relative to its C₁₀ analogue 18, the C₁₁ cis olefin 37 displayed equivalent activity to its C₁₀ analogue 21 in cell assays. However, enzyme assays utilizing recombinant GGDPS showed marked potency of both the C₁₁ compounds with IC₅₀ values of approximately 50 nM (Table 1), while the C₁₀ series had activity consistent with the cellular studies: no activity was observed with the trans olefin 18 and the mixture 15 was less potent than the pure cis olefin 21.

Figure 3. The α-amino bisphosphonates 15, 21, 29 and 37 disrupt protein geranylgeranylation in myeloma cells.

RPMI-8226 cells were incubated for 48 hours in the presence or absence of lovastatin (10 μM, Lov) or varying concentrations of the α-amino bisphosphonate triazoles 15, 18, 21, 29 or 37. A and C) Immunoblot analysis of unmodified Rap1a (uRap1a; antibody detects only unprenylated protein) and β-tubulin (a loading control) was performed. The gels are representative of three independent studies. B and D) Intracellular lambda light chain concentrations were determined via ELISA. Data are expressed as percentage of control (mean ± SD, n=3 independent experiments). The * denotes p<0.05 per t-test and compares treated cells to untreated control cells.

Table 1:

Summary of the bioassay results of the novel α-amino bisphosphonate triazoles and historical control GGDPS inhibitors

Compound	GGDPS IC501 (nM)	Cellular LEC2 (μM)
5 16	45 ± 16	0.03
6 17	125 ± 27	0.02
7 17	86 ± 22	0.025
15	3760 ± 270	0.5
18	>100,000	No activity up to 10 μM
21	270 +120	0.25
29	52.1 ± 3.9	1
37	49.4 ± 14.3	0.25

¹Data are expressed as mean ± SD (n=3 independent experiments)

²Cellular LEC (lowest effective concentration) is defined as the lowest concentration for which an unmodified Rap1a band is visible in the immunoblot and a statistically significant increase in intracellular lambda light chain is observed in the ELISA in RPMI-8226 cells.

In conclusion, the data obtained from this family of α-amino bisphosphonate triazoles further informs the complex structure-function relationship of the larger family of isoprenoid triazole bisphosphonate as GGDPS inhibitors. Themes that have emerged from this body of work thus far have included the importance of the alkyl chain length (with C₁₁ substituents being more active than C₁₀), the impact of the olefin stereochemistry (with the Z isomer generally being more active than the E isomer) and that modifications at the alpha position can have substantial impact on inhibitor potency as well as in vitro and in vivo activity.^{8, 15–18, 27–29} Notably, while the two C₁₁ amino bisphosphonates (29 and 37) were found to have the lowest IC₅₀ against GGDPS of any of the single isomer inhibitors prepared to date, their in vitro activity was much more modest. This suggests that mechanisms regulating cellular uptake of these inhibitors are impacted by the presence of the amino group relative to non-charged substituents such as the methyl group found in inhibitors 6 and 7. However, the advantage of the α-amino group over the methyl group is that it is amenable to conjugation to other agents, such as hyaluronic acid (HA).

We previously demonstrated that an ω-modified isoprenoid triazole bisphosphonate could be conjugated to HA and that this resulted in superior disruption of protein geranylgeranylation relative to the parent compound in two myeloma cell lines.²⁸ The glycosaminoglycan HA is an attractive candidate for conjugation because it is the native ligand for CD44, a receptor that is strongly expressed in multiple myeloma cell lines³⁰ and primary samples³¹ and because HA itself already is in clinical use.³² While our previously published HA-conjugated GGDPS inhibitor provided a proof of concept for myeloma-targeted delivery and subsequent intracellular drug release, our work also demonstrated that modification at the ω-position (e.g., compound 38) lessened GGDPS inhibitory activity.²⁸ Thus, our current efforts have focused on the development of analogues that can be conjugated and also display superior potency. The amino bisphosphonates 29 and 37 both display greater potency (IC₅₀’s ~50 nM) than compound 38 (IC₅₀ 670 nM) and have the potential to be linked to HA through the amine.

Given the size of HA, it is unlikely that direct linkage to the α-amino group would afford a compound where the bisphosphonate could reach the active site of the enzyme if the conjugate remains intact after cellular uptake. Cellular metabolism may cleave a direct amide linkage and release the active drug. However, if that cleavage is not facile, it may be possible to use a more advanced prodrug approach to exploit the selective HA uptake along with the improved drug activity, including the use of aminopeptidase-mediated cleavage.³³ Various prodrug forms of many phosphonates are known,³⁴ and prodrug activity often can be tuned to a specific cell type or subcellular localization, so extensive studies may still be necessary to explore fully the activity of conjugated derivatives of compounds 29 and 37. Furthermore structural biology approaches will be critical to better understand the manner by which these inhibitors engage the multiple potential binding sites within the oligomeric GGDPS enzyme.³⁵ Nonetheless, the activity of these new amino bisphosphonates affords a unique opportunity for the development of selective anti-myeloma agents and establishes the foundation for those studies.

Figure 1.

The linear isoprenoids FDP (1) and GGDP (2) and some inhibitors of GGDPS (3 – 7).

Scheme 2.

Synthesis of the E-olefin 29.