Mixed Aryl Phosphonate Prodrugs of a Butyrophilin Ligand

Abstract

Studies of aryl phosphonate derivatives of a butyrophilin 3A1 ligand have resulted in identification of a potent stimulant of Vγ9 Vδ2 T cells. This compound, a mixed ester bearing one pivaloyloxymethyl substituent and one 1-naphthyl ester displayed an EC₅₀ of 0.79 nM as a stimulant of T cell proliferation, and a 9.0 nM EC₅₀ in an assay designed to measure interferon gamma production. In both assays, this is the most potent butyrophilin ligand prodrug yet reported, and thus it should be a valuable tool for studies of T cell function. Furthermore, mixed aryl/acyloxyalkyl esters may represent a new class of phosphonate prodrugs with high efficacy.

Proliferation of Vγ9 Vδ2 T cells¹ is stimulated by the presence of small organophosphorus compounds² that bind to the signaling protein butyrophilin 3A1 (BTN3A1).³⁻⁶ The most potent natural ligand for this protein is (E)-4-hydroxy-3-methyl-but-2-enyl diphosphate (HMBPP, 1, Figure 1), which is the last intermediate in the biosynthesis of isoprenoids from 1-deoxy-d-xylulose 5-phosphate that is not found naturally in the mammalian route to isoprenoids from mevalonate.⁷ Synthetic analogues of HMBPP that stimulate T cell proliferation also have been reported, most bearing one (or more) C–P bond(s) in place of an O–P bond (or bonds) to endow greater metabolic stability.^6,8 For example, the phosphonate 2 has much greater metabolic stability than the phosphodiester 1, although it only displays cellular activity at much higher concentrations than HMBPP.^9,10 A more complete understanding of butyrophilin ligands may enhance the natural anticancer activities of this unique T cell subset.^11,12

Figure 1.

Most potent natural ligand for BTN3A1 (1), its phosphonate analogue C-HMBP (2), and its cell permeable prodrug POM₂-C-HMBP (3).

Several recent studies have explored the binding of HMBPP and its biologically active analogues to BTN3A1. Originally, binding was viewed as an extracellular event.¹³ However, more recently these small organophosphorus compounds have been shown to bind to an intracellular domain through a variety of structural, sequence swapping, and crystallographic techniques,¹⁴ as well as through isothermal titration calorimetry (ITC) and NMR studies.¹⁰ Given the growing body of evidence that the organophosphorus ligands must cross the cell membrane to demonstrate biological activity, the high charge-to-mass ratio of HMBPP may reduce its effectiveness when extracellularly dosed. While phosphonates such as compound 2 can improve metabolic stability, to improve cell permeability we have turned to preparation of phosphonate prodrugs. The first such example, the bispivaloyloxymethyl compound 3 (POM₂-C-HMBP), demonstrated an improved potency of 740-fold relative to the disodium salt 2.¹⁰ However, the question remains as to whether other prodrug forms might afford still more potent compounds. Furthermore, if prodrugs of the phosphonate ligand were to be advanced to animal studies, concerns have been expressed that pivalic acid may impact carnitine metabolism^15,16 and that the serum half-lives of bis-POM prodrugs may be limited.¹⁷ For these reasons, continued investigations of other prodrug forms are justified.

In theory, the simplest and most synthetically accessible prodrug form of a biologically active phosphonate might be a diester of a simple alcohol such as methanol or ethanol. However, despite the discovery of organophosphorus hydrolases in some bacteria,¹⁸ there is little evidence to support metabolic cleavage of dialkyl phosphonate esters of small alcohols in mammalian systems,¹⁶ and the dimethyl esters may be particularly stable.¹⁹ However, aryl esters may have more promise as prodrugs,²⁰⁻²² and both phenyl and naphthyl systems are frequent components of phosphoramidate prodrugs.¹⁶ To determine the ability of different aryl esters of phosphonate 2 to function as prodrugs, we have prepared and determined the biological activity of a small set of derivatives and report here our findings.

Synthesis of the target compounds began with conversion of dimethyl homoprenylphosphonate (4) to the phosphonic acid chloride 5 (Scheme 1).²³ Given its anticipated reactivity, the acid chloride was employed in reactions with phenol or 1-naphthol after only minimal purification. Nonetheless, the expected esters 6a and 6b were obtained in good yields.

Scheme 1. Synthesis of Homoprenylphosphonate Derivatives.

To explore the lability of these aryl groups in this cell system, three derivatives of both ester 6a and ester 6b were prepared to obtain compounds that might release a biologically active form of phosphonate 2 within cells. Based on our prior studies, we hypothesized that the mixed aryl/POM diesters 8a/8b would readily cross the cell membrane and undergo at least POM cleavage once inside the cell. The mixed aryl/methyl diesters 9a/9b also would be expected to cross the cell membrane readily, with the methyl ester more stable to metabolic cleavage. The salt forms of the aryl esters 10a/10b would be expected to have more difficulty crossing the cell membrane due to a negative charge at physiological pH.

The desired compounds were available through short synthetic sequences. Reaction of the esters 6a and 6b with pivaloyloxymethyl chloride (POMCl)²⁴ gave the corresponding racemic mixed diesters 7a and 7b in modest yields. Installation of the allylic alcohol through selenium dioxide oxidation, followed by brief treatment with sodium borohydride to reduce any aldehyde formed, also proceeded in low yield. However, the desired products 8a and 8b were obtained in amounts sufficient for the necessary bioassays. Both compounds were obtained as racemates, which was sufficient to establish whether they demonstrate biological activity.

The parent esters 6a and 6b also served as the precursors to the other pairs of target compounds. Direct selenium dioxide oxidation gave the expected allylic alcohols 9a and 9b in low yields. Fortunately treatment of the resulting esters 9a and 9b with sodium iodide in acetonitrile²⁵ gave nearly quantitative yields of the salts 10a and 10b, with no evidence of cleavage of the aryl groups. In this way the six compounds 8a/b, 9a/b, and 10a/b were obtained for evaluation of their biological activity.

Each of the six new compounds was tested first for the ability to stimulate expansion of Vγ9 Vδ2 T cells (Table 1). In this assay, cells are exposed to test compounds for 72 h to maximize cellular uptake and subsequent T cell expansion. The phenyl/methyl-protected phosphonate 9a exhibited weak activity with an EC₅₀ in the mid micromolar range. Even weak activity of this compound was surprising, as typically dimethyl protected phosphonates are inactive.^10,26 Synthetic deprotection of the methyl group gave compound 10a, which showed increased cellular activity, with an EC₅₀ value near 1 μM. Based on prior results, we suspected that cell permeability would remain a barrier to the monosalt form 10a, and therefore, we assessed the phenyl/POM protected form 8a. This compound again displayed significant potency gains, with activity in the low nanomolar range.

Table 1. Activity for Expansion of Vγ9 Vδ2 T Cells from Human PBMC.

			fold differencea
compd	LogP	EC50 [μM] (95% CI)	vs 2	vs 3
2 Na/Na10	–0.24	4.0	NA	NA
3 POM/POM10	3.42	0.0054	740	NA
8a Phe/POM (n = 3)	3.56	0.014 (0.0040 to 0.051)	290	ND
9a Phe/Me (n = 3)	2.01	33 (7.9 to 140)	ND	ND
10a Phe/Na (n = 3)	1.73	0.87 (0.42 to 1.8)	4.6	ND
8b Nap/POM (n = 3)	4.75	0.00079 (0.00060 to 0.0010)	5100	6.8
9b Nap/Me (n = 3)	3.19	5.5 (0.029 to 1000)	ND	ND
10b Nap/Na (n = 3)	2.92	0.61 (0.14 to 2.6)	6.6	ND

^aND = not determined. NA = not applicable.

While the activity of the phenyl/POM protected compound 8a was impressive, it was slightly less potent than the bis-POM analogue (3, POM₂-C-HMBP), even though both compounds exhibit similar calculated LogP values and presumed cell permeability. To explore further this relationship, we hypothesized that the naphthyl group would afford further potency gains because it is more hydrophobic, and given that naphthol is more acidic than phenol by approximately a factor of 4,²⁷ it also could serve as a better leaving group. Indeed, all three forms, the naphthyl/methyl (9b), naphthyl/salt (10b), and naphthyl/POM (8b) forms were more potent than the analogous phenyl-protected compounds. Surprisingly, the naphthyl/POM form of this compound (8b) is the most potent synthetic phosphoantigen we have identified to date, with an EC₅₀ of 790 pM, and comparable to the natural butyrophilin ligand HMBPP (510 pM).¹⁰ Although a phenyl/POM derivative of an acyclic nucleoside phosphonate has been reported in the patent literature,²⁸ and a phenyl/acyloxy derivative of methyl phosphonate was reported in a recent patent as a prodrug form of a carboxylic acid,²⁹ the naphthyl/POM phosphonate protecting strategy as shown in compound 8b has not been reported prior to this work.

Because it remained a possibility that the phenyl or naphthyl groups are present in cell-active forms of the parent compound, we assessed the sodium salts of these two compounds (10a and 10b) for their ability to bind to the molecular target, BTN3A1, by ITC. In these assays, no binding was observed (Figure S1). Additionally, the more potent compound 10b was tested and found to be unable to compete with HMBPP for binding to BTN3A1, even at a concentration that was 100-fold higher (Table S1). This lack of binding of protected phosphonates is consistent with our previous studies on the tris-POM versions of phosphinophosphonates, which were also unable to bind to the protein in their prodrug forms.²³ In our view, this suggests that both the phenyl and naphthyl groups are susceptible to cellular hydrolysis resulting in metabolic conversion to the phosphonate dianion 2.

To assess further the unique cellular potency of the naphthyl/POM compound 8b, we compared it to HMBPP (1) and POM₂-C-HMBP (3), our previous most-potent synthetic prodrug, in an ELISA assay of T cell interferon gamma production. In this assay, K562 cells were pre-exposed for 4 h to the test compound, washed, then cocultured for 20 h with T cells to stimulate cytokine production by the T cells. This exposure time is intentionally limiting to cellular uptake and meant to maximize differences between cell uptake by diffusion and by transport.³⁰ Here (Table 2), as expected, we found that POM₂-C-HMBP (3) was significantly more potent than HMBPP. Again, the naphthyl/POM compound 8b provided further gains in potency relative to POM₂-C-HMBP (3) and retained low nanomolar potency even when uptake times were restricted.

Table 2. Activity for Stimulation of Purified Vγ9 Vδ2 T Cells To Produce Interferon Gamma in Response to K562 Cells Preloaded with Test Compounds.

		fold differencea
compd	EC50 [μM] (95% CI)	vs 1 (HMBPP)	vs 3
HMBPP (1)	5.1 (3.7 to 7.1)	NA	NA
POM/POM (3) (n = 3)	0.024 (0.019 to 0.031)	210	NA
Nap/POM (8b) (n = 3)	0.0090 (0.0075 to 0.011)	570	2.7

^aNA = not applicable.

In conclusion, these studies have shown that aryl phosphonate prodrugs of the homoprenyl phosphonate 1 effectively stimulate T cell proliferation. In the best case, compound 8b demonstrated an EC₅₀ in the high picomolar range. Furthermore, compound 8b is ∼570-fold more potent than HMBPP (1) in an assay designed to measure the ability of a compound to stimulate interferon gamma production. These results clearly encourage further studies of prodrug forms of the phosphonates that serve as butyrophilin ligands to identify still more effective compounds. In the meantime, these new compounds represent valuable tools for studies of Vγ9 Vδ2 T cell proliferation and ligand binding to the signaling protein BTN3A1.

Finally, while the aryl phosphonate protecting strategy has been successfully incorporated into some phosphoramidate prodrugs, including the clinical compound Sofosbuvir,³¹ the potency gains on the acyloxyalkyl scaffold described herein suggest this protecting group combination may be a viable option for efficient cellular delivery of other phosphonate-containing payloads.

Biographies

Andrew J. Wiemer earned his B.S. from the University of Notre Dame. He received his Ph.D. from the University of Iowa in 2008 under the guidance of Professor Raymond J. Hohl and performed postdoctoral research at the University of Wisconsin–Madison in the laboratory of Professor Anna Huttenlocher. Andrew joined the faculty at the University of Connecticut as an Assistant Professor in 2012. He has published over 30 articles. His research interests are in medicinal chemistry and chemical biology, with a particular focus on small molecules that affect anti-cancer immunity.

David F. Wiemer is a Midwestern native who received a B.S. degree from Marquette, his Ph.D. from the University of Illinois (under Nelson Leonard), and an NIH postdoctoral fellowship at Cornell University (with Jerrold Meinwald). At the University of Iowa he holds appointments as F. Wendell Miller Professor of Chemistry and Professor of Pharmacology. His research is focused on the design and synthesis of organophosphorus compounds that mimic the intermediates of isoprenoid biosynthesis, and he collaborates to determine the biomedical potential of these compounds. He has nearly 200 publications and is a fellow of both the AAAS and the ACS.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.7b00245.

• Experimental procedures for the synthetic chemistry, bioassay protocols, and NMR spectra (PDF)

Author Contributions

The manuscript was written through contributions of all authors and all authors have given approval to the final version of the manuscript.

The authors declare the following competing financial interest(s): A.J.W. and D.F.W. own shares in Terpenoid Therapeutics, Inc. The current work did not involve the company. The other authors have no financial conflicts of interest.