N-(Pivaloyloxy)alkoxy-carbonyl Prodrugs of the Glutamine Antagonist 6-Diazo-5-oxo-L-norleucine (DON) as a Potential Treatment for HIV Associated Neurocognitive Disorders

Abstract

Aberrant excitatory neurotransmission associated with overproduction of glutamate has been implicated in the development of HIV-associated neurocognitive disorders (HAND). The glutamine antagonist 6-diazo-5-oxo-L-norleucine (DON, 14) attenuates glutamate synthesis in HIV-infected microglia/macrophages, offering therapeutic potential for HAND. We show that 14 prevents manifestation of spatial memory deficits in chimeric EcoHIV-infected mice, a model of HAND. 14 is not clinically available, however, because its development was hampered by peripheral toxicities. We describe the synthesis of several substituted N-(pivaloyloxy)alkoxy-carbonyl prodrugs of 14 designed to circulate inert in plasma and be taken up and biotransformed to 14 in the brain. The lead prodrug, isopropyl 6-diazo-5-oxo-2-(((phenyl(pivaloyloxy)methoxy)-carbonyl)amino)hexanoate (13d), was stable in swine and human plasma but liberated 14 in swine brain homogenate. When dosed systemically in swine, 13d provided a 15-fold enhanced CSF-to-plasma ratio and a 9-fold enhanced brain-to-plasma ratio relative to 14, opening a possible clinical path for the treatment of HAND.

Graphical abstract

INTRODUCTION

HIV-associated neurocognitive disorders (HAND) remain one of the central health issues in patients with chronic HIV infection¹ despite viremic control by combined antiretroviral therapy (cART).² While broad use of cART has markedly reduced the prevalence of HIV-associated dementia (HAD), it has had limited effects on milder forms of HAND which now represent the majority of new cases of HIV neurological disease.^3–5

Although the mechanisms of HAND pathogenesis are unclear, accumulating evidence suggests that HAND is associated with aberrant glutamate metabolism in the central nervous system (CNS) culminating in elevated extracellular glutamate and disrupted excitatory neurotransmission.^6,7 In cultured human macrophages and microglia, HIV infection induces elevated expression of the glutamate-synthesizing enzyme glutaminase which directly contributes to the toxic overproduction of glutamate.^6,8–19 Activated microglia and macrophages also release viral proteins and pro-inflammatory cytokines^6,7 that independently promote further glutamate synthesis and release²⁰ and inhibit glutamate reuptake in astrocytes and neurons.⁶ These cumulative insults result in impaired glutamate-dependent synaptic function^21–28 and cognition.^25–28

Studies of HAND patients corroborate preclinical findings of disrupted glutamate metabolism in the CNS. Recent examination of cART-treated HIV patients revealed increased cerebrospinal fluid (CSF) glutamate in patients with HAND compared to those without neurocognitive impairment.²⁹ Several magnetic resonance spectroscopy (MRS) studies have also shown dysregulated glutamate homeostasis in HIV patients that correlate with neurocognitive impairment including alterations in white and gray matter glutamate and glutamine.^30–35 In addition, microarray analyses in post-mortem brain samples from HAND patients show frontal cortical alterations in the expression of genes critical for the regulation of glutamatergic signaling and glutamate-dependent synaptic plasticity, including glutaminase.^18,36–38

Collectively, these studies suggest that interventions aimed at normalizing glutamate homeostasis offer therapeutic potential for HAND treatment. The glutamine antagonist 6-diazo-5-oxo-L-norleucine (DON, 14), which is known to inhibit glutaminase, has been shown to block glutamate over-production induced by HIV infection or immune challenge, and to mitigate excitotoxic neuronal damage both in vitro and in vivo,^13,39–41 but has not yet been tested in an animal model of HAND.

In the present work, we tested the ability of 14 to prevent cognitive decline in EcoHIV-infected wild-type mice which serve as a model of HAND.^42,43 EcoHIV is a chimeric HIV in which the coding region for envelope protein gp120 has been replaced with gp80 from murine leukemia virus, rendering EcoHIV capable of infecting conventional mice.⁴² EcoHIV infection results in expression of HIV viral proteins in relevant tissues including microglial cells in the brain⁴³ and recapitulates many of the neuropathological features of HAND, bolstering the use of these mice as a model system for testing possible HAND therapeutics.⁴² We report that when 14 was administered to the mice prior to and during EcoHIV infection, it was able to completely prevent their cognitive decline.

Although exciting, the clinical utility of 14 is limited by its peripheral toxicities mediated primarily by the gastrointestinal (GI) system,^44–46 which is known to be highly glutamine-utilizing. Recently, we described the synthesis of several N-(pivaloyloxy)alkoxy-carbonyl 14 prodrugs designed to enhance CNS penetration and brain delivery of 14 while reducing peripheral exposure and thus GI toxicity.⁴⁷ Systemic administration in nonhuman primate of one such prodrug, compound 13b, with isopropyl ester on the carboxylate and N-(pivaloyloxy)ethoxy-carbonyl on the amine, resulted in about 10-fold enhancement in the CSF-to-plasma ratio of 14 exposure relative to administration of equimolar 14.⁴⁷ These findings provided the first evidence that a prodrug strategy could significantly alter the tissue distribution of 14.

In the current study, we hypothesized that modification of the N-(pivaloyloxy)methoxy-carbonyl pro-moiety with additional steric bulk on the methylene bridge could slow peripheral bioconversion of 14 prodrugs and confer further improvement in CNS penetration, as has been shown for other prodrugs utilizing carboxylesterase-mediated hydrolyses.^48–51 In addition to increasing steric hindrance and metabolic stability, our modification approach could also enhance lipophilicity.^52,53 The new 14 prodrugs, compounds 13b–e, with addition of methyl, isopropyl, phenyl, and dimethyl groups to the N-(pivaloyloxy)methoxy-carbonyl pro-moiety, respectively, had substantially increased calculated partition coefficients (cLogP) versus 14 (1.50, 2.42, 2.75, and 1.81 vs –2.50, respectively). However, when evaluated in mouse plasma, all of the prodrugs were rapidly and completely metabolized to 14, precluding attempts to meaningfully differentiate their efficacy/toxicity from 14 in the EcoHIV murine model. Importantly, this conundrum is not uncommon in prodrug discovery, where increased metabolism in rodents is well documented and higher species are necessary to mimic human metabolism.^54,55 Thus, these prodrugs were subsequently tested for stability in human plasma as well as swine, a model organism that closely recapitulates human drug metabolism and pharmacokinetics.⁵⁶ Although the N-(pivaloyloxy)methoxy-carbonyl derivative of 14, compound 13a, was also labile in human and swine plasma, all of the substituted analogues, compounds 13b–e, were found to be metabolically stable. Furthermore, when incubated in homogenates of swine brain, the target tissue for 14 release, compounds 13b and 13d were readily converted. These compounds were then tested in vivo in swine. Relative to administration of equimolar 14, systemic administration of 13b and 13d resulted in lower 14 plasma exposure, higher CSF exposure, and a greater than 7- and 15-fold increased CSF-to-plasma ratio, respectively. Having shown the best profile, 13d and 14 were subsequently assessed in terminal swine studies in which brain levels were directly measured. 13d afforded a 9-fold enhancement in brain-to-plasma ratio relative to equimolar 14.

CHEMISTRY

As shown in Scheme 1, the intermediates 4a–c were synthesized by a three-step procedure starting from the appropriate 1-chloroalkyl carbonochloridate (1a–c), which was reacted with ethanthiol in the presence of triethylamine to yield corresponding O-(1-chloroalkyl)-S-ethyl carbonothioates (2a–c). These intermediates were transformed to 1-(((ethylthio)carbonyl)oxy)alkyl pivalate derivatives (3a–c) by reaction with pivalic acid in the presence of base (DIPEA). Oxidation of the thioester moiety by peracetic acid followed by reaction with N-hydroxysuccinimide^57,58 provided 4a–c in a yield of 56–64% after column chromatography (total over three steps). The attempt to use the same methodology for phenyl analogue 13d in the last step resulted in a complex mixture, likely due to reactivity of the corresponding benzyl cation intermediate. The attempt to prepare the analogous intermediate with a phenyl was unsuccessful. Although it was not confirmed, this failure could be due to the oxidation of the benzoacetal group by peracetic acid in the third step.

Scheme 1. Synthesis of N-Hydroxysuccinimide Esters of Pivaloyloxy-alkoxycarbonates Intermediates 4a–c^a.

^aReagents and conditions: (a) EtSH, Et₃N, Et₂O, 0 °C to rt, 17–25 h; (b) pivalic acid, DIPEA, 60–70 °C, 20–70 h; (c) N-hydroxysuccinimide, peracetic acid, DCM, 0 °C to rt, 3 h.

We thus chose to explore the para-nitrophenoxy derivative 8, which can be synthesized via a route devoid of an oxidant as shown in Scheme 2. First, the chloro(phenyl)methyl carbonochloridate (6) was prepared by reaction of benzaldehyde (5) with triphosgene in the presence of base (pyridine). Intermediate 6 was reacted with 4-nitrophenol to give the chloro(phenyl)methyl (4-nitrophenyl) carbonate (7), which was then subjected to reaction with freshly prepared mercury-(II) salt of pivalic acid to yield analogue 8. Pivalate 8 was then used to prepare 13d (Scheme 4). Despite several attempts with different pivalates (potassium, silver, palladium) we were not successful in eliminating the use of mercury in this step.

Scheme 2. Synthesis of 4-Nitrophenyl Esters of Pivaloyloxy-alkoxycarbonate Intermediate 8^a.

^aReagents and conditions: (a) triphosgene, pyridine, Et₂O, –20 °C to rt, 20 h; (b) 4-nitrophenol, pyridine, DCM, 0 °C to rt, 2 h; (c) Hg(OPiv)₂, DCM, rt, 16 h.

Scheme 4. Synthesis of 14 Prodrugs 13a –e^a.

^aReagents and conditions: (a) DCM or DMF, 0 °C to rt, 2–5 h.

As shown in Scheme 3, compound 11 was prepared in an analogous way. 4-Nitrophenyl prop-1-en-2-yl carbonate (9) was prepared from commercially available 2-propenyl chloroformate as previously reported⁵⁹ and reacted with hydrogen chloride. The obtained 2-chloropropan-2-yl (4-nitrophenyl) (10) was then reacted with mercury(II) pivalate to yield the desired intermediate 11, which was then used to prepare analogue 13e.

Scheme 3. Synthesis of Intermediate 11^a.

^aReagents and conditions: (a) 4 M HCl, dioxane, rt, 21 h; (b) Hg(OPiv)₂, DCM, rt, 20 h.

As shown in Scheme 4, the prodrugs 13a–e were prepared by reaction of 14 isopropyl ester (12) with 4a–c, 8, or 11, respectively, in dichloromethane (DCM) or dimethylformamide (DMF) at 0 °C or room temperature. Compound 12 was prepared as previously described in detail with yield and purity in agreement with the published data.⁴⁷ The compounds 13a–e were obtained in 40–83% yield after purification by liquid chromatography (LC). To improve yields of 13d and 13e, an excess of compound 12⁴⁷ and longer reaction time were necessary (see Experimental Section).

RESULTS AND DISCUSSION

Compound 14 Prevented Cognitive Decline, a Major Manifestation of HAND, in EcoHIV-Infected Mice

Similar to HAND patients,⁶⁰ mice inoculated with EcoHIV exhibited impaired spatial learning and memory as measured by radial arm water maze (RAWM) 30 days postinfection (Figure 1). EcoHIV infection resulted in a significant increase in the number of errors (Figure 1A) and latency to escape (Figure 1B) onto a hidden platform in the maze. 14 treatment (1 mg/kg, ip, qod) beginning 1 day prior to EcoHIV or sham inoculation and continued throughout the infection period and RAWM testing, fully normalized cognitive performance as measured by both errors (main effect of treatment [F(3,140) = 261.8, p < 0.0001], trial [F(4,140) = 146.2, p < 0.0001], interaction [F(12,140) = 1.93, p = 0.0355]) and latency to escape (main effect of treatment [F(3,140) = 37.00, p < 0.0001], trial [F(4,140) = 56.03, p < 0.0001], interaction [F(12,140) = 1.832, p = 0.0484]). Neither EcoHIV infection nor 14 treatment affected latency to escape to a visible platform (Figure 1C; main effect of trial [F(3,112) = 34.44, p < 0.0001]), indicating no impairment in visual or motor function. Additionally, no mice exhibited overt signs of toxicity (i.e., diarrhea, weight loss). The EcoHIV-infected mice exhibited measurable viral loads in the spleen and brain (Figure 1D). 14 treatment actually caused an increase in peripheral viral load (t(14) = 2.58, p = 0.022), likely due to the well described 14-mediated inhibition of T cell activity/proliferation,^61,62 which is known to be required for endogenous suppression of EcoHIV replication.⁴³ Therefore, despite enabling a modest increase in EcoHIV replication, 14 still prevented cognitive decline, suggesting its mechanism of action to be secondary to the infection itself. These findings suggest that glutaminase inhibition through 14 delivery to the CNS prior to or during HIV infection may prevent the development of cognitive impairment in HAND patients.

Figure 1.

DON (14) prevented cognitive decline in the EcoHIV model of HAND. DON (14) treatment (1 mg/kg, ip) was begun prior to EcoHIV inoculation and continued every other day throughout the 30 day infection period and during radial arm water maze (RAWM) testing. 14 significantly attenuated spatial learning and memory deficits in the RAWM as measured by (A) number of errors across learning trials (LT) 1–4 and the retention trial (RT) and (B) latency to escape to a hidden platform relative to sham-inoculated control mice. 14 had no effect on (C) RAWM escape latency to a visible platform. 14 treatment also caused (D) a slight increase in EcoHIV viral load as measured by DNA copies in the spleen but had no effect on viral load in the brain. Behavioral comparison conducted by two-way ANOVA, posthoc comparison by Tukey’s test; ****p < 0.0001, EcoHIV/Veh vs Sham/Veh, EcoHIV/DON, and Sham/DON. Viral load comparison conducted by t test, *p < 0.05, n = 8/group.

N-(Pivaloyloxy)alkoxy-carbonyl Prodrugs of 14 Were Synthesized with Improved Lipophilicity

Despite robust efficacy in the EcoHIV model of HAND, the GI toxicity induced by 14 in humans has hampered its clinical development.^44–46 Previous efforts by our group to develop more tolerable prodrugs of 14 yielded compound 13a (Table 1) through coverage of 14’s carboxylate and amine groups with isopropyl ester and a primary N-(pivaloyloxy)methoxy-carbonyl group, respectively.⁴⁷ We rationalized that additional bulky, lipophilic substituents would increase cLogP and further improve CNS penetration. Thus, methyl-, isopropyl-, phenyl-, and dimethyl-analogues of 13a (compounds 13b–e) were synthesized as described above. cLogP for each of these prodrugs was incrementally and substantially increased relative to 14 (Table 1). We then initiated testing of their ex vivo metabolic stability.

Table 1.

Lipophilicity and Stability of N-(Pivaloyloxy)alkoxy-carbonyl Prodrugs of DON (14)

Cmpd	R1	R2	cLogP	Plasma Stability (% remainin at 60 min)			Brain Stability (% remaining at 60 min)
				Mouse	Human	Swine	Swine
14	$\mathrm{H}$	$\mathrm{H}$	−2.50	–	–	–	–
13a			1.19	0	947	–	–
13b			1.50	0	9147	86±5	58±1
13c			2.42	0	100±7	100±3	31±1
13d			2.75	0	88±5	83±1	26±0
13e			1.81	0	93±5	85±3	87±2

N-(Pivaloyloxy)alkoxy-carbonyl Prodrugs of 14 Were Completely Unstable in Mouse Plasma, Restricting Their Evaluation in Mouse Models

All prodrugs 13a–e were found to be completely metabolized during a 60 min incubation in mouse plasma (Table 1). A representative chromatogram for the rapid metabolism of 13d in mouse plasma is provided in Supporting Information, Figure S1, showing complete conversion to 14 within 10 min. The rapid metabolism of the 14 prodrugs precluded the examination of their efficacy and toxicity in mice as they would be immediately converted to 14. As mentioned above, enhanced metabolism in rodents is a common issue with prodrugs of various classes,^54,55 necessitating the use of higher species, such as swine, that more closely model human metabolism.⁵⁶

N-(Pivaloyloxy)alkoxy-carbonyl Prodrugs of 14 Were Stable in Human and Swine Plasma but Labile in Brain Homogenate

Consistent with previous reports of the metabolic lability of N-(pivaloyloxy)methoxy-carbonyl group-containing prodrugs,^47–51,63 compound 13a was found to be rapidly metabolized during a 60 min incubation in human plasma (Table 1). In contrast, we found that modification of 13a with methyl-, isopropyl-, phenyl-, and dimethyl- substitutions (13b–e) resulted in a significant improvement in human plasma stability (Table 1). As our goal was to evaluate these prodrugs in an animal model which mimicked human metabolism, we next tested the stability of compounds 13b–e in swine plasma. We found that all of the prodrugs, similar to human plasma, were stable in swine plasma (Table 1). Given that the target compartment for 14 activity in HAND is the brain, 13b–e were then tested for metabolic lability in swine brain homogenate. Compounds 13c and 13d were found to be readily biotransformed in swine brain homogenate, whereas 13b was moderately labile and 13e was mostly stable (Figure 2). Swine brain homogenate half-lives were calculated for each compound and found to be significantly different from each other [F(3,8) = 31.42, p < 0.0001].

Figure 2.

N-(Pivaloyloxy)alkoxy-carbonyl prodrugs of DON (14) showed differential rates of metabolism in swine brain homogenate. Compounds 13b–e (10 μM) were spiked into swine brain homogenate (1% w/v); concentrations of the prodrug remaining were measured by LC-MS/MS at 0, 30, or 60 min post incubation. The prodrugs were metabolized at varying rates, with 13c and 13d showing the highest lability. Data are depicted as mean ± SEM. Half-lives (t_1/2) compared by one-way ANOVA, p < 0.0001, n = 3/group.

When Tested in Swine, Compounds 13b and 13d Resulted in Enhanced 14 CSF-to-Plasma Ratios; Compound 13d Showed an Optimal Profile

To determine if the ex vivo metabolism data translated in vivo, 14, 13b, and 13d were selected for pharmacokinetic evaluation in swine. Consistent with their observed metabolic stability, iv infusion of 13b and 13d (1.6 mg/kg 14 equivalent dose) resulted in 3–5-fold lower 14 plasma exposures relative to an equimolar dose of 14 (Figure 3A). Plasma AUC_0–t for 14, 13b, and 13d were 29.9, 8.00, and 5.70 nmol·h/mL, respectively. The opposite trend occurred in CSF, where the prodrugs delivered substantially higher amounts of 14 to the CSF (Figure 3B; Table 2), resulting in significantly increased CSF-to-plasma ratios (Figure 3C; Table 3). The improvement over 14 in CNS penetration correlated with the lipophilicity of each prodrug. The cLogP for 13d was nearly twice that calculated for 13b (2.75 vs 1.50) and was associated with a doubling of the improvement in CSF-to-plasma ratio in vivo (approximately 15-fold increase for 13d vs 7-fold increase for 13b relative to equimolar 14). These experiments confirm that POM-based ester prodrugs substantially improve CNS delivery of 14 and support the hypothesis that iterative derivatization of this promoiety with sterically hindering, lipophilic residues promotes better CNS targeting. It should be noted that the 13b diastereomers were separable by standard column chromatography,⁴⁷ whereas the 13d diastereomers were not readily separable by the same method and thus were administered as a mixture. This may not influence the clinical relevance of these findings, as human plasma stabilities of the 13b diastereomers were comparable (91% vs 89% remaining at 60 min for 13b-1 and 13b-2, respectively).⁴⁷ However, as 13d moves toward clinical development, stereochemical purity will be prioritized.

Figure 3.

In vivo pharmacokinetics of DON following iv administration of DON (14), 13b, and 13d in swine plasma and CSF. DON (14, 1.6 mg/kg, iv) or an equivalent dose of either 13b or 13d were administered to swine. Plasma (0–60 min) and CSF (60 min) concentrations of 14 were evaluated via LC-MS/MS. Relative to 14, compounds 13b or 13d delivered (A) lower 14 plasma exposure and (B) higher 14 CSF concentrations, resulting in (C) more than 7-fold or 15-fold enhanced CSF:plasma ratio at 60 min postadministration, respectively.

Table 2.

Total and Free DON (14) Concentrations 60 min Post-Administration of 14 or 13d in Swine (1.6 mg/kg Equivalent, iv)

	brain (nmol/g)		plasma (nmol/mL)		CSF (nmol/mL)
	total	free	total	free	total
14	5.15 ± 1.40	0.464 ± 0.126	41.2 ± 12.2	32.5 ± 9.64	0.252
13d	5.90 ± 0.590	0.531 ± 0.053	4.88 ± 1.50	3.86 ± 1.19	1.10

Table 3.

Total and Free DON (14) Ratios 60 min Post-Administration of 14 or 13d in Swine (1.6 mg/kg Equivalent, iv)

	brain:plasma		CSF:plasma		CSF:brain
	total:total	free:free	total:total	total:free	total:free
14	0.148 ± 0.078	0.017 ± 0.001	0.005	0.006	0.543
13d	1.38 ± 0.544	0.157 ± 0.062	0.069	0.087	2.07

In Terminal Swine Studies, Compound 13d Resulted in Enhanced 14 Brain-to-Plasma Ratio

Because CSF is not always an accurate surrogate for brain concentrations, we conducted terminal studies in swine with the lead compound 13d where plasma and brain tissue were collected. At 60 min postinfusion of 13d (1.6 mg/kg 14 equivalent dose), 14 delivery to the plasma was substantially lower than equimolar 14 (Figure 4A) while brain levels were comparable (Figure 4B; Table 2). This resulted in a 9-fold enhancement in the brain-to-plasma ratio of 13d relative to equimolar 14 (Figure 4C), similar to its enhanced CSF-to-plasma ratio (Table 3). Given that the 14 CSF levels were significantly lower than that observed in brain tissues, we next conducted protein binding studies in an attempt to elucidate the reason for these differences. Plasma protein binding of 14 was low (F_b = 21%), consistent with previous reports.⁴⁶ However, higher brain protein binding of 14 was observed (F_b= 91%), providing a possible explanation for the CSF and brain level discrepancies. On the basis of this value, free levels of 14 in the swine brain were calculated to be 0.531 nmol/g at 60 min postadministration of 13d, closer to observed levels of 14 delivered to the CSF by 13d at the same time point (~1.0 nmol/mL; Table 2).

Figure 4.

In vivo pharmacokinetics of DON following iv administration of DON (14) and 13d in swine plasma and brain. DON (14, 1.6 mg/kg, iv) or an equivalent dose of 13d was administered to swine. Swine were sacrificed, and plasma and brain (60 min) concentrations of 14 were evaluated via LC-MS/MS. Relative to 14, compound 13d delivered (A) lower 14 plasma concentrations and (B) comparable 14 brain concentrations, resulting in (C) a 9-fold enhanced brain:plasma ratio at 60 min postadministration. Data are depicted as mean ± SEM (n = 2/group). Free levels of 14 in the swine brain (F_b = 91%) were calculated to be 0.46 and 0.53 nmol/g following 14 and 13d administration, respectively.

CONCLUSION

14 was robustly efficacious in preventing cognitive decline in EcoHIV-infected mice used as a model of HAND. However, 14 is not suitable for clinical use due to its dose-limiting peripheral toxicities. To translate these findings to the clinic, we describe the synthesis of several substituted N-(pivaloyloxy)alkoxy-carbonyl prodrugs of 14 that improve CNS penetration. Rational design of 14 prodrugs by addition of lipophilic substituents to the N-(pivaloyloxy)alkoxy-carbonyl pro-moiety yielded sterically hindered compounds with increased metabolic stability in swine and human plasma, favorable conversion in brain homogenate, and substantially improved CNS delivery in vivo in swine, exemplified by compound 13d. Metabolic stability of these prodrugs in human plasma was very similar to that observed in swine, suggesting strong translational potential for this strategy. This approach may therefore enable the clinical use of brain glutaminase inhibition for the treatment of HAND and other neuropsychiatric disorders characterized by aberrant glutamate metabolism.

EXPERIMENTAL SECTION

The commercially available HPLC grade acetonitrile, catalysts, and reagent grade materials were used as received. TLC was performed on Silica gel 60 F254-coated aluminum sheets (Merck), and spots were detected by the solution of Ce(SO₄)₂.4H₂O (1%) and H₃P(Mo₃O₁₀)₄ (2%) in sulfuric acid (10%). Flash chromatography was performed on Silica Gel 60 (0.040–0.063 mm, Fluka) or on Biotage KP-C18-HS or KP-Sil SNAP cartridges using the Isolera One HPFC system (Biotage, Inc.). All chemicals were purchased from Sigma-Aldrich or TCI and were used without further purification. The ¹H NMR spectra were measured at 400.1 MHz and ¹³C NMR spectra at 100.8 MHz. For standardization of ¹H NMR spectra, the internal signal of TMS (δ 0.0, CDCl₃) or residual signals of CDCl₃ (δ 7.26) were used. In the case of ¹³C NMR spectra, the residual signal of CDCl₃ (δ 77.00) was used. The chemical shifts are given in δ scale; the coupling constants J are given in Hz. The ESI mass spectra were recorded using ZQ micromass mass spectrometer (Waters) equipped with an ESCi multimode ion source and controlled by MassLynx software. Alternatively, the low resolution ESI mass spectra were recorded using a quadrupole orthogonal acceleration time-of-flight tandem mass spectrometer (Q-Tof Micro, Waters) and high resolution ESI mass spectra using a hybrid FT mass spectrometer combining a linear ion trap MS and the Orbitrap mass analyzer (LTQ Orbitrap XL, Thermo Fisher Scientific). The conditions were optimized for suitable ionization in the ESI Orbitrap source (sheat gas flow rate 35 au, aux gas flow rate 10 au of nitrogen, source voltage 4.3 kV, capillary voltage 40 V, capillary temperature 275 °C, tube lens voltage 155 V). The samples were dissolved in methanol and applied by direct injection. The purity of all compounds subjected to biological testing was established using HPLC (Jasco Inc.) equipped with a Reprosil 100 C18, 5 μm, 250 mm × 4 mm column. The analysis was performed using a gradient of 2% CH₃CN/98% H₂O with 0.1% TFA → 100% CH₃CN, with UV detection, λ = 210 nm. Purity of all compounds subjected to biological testing was over 95%. Optical rotations were measured in CHCl₃ using an Autopol IV instrument (Rudolph Research Analytical). IR spectra were measured in CHCl₃ on an FT-IR spectrometer.

((((2,5-Dioxopyrrolidin-1-yl)oxy)carbonyl)oxy)methyl Pivalate (4a)

Chloromethyl carbonochloridate 1a (2.00 g, 1.38 mL, 15.5 mmol) was dissolved in anhydrous Et₂O (20 mL). The reaction mixture was cooled to 0 °C, and the mixture of Et₃N (1.57 g, 2.16 mL, 15.5 mmol, 1 equiv) and EtSH (964 mg, 1.15 mL, 15.5 mmol, 1 equiv) in anhydrous Et₂O (5 mL) was added dropwise over 5 min. The resulting mixture was then stirred overnight (25 h) at rt. Precipitate was filtered through a pad of Celite, and solvent was removed under reduced pressure. The crude O-(chloromethyl) S-ethyl carbonothioate 2a (2.4 g, colorless liquid) was used in the next step without purification. 2a (2.40 g, 15.5 mmol) was dissolved in pivalic acid (9.51 g, 93.1 mmol, 6 equiv), and freshly prepared salt of pivalic acid (4.76 g, 46.6 mmol, 3 equiv) and DIPEA (6.02 g, 8.10 mL, 46.6 mmol, 3 equiv) were added in few portions. The reaction mixture was heated to 60 °C for 22 h. EtOAc (100 mL) was added, and the organic phase was extracted with water (100 mL), satd NaHCO₃ (3 × 100 mL), and satd NaCl (100 mL), dried over MgSO₄, solvent was removed under reduced pressure, and the crude product 3a (3.30 g, 97%, light-yellow liquid) was used in the next step without purification. (((Ethylthio)-carbonyl)oxy)methyl pivalate 3a (3.20 g, 14.5 mmol) was dissolved in anhydrous DCM (40 mL), N-hydroxysuccinimide (3.34 g, 29.1 mmol, 2 equiv) was added, and the suspension was cooled to 0 °C. Peracetic acid (3.31 g (100%), 9.21 g (36%), 43.6 mmol, 3 equiv, 36% solution in acetic acid) was added dropwise over 15 min. The resulting mixture was stirred for 1 h at 0 °C and 2 h at rt. DCM (50 mL) was added, and the organic phase was washed with water (30 mL) and satd NaCl (30 mL), dried over MgSO₄, solvent was removed under reduced pressure, and column chromatography of the residue (EtOAc/hexane 1:2, R_f 0.27) yielded 4a as a colorless solid (2.54 g, 64% over three steps). ¹H NMR (400 MHz, CDCl₃): δ 1.24 (9H, s), 2.84 (4H, s), 5.86 (2H, s). ¹³C NMR (101 MHz, CDCl₃): δ 25.56 (2C), 26.86 (3C), 38.96, 83.67, 150.90, 168.34 (2C), 176.54. IR (CHCl₃): 2979 m, 2939 w, 2876 w, 1823 s, 1796 vs, 1649 vs, 1481 m, 1463 m, 1456 w, 1431 m, 1398 w, 1371 m, 1367 m, 1280 m, 1199 vs, 1110 vs, 1047 m, 998 s, 986 s, 942 m, sh, 924 s, 853 w, cm⁻¹. ESI MS: 296 ([M + Na]⁺). HR ESI MS: calcd for C₁₁H₁₅O₇NNa 296.07407; found 296.07410.

((((2,5-Dioxopyrrolidin-1-yl)oxy)carbonyl)oxy)ethyl Pivalate (4b)

Chloroethyl carbonochloridate 1b (2.00 g, 1.51 mL, 14.1 mmol) was dissolved in anhydrous Et₂O (20 mL). The reaction mixture was cooled to 0 °C, and the mixture of Et₃N (1.43 g, 1.97 mL, 14.1 mmol, 1 equiv) and EtSH (876 mg, 1.02 mL, 14.1 mmol, 1 equiv) in anhydrous Et₂O (5 mL) was added dropwise over 5 min. The resulting mixture was then stirred overnight (17 h) at rt. Precipitate was filtered through a pad of Celite, and solvent was removed under reduced pressure. The crude O-(1-chloroethyl) S-ethyl carbonothioate 2b (2.20 g, colorless liquid) was used in the next step without purification. 2b (2.20 g, 13.1 mmol) was dissolved in pivalic acid (8.02 g, 78.6 mmol, 6 equiv), and freshly prepared salt of pivalic acid (4.01 g, 39.3 mmol, 3 equiv) and DIPEA (5.08 g, 6.85 mL, 39.3 mmol, 3 equiv) was added in few portions. The reaction mixture was heated to 70 °C for 20 h. EtOAc (100 mL) was added, and the organic phase was extracted with water (100 mL), satd NaHCO₃ (3 × 100 mL), and satd NaCl (100 mL), dried over MgSO₄, solvent was removed under reduced pressure, and the crude product 3b (2.82 g, 92%, light-yellow liquid) was used in the next step without purification. (((Ethylthio)carbonyl)oxy)methyl pivalate 3b (2.82 g, 12.1 mmol) was dissolved in anhydrous DCM (40 mL), N-hydroxysuccinimide (2.77 g, 24.1 mmol, 2 equiv) was added, and the suspension was cooled to 0 °C. Peracetic acid (2.76 g (100%), 7.67 g (36%), 36.3 mmol, 3 equiv, 36% solution in acetic acid) was added dropwise over 15 min. The resulting mixture was stirred for 1 h at 0 °C and 2 h at rt. DCM (50 mL) was added, and the organic phase was washed with water (30 mL) and satd NaCl (30 mL), dried over MgSO₄, solvent was removed under reduced pressure, and column chromatography of the residue (EtOAc/hexane 1:2, R_f 0.27) yielded 4b as a colorless oil (2.02 g, 50% over three steps). ¹H NMR (400 MHz, CDCl₃): δ 1.21 (9H, s), 1.60 (3H, d, J = 5.4), 2.83 (4H, s), 6.81 (1H, q, J = 5.4). ¹³C NMR (101 MHz, CDCl₃): δ 19.43, 25.58 (2C), 6.88 (3C), 38.89, 93.83, 150.01, 168.43, 176.10. IR (CHCl₃): 2978 m, 2939 w, 2875 w, 1821 s, 1795 s, 1748 vs, br, 1480 m, 1432 m, 1396 m, 1371 m, 1365 m, sh, 1298 w, sh, 1280 s, 1199 s, 1054 vs, 1028 s, 944 vw, sh, 935 w, sh, 811 w, cm⁻¹. ESI MS: 310 ([M + Na]⁺). HR ESI MS: calcd for C₁₂H₁₇O₇NNa 310.08972; found 310.08979.

1-((((2,5-Dioxopyrrolidin-1-yl)oxy)carbonyl)oxy)-2-methylpropyl Pivalate (4c)

1-Chloro-2-methylpropyl carbonochloridate 1c (2.00 g, 1.71 mL, 11.7 mmol) was dissolved in anhydrous Et₂O (20 mL). The reaction mixture was cooled to 0 °C, and the mixture of Et₃N (1.18 g, 1.63 mL, 11.7 mmol) and EtSH (727 mg, 866 μL, 11.7 mmol) in anhydrous Et₂O (10 mL) was added dropwise over 10 min. The resulting mixture was stirred overnight (23 h) at rt. Precipitate was filtered through a pad of Celite, and solvent was removed under reduced pressure. The crude product O-(1-chloro-2-methylpropyl) S-ethyl carbonothioate 2c (2.20 g, 96%, colorless liquid) was used in the next step without purification. 2c (1.20 g, 6.10 mmol) was dissolved in pivalic acid (3.74 g, 4.20 mL, 36.6 mmol, 6 equiv), and freshly prepared salt of pivalic acid (1.87 g, 2.10 mL, 18.3 mmol, 3 equiv) and DIPEA (2.37 g, 3.19 mL, 18.3 mmol, 3 equiv) were added in few portions. The reaction mixture was heated to 60 °C for 70 h. EtOAc (100 mL) was added, and the organic phase was extracted with water (50 mL), satd NaHCO₃ (3 × 50 mL), satd NaCl (50 mL), dried over MgSO₄, solvent was removed under reduced pressure, and the crude product 1-(((ethylthio)carbonyl)oxy)-2-methylpropyl pivalate 3c (1.32 g, 83%, light-yellow liquid) was used in the next step without purification. 3c (1.28 g, 4.88 mmol) was dissolved in anhydrous DCM (13 mL), N-hydroxysuccinimide (1.12 g, 9.76 mmol, 2 equiv) was added, and the suspension was cooled to 0 °C. Peracetic acid (1.11 g (100%), 3.09 g (36%), 14.6 mmol, 3 equiv, 36% solution in acetic acid) was added dropwise over 10 min. The resulting mixture was stirred for 1 h at 0 °C and for 2 h at rt. DCM (40 mL) was added, and the organic phase was washed with water (20 mL) and satd NaCl (20 mL), dried over MgSO₄, solvent was removed under reduced pressure, and column chromatography of the residue (EtOAc/hexane 3:5, R_f 0.26) yielded 4c as a light-yellow oil (863 mg, 56% over 3 steps). ¹H NMR (400 MHz, CDCl₃): δ 1.00 (6H, d, J = 6.9), 1.21 (9H, s), 2.08–2.19 (1H, m), 2.81 (4H, s), 6.55 (1H, d, J = 5.0). ¹³C NMR (101 MHz, CDCl₃): δ 16.06, 16.38, 25.55, 26.89, 31.84, 39.05, 98.19, 150.37, 168.48, 176.14. IR (CHCl₃): 2978 m, 2938 w, 2878 w, 1821 s, 1795 s, 1748 vs, br, 1481 m, 1463 w, 1432 m, 1396 w, 1373 m, 1366 m, sh, 1279 m, 1199 s, 1046 m, 998 m, sh, 987 m, 932 s cm⁻¹. ESI MS: 338 ([M + Na]⁺). HR ESI MS: calcd for C₁₄H₂₁O₇NNa 338.12102; found 338.12115.

Chloro(phenyl)methyl(4-nitrophenyl)carbonate (7)

Chloro-(phenyl)methyl carbonochloridate (6) was prepared from benzaldehyde (5) by a previously reported method.⁶⁴ Compound 6 (900 mg, 4.39 mmol) was dissolved in anhydrous DCM (20 mL), 4-nitrophenol (611 mg, 4.39 mmol, 1 equiv) was added, and the mixture was cooled to 0 °C. Pyridine (347 mg, 355 μL, 4.39 mmol, 1 equiv) in anhydrous DCM (5 mL) was added dropwise over 5 min. Reaction mixture was stirred for 2 h at rt. DCM was evaporated, and the crude product was purified by column chromatography (DCM/hexane 1:1). Compound 7 was obtained as a colorless solid (520 mg, 39%). ¹H NMR (400 MHz, CDCl₃): δ 7.33 (1H, s), 7.41–7.50 (5H, m), 7.58–7.63 (2H, m), 8.28–8.34 (2H, m). ¹³C NMR (101 MHz, CDCl₃): δ 87.37, 121.83 (2C), 125.59 (2C), 126.41 (2C), 129.07 (2C), 130.56, 136.35, 145.91, 150.50, 155.07. IR (CHCl₃): 3119 w, 3088 w, 3071 vw, 3032 w, 1788 vs, 1772 s, sh, 1619 m, 1595 m, 1530 vs, 1492 s, 1456 m, 1349 vs, 1317 m, 1296 m, 1232 vs, sh, 1178 m, sh, 1165 m, 1111 m, 1105 w, sh, 1078 m, 1054 s, 1029 m, 1014 m, 1002 w, 978 s, 920 w, 872 s, 854 s, 830 vw, 708 s, 695 m, sh, 680 w, 626 vw, 618 vw, 530 vw, 495 w, 403 w cm⁻¹. ESI MS: 329 ([M + Na]⁺). HR ESI MS: calcd for C₁₄H₁₀O₅NClNa 330.01397; found 330.01367.

(((4-Nitrophenyloxy)carbonyl)oxy)(phenyl)methyl Pivalate (8)

Compound 7 (100 mg, 0.325 mmol) and mercury pivalate (157 mg, 0.390 mmol, 1.2 equiv) were dissolved in anhydrous DCM (6 mL). Reaction mixture was stirred under inert atmosphere at rt overnight (16 h). DCM (10 mL) was added, and reaction mixture was washed with satd NaHCO₃ (10 mL) and brine (10 mL), the organic phase was dried over MgSO₄, and DCM was evaporated. The product 8 (115 mg, 95%) was used in the next step without purification. ¹H NMR (400 MHz, CDCl₃): δ 1.28 (9H, s), 7.38–7.43 (2H, m), 7.44–7.50 (3H, m), 7.57–7.60 (1H, m), 7.61 (1H, s), 8.23–8.33 (2H, m). ¹³C NMR (101 MHz, CDCl₃): δ 27.02 (3C), 39.11, 93.80, 121.86 (2C), 125.47 (2C), 126.84 (2C), 128.97 (2C), 130.48, 134.39, 145.69, 150.73, 155.32, 176.44. IR (CHCl₃): 3118 w, 3087 w, 3072 w, 3031 m, 2980 m, 2875 w, 1775 vs, 1747 s, 1618 m, 1595 m, 1529 vs, 1493 s, 1480 m, 1459 m, 1399 m, 1365 m, 1349 vs, 1279 vs, 1248 vs, 1165 s, 1123 vs, 1112 s, sh, 1030 s, 1014 m, 1003 m, 970 s, br, 943 s, 918 m, 865 s, 860 s, 832 w, 697 s, 682 w, 633 w, 619 vw, 530 vw, 495 w, 403 vw cm⁻¹. ESI MS: 396 ([M + Na]⁺). HR ESI MS: calcd for C₁₉H₁₉O₇NNa 396.10537; found 396.10546.

2-(((4-Nitrophenyloxy)carbonyl)oxy)propan-2-yl Pivalate (11)

2-Chloropropan-2-yl (4-nitrophenyl) carbonate 10⁵⁹ (300 mg, 1.16 mmol) was dissolved in anhydrous DCM (15 mL). Mercury pivalate (559 mg, 1.39 mmol, 1.2 equiv) was added, and the reaction mixture was stirred overnight (20 h) at rt under inert atmosphere. The solid precipitate (HgCl₂) was removed by filtration, DCM (15 mL) was added, and organic phase was washed with satd NaHCO₃ (15 mL) and satd NaCl (15 mL), dried over MgSO₄, solvent was removed under reduced pressure, and the product 11 was obtained as a light-yellow oil (301 mg, 80%). ¹H NMR (400 MHz, CDCl₃): δ 1.20 (9H, s), 1.91 (6H, s), 7.34–7.40 (2H, m), 8.24–8.30 (2H, m). ¹³C NMR (101 MHz, CDCl₃): δ 25.28 (2C), 27.02 (3C), 39.59, 107.71, 121.98 (2C), 125.40 (2C), 145.50, 149.07, 155.41, 175.97. IR (CHCl₃): 3031 w, 2976 w, 2875 w, 1777 m, 1736 m, 1618 w, 1595 w, 1528 m-s, 1493 m, 1481 w, 1439 w, 1396 w, 1376 w, 1349 m, 1322 w, 1264 m, 1191 m, 1112 vs, 1094 s, sh, 1030 w, 980 w, 859 m, 682 vw, 491 vw cm⁻¹. ESI MS: 348 ([M + Na]⁺). HR ESI MS: calcd for C₁₅H₁₉O₇NNa 348.10537; found 348.10543.

Isopropyl 6-Diazo-5-oxo-2-((((pivaloyloxy)methoxy)-carbonyl)amino)hexanoate (13a)

Compound 4a (320 mg, 1.17 mmol) was suspended in anhydrous DCM (6 mL). The reaction mixture was cooled to 0 °C, and compound 12⁴⁷ (250 mg, 1.17 mmol, 1 equiv) in anhydrous DCM (3 mL) was added dropwise. The mixture was stirred for 15 min at 0 °C and then 2 h at rt. Solvent was removed under reduced pressure, and column chromatography of the residue (EtOAc/hexane 1:2, R_f 0.21) yielded the desired compound 13a (175 mg, 40%) as yellow oil. ¹H NMR (400 MHz, CDCl₃): δ 1.19 (9H, s), 1.23 (3H, d, J = 6.2), 1.24 (3H, d, J = 6.2), 1.90–2.05 (1H, m), 2.14–2.25 (1H, m), 2.31–2.51 (2H, m), 4.28 (1H, td, J = 8.2, 4.7), 5.03 (1H, hept, J = 6.2), 5.27 (1H, bs), 5.65 (1H, d, J = 8.1), 5.69 (1H, d, J = 5.7), 5.73 (1H, d, J = 5.7). ¹³C NMR (101 MHz, CDCl₃): δ 21.79, 21.81, 26.97, 36.34, 38.86, 53.63, 54.90, 69.75, 80.33, 154.40, 171.01, 177.51, 193.43. Optical rotation: [α]²²_D +13.0° (c 0.184, CHCl₃). IR (CHCl₃): 3424 m, 3354 w, br, 3116 w, 2984 s, 2937 m, 2875 s, 2110 vs, 1747 vs, 1730 vs, sh, 1642 s, 1512 s, 1481 m, 1466 m, 1453 m, 1377 s, 1282 s, 1182 m, 1145 s, 1105 s, 994 s, 942 m, cm⁻¹. ESI MS: 394 ([M + Na]⁺). HR ESI MS: calcd for C₁₆H₂₅O₇N₃Na 394.15847; found 394.15855.

Isopropyl 6-Diazo-5-oxo-2-(((1-(pivaloyloxy)ethoxy)-carbonyl)amino)hexanoate (13b)

This compound was prepared according to a procedure previously described in detail.⁴⁷ Yields and ¹H NMR and ¹³C NMR spectra were in agreement with the published data.

Isopropyl 6-Diazo-2-(((2-methyl-1-(pivaloyloxy)propoxy)-carbonyl)amino)-5-oxohexanoate (13c)

Compound 4c (399 mg, 1.27 mmol, 0.9 equiv) was suspended in anhydrous DCM (7 mL). The reaction mixture was cooled to 0 °C, and compound 12⁴⁷ (300 mg, 1.41 mmol, 1 equiv) in anhydrous DCM (3 mL) was added dropwise. The mixture was stirred for 15 min at 0 °C and then 2 h at rt. Solvent was removed under reduced pressure, and column chromatography of the residue (EtOAc/hexane 1:2, R_f 0.30) yielded 13c (285 mg, 54%) as yellow oil (mixture of two stereoisomers 1:1). ¹H NMR (400 MHz, CDCl₃, first stereoisomer): δ 0.94 (6H, d, J = 6.8), 1.16 (9H, s), 1.23 (6H, t, J = 6.3), 1.83–2.50 (4H, m), 4.22–4.31 (1H, m), 5.02 (1H, hept, J = 6.8), 5.29 (1H, bs), 5.48 (1H, d, J = 8.3), 6.52 (1H, d, J = 4.9). ¹³C NMR (101 MHz, CDCl₃, first stereoisomer): δ 16.40, 16.54, 27.00, 28.05, 31.87, 36.29, 38.96, 53.38, 54.82, 69.64, 94.21, 154.28, 171.31, 176.56, 193.87. ¹H NMR (400 MHz, CDCl₃, second stereoisomer): δ 0.93 (6H, d, J = 6.8), 1.18 (9H, s), 1.22 (6H, t, J = 6.3), 1.83–2.50 (4H, m), 4.22–4.31 (1H, m), 5.00 (1H, sept, J = 6.8), 5.37 (1H, bs), 5.45 (1H, d, J = 8.3), 6.48 (1H, d, J = 4.9). ¹³C NMR (101 MHz, CDCl₃, second stereoisomer): δ 16.37, 16.54, 26.98, 27.74, 31.91, 36.55, 38.92, 53.54, 54.82, 69.66, 93.87, 154.22, 171.17, 176.81, 193.58. Optical rotation: [α]²²_D +11.5° (c 0.261, CHCl₃). IR (CHCl₃): 3428 m, 3116 w, 2982 s, 2936 m, 2878 m, 2110 vs, 1741 vs, br, 1731 vs, sh, 1641 s, 1508 s, 1480 m, 1463 m, 1400 m, sh, 1385 s, sh, 1377 s, 1365 s, sh, 1281 s, 1231 s, 1183 m, 1146 s, 1105 s, 990 s, 941 m cm⁻¹. ESI MS: 436 ([M + Na]⁺). HR ESI MS: calcd for C₁₉H₃₁O₇N₃Na 436.20542; found 436.20553.

Isopropyl 6-Diazo-5-oxo-2-(((phenyl(pivaloyloxy)methoxy)-carbonyl)amino)hexanoate (13d)

Compound 8 (817 mg, 2.19 mmol) was dissolved in anhydrous DMF (15 mL). Compound 12⁴⁷ (700 mg, 3.18 mmol, 1.5 equiv) in anhydrous DMF (7 mL) was added dropwise. Reaction mixture was stirred at rt under inert atmosphere for 5 h. DMF was evaporated, and column chromatography of the residue (EtOAc/hexane 1:2, R_f 0.18) yielded product 13d (648 mg, 66%) as a light-yellow oil (mixture of two stereoisomers 1:1). ¹H NMR (400 MHz, CDCl₃): δ 1.17–1.31 (15H, m), 1.90–2.06 (1H, m), 2.12–2.31 (1H, m), 2.31–2.54 (2H, m), 4.27–4.36 (1H, m), 5.03 (1H, hept, J = 6.3), 5.29 (1H, bs), 5.59 (1H, d, J = 8.1), 7.36–7.42 (3H, m), 7.46–7.51 (2H, m), 7.61 (s, 1H). ¹³C NMR (101 MHz, CDCl₃): δ 21.82 (2C), 26.99 (3C), 27.70, 36.35, 38.99, 53.55, 54.90, 69.79, 90.93, 126.57 (2C), 128.66 (2C), 129.64, 135.95, 153.82, 171.10, 176.33, 193.51. Optical rotation: [α]²²_D +12.5° (c 0.246, CHCl₃). IR (CHCl₃): 3425 w, 3116 w, 3098 vw, 3070 vw, 3029 m, 2984 m, 2937 m, 2875 w, 2110 s, 1735 vs, br, 1641 s, 1590 w, 1507 s, 1480 m, 1457 m, 1398 m, 1377 s, 1367 s, sh, 1366 s, sh, 1280 s, 1182 m, 1146 s, sh, 1133 s, 1105 s, 1085 m, 1057 s, 1027 s, 1003 m, 942 m, 918 w, 697 m, 619 vw cm⁻¹. ESI MS: 470 ([M + Na]⁺). HR ESI MS: calcd for C₂₂H₂₉O₇N₃Na 470.18977; found 470.18985.

Isopropyl 6-Diazo-5-oxo-2-((((2-(pivaloyloxy)propan-2-yl)-oxy)carbonyl)amino)hexanoate (13e)

Compound 11 (936 mg, 2.88 mmol) was dissolved in anhydrous DMF (18 mL), and the reaction mixture was cooled to 0 °C. Compound 12⁴⁷ (920 mg, 4.32 mmol, 1.5 equiv) in anhydrous DMF (6 mL) was added dropwise. Reaction mixture was stirred at 0 °C under inert atmosphere for 5 h. DMF was evaporated, and column chromatography of the residue (EtOAc/hexane 1:2, R_f 0.22) yielded product 13e (955 mg, 83%) as a light-yellow oil. ¹H NMR (400 MHz, CDCl₃): δ 1.19 (9H, s), 1.23 (3H, d, J = 6.3), 1.24 (3H, d, J = 6.3), 1.80 (3H, s), 1.83 (3H, s), 1.90–2.01 (1H, m), 2.14–2.27 (1H, m), 2.29–2.51 (2H, m), 4.24 (1H, dt, J = 8.3, 4.7), 5.05 (1H, hept, J = 6.3), 5.31 (1H, bs), 5.44 (1H, d, J = 8.2). ¹³C NMR (101 MHz, CDCl₃): δ 21.80, 21.82, 25.76, 25.91, 27.07 (3C), 27.79, 36.46, 39.48, 53.26, 54.87, 69.59, 105.44, 153.16, 171.31, 176.21, 193.63. Optical rotation: [α]²²_D +12.8° (c 0.133, CHCl₃). IR (CHCl₃): 3430 w, 3116 w, 2984 m, 2936 m, 2874 m, 2110 s, 1732 vs, br, 1641 m, 1502 s, 1481 m, 1466 m, 1462 m, 1455 m, 1452 m, 1397 m, sh, 1384 s, 1374 s, 1365 s, sh, 1198 s, 1184 s, 1147 m, sh, 1128 s, 1112 s, 1105 s, 1045 m, 942 w cm⁻¹. ESI MS: 329 ([M + Na]⁺). HR ESI MS: calcd for C₁₈H₂₉O₇N₃Na 422.18977; found 422.18982.

EcoHIV-Infected Mice

All mouse efficacy studies were conducted in full compliance with NIH guidelines and with the approval of the Institutional Animal Care and Use Committee at the Icahn School of Medicine at Mount Sinai. Male C57BL/6J mice at 5 weeks old (The Jackson Laboratory, Bar Harbor, ME) were maintained on a 12 h light–dark cycle with ad libitum access to food and water. EcoHIV chimeric virus was generated by transfecting HEK293T cells with plasmid DNA containing the EcoHIV construct, specifically EcoNDK containing the V5C5 fragment of gp120.⁴² EcoHIV was then collected from the culture media, concentrated by centrifugation, and titered by p24 ELISA (Advanced Biosciences Laboratory, Rockville, MD). Each mouse was randomized to begin 14 (1 mg/kg, ip, qod) or saline treatment 1 day prior to inoculation with either EcoHIV (2 × 10⁶ pg p24, ip) or sham inoculation with PBS (n = 8/group) as previously described.⁶⁵ The dose of 14 was chosen based on several published efficacy studies in CNS models including multiple sclerosis,⁶⁶ brain cancer,⁴⁷ and sindbus virus^62,67,68 as well as our lab’s mouse pharmacokinetic studies showing μM brain levels of 14 after 1 mg/kg systemic dosing.⁶⁹ On day 30 postinoculation (p.i.), having allowed time for stable virus infection and emergence of cognitive deficits, all mice began behavioral training and testing.

Radial Arm Water Maze

The radial arm water maze (RAWM) test was administered in a pool of opaque water containing six swimming lanes and a hidden platform with visual cues essentially as described.⁷⁰ The visual cues consisted of both two- and three-dimensional objects that were affixed to the side of the maze pool using clear packing tape. The two-dimensional cues were six different brightly colored 8½ × 11 sheets of paper each printed with a different black shape (square, triangle, star, etc.), and the three-dimensional cues were commonly found laboratory items (funnel, culture flask, pipet tip box, etc.). Briefly, RAWM consisted of four learning trials (LT) of 60 s and one post-training 60 s retention test (RT) administered after 30 min rest every day until test completion. Testing was considered complete when control mice reached asymptotic performance of one error or fewer in finding the hidden platform on trials LT4 and RT. Errors for the last 3 days of testing for all groups were then averaged and used for statistical analysis. The hidden platform was rotated randomly to a different arm each test day to ensure that mice used working memory to locate the platform. Each of the training trials began by placing a mouse randomly into one of the six swimming arms and allowing the mouse to swim for 60 s to find the hidden platform, during which time the number of errors (entering an arm without the platform and/or 20 s of immobility) and latency to locate the hidden platform were recorded. The retention test was performed in the same manner as the learning trials. The hidden platform tests were followed by measuring the latency it took for treated and control mice to find a visible platform in the same context, as a control for possible effects of treatment on animal vision, motivation, or ability to swim to the platform.

Viral Load

After completion of behavioral testing, all mice were sacrificed by carbon dioxide asphyxiation after brief isoflurane anesthesia. Spleen and brain were harvested, DNA was isolated, and viral load was measured as previously described.^65,71–73 Briefly, total DNA was isolated from mouse tissues by sequentially homogenizing in Trizol, mixing with DNAzol, precipitating with ethanol, washing, and treating with NaOH. Quantitative PCR (qPCR) was then conducted to detect EcoHIV/NDK DNA from the gag gene region using Taqman chemistry with MGB probes with forward primer 5′-TGGGACCACAGGCTACACTAGA-3′, reverse primer 5′-CAGCCAAAACTCTTGCTTTATGG-3′, and probe 5′-TGATGACAGCATGCCAGGGAGTGG-3′ (ThermoFisher Scientific). A standard curve for quantitation of copy number was constructed using graded numbers of a plasmid containing the EcoNDK gag amplicon. Copy number was then normalized to cell count by amplification of murine gapdh (Mm99999915_g1). All qPCR amplification was performed in an Applied Biosystems 7500 instrument.

In Vitro Metabolic Stability

Metabolic stability of prodrugs were evaluated as previously described.⁴⁷ Briefly, prodrugs (10 μM) were spiked in mouse, swine, and human plasma or swine brain homogenate and incubated in each matrix in an orbital shaker at 37 °C. Relative prodrug amounts at 0 and 60 min for plasma and 0, 30, and 60 min for brain were measured using liquid chromatography and tandem mass spectrometry (LC–MS/MS) by removing 100 μL aliquots of each mixture in triplicate and quenching each reaction by addition of three times the volume of ice-cold acetonitrile spiked with the internal standard (losartan 5 μM). The samples were vortexed for 30 s and centrifuged at 12000g for 10 min. Then 50 μL of each supernatant was diluted with 50 μL of water and transferred to a 250 μL polypropylene vial sealed with a Teflon cap. Prodrug detection was performed on a Thermo Scientific Accela UPLC system coupled to Accela open autosampler on an Agilent C18 (100 mm × 2.1 mm i.d.) UPLC column. The autosampler was temperature controlled and was operated at 10 °C. The mobile phase used for the chromatographic separation was composed of acetonitrile/water containing 0.1% formic acid and flow rate of 0.5 mL/min for 4.5 min using gradient elution. The column effluent was monitored using TSQ Vantage triple-quadrupole mass-spectrometric detector, equipped with an electrospray probe set in the positive ionization mode. Samples were introduced into the ionization source through a heated nebulized probe (350 °C). Relative prodrug amounts were measured from ratio of peak areas of analyte to IS; percent remaining was calculated by normalizing this value at 60 min to the value obtained at 0 min.

Protein Binding

The free 14 fraction (f_u) in swine plasma and swine brain homogenate was determined using a previously described ultrafiltration method.⁷⁴ Briefly, the 14 (10 μM) spiked plasma and/or brain homogenate was incubated at 37 °C for 30 min, following which the samples were loaded onto an ultrafiltration column tube (Corning Spin-X UF with a MW cut off of 30 kDa), whose base is impermeable to the plasma and brain proteins. The samples were centrifuged for 45 min at 37 °C and 4000 rpm (1800g). The concentrations were determined in both the filtrate and plasma before filtration, following extraction and quantified via LC-MS/MS. Fraction unbound was calculated from the equation C_u/C_t × 100, where C_u was the concentration in the ultrafiltrate and C_t was the concentration in the plasma or brain homogenate prior to filtration.

In Vivo Pharmacokinetics

Swine studies were conducted under a protocol approved by the Johns Hopkins Animal Care and Use Committee. Adult, female Göttingen × Yucutan miniature swine (Massachusetts General Hospital, MA) were housed in Johns Hopkins University facilities accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International in compliance with the Animal Welfare Act, Animal Welfare Regulations, and the Public Health Service Policy on the Humane Care and Use of Laboratory Animals. Animals were maintained on a 14 h light and 10 h dark schedule, and provided ad libitum water and a commercial miniswine diet (Teklad, Madison, WI) with environmental enrichment (fruit/vegetables) twice daily. Animals were individually housed while on study in order to monitor behavior and clinical health following drug administration. Whole blood for drug pharmacokinetic evaluation was collected from a dual lumen central venous catheter (CVC) implanted in the external jugular vein prior to study initiation. Animals were anesthetized with a combination of ketamine hydrochloride (20–30 mg/kg, im) and xylazine (2 mg/kg, im), intubated, and maintained under isoflurane (1–2%) inhalant anesthesia. A temporary peripheral saphenous vein catheter was placed in the hind limb to allow for anatomical separation of drug infusion and whole blood sampling via CVC. 14, 13b, or 13d were dissolved in a sterile saline solution containing 5% ethanol and 5% Tween 80 prior to iv infusion via saphenous vein catheter over 1 h (1 mL/min) for a final dose of 1.6 mg/kg or molar equivalent administered at 1 mL/kg (n = 1/dose). Blood samples (1 mL) were taken from CVC at predose, 5, 15, 30, 45, and 60 min. Plasma was separated by low speed centrifugation at 3000g for 10 min at 4 °C. CSF was obtained from the cisterna magna using a 3.5 in. × 22 gauge spinal needle (Becton Dickinson Health Care, Franklin Lakes, New Jersey, USA) at 60 min postdose. All samples were flash frozen upon harvest and stored at –80°C until bioanalysis.

In a separate study, the above dosing procedure was repeated except that the animals (n = 2/dose) were euthanized at 60 min postinfusion using Euthasol (Virbac Animal Health, Fort Worth, TX, USA) containing pentobarbital sodium and phenytoin sodium at a dose of 1 mL per 10 lbs of body weight and plasma and brain tissues were harvested. To obtain brain tissues, a horizontal incision connecting both orbits was extended toward the base of the ears and skin, and subcutaneous tissue over the swine skull were removed. A reciprocating saw was used to cut through the frontal aspect of the calvarium to create a window for intracranial access. Next, the dura mater was cut and a 2 cm × 2 cm sample of full thickness frontal cortex was obtained. All samples were flash frozen and stored at –80 °C until bioanalysis.

Bioanalysis

Quantitation of 14 in plasma, CSF, and brain homogenate by LC-MS/MS was performed as previously described.⁴⁷ Briefly, 14 was extracted from plasma, CSF, and brain samples with methanol containing glutamate-d₅ (10 μM ISTD) by vortexing followed by centrifugation 16000g for 5 min. Supernatants were aliquoted and dried at 45 °C for under vacuum for 1 h. Sodium bicarbonate buffer (0.2M, pH 9.0) and dabsyl chloride (10 mM) in acetone were added to each tube, mixed, and incubated for 15 min at 60 °C to derivatize. Samples were then injected and separated on an Agilent 1290 equipped with an Agilent Eclipse plus C18 RRHD 2.1 mm × 100 mm column over a 2.5 min gradient from 20 to 95% acetonitrile +0.1% formic acid and quantified on an Agilent 6520 QTOF mass spectrometer. Peak area ratio of the analyte to the internal standard was plotted against a 14 standard curve to yield 14 concentrations for each sample.

Pharmacokinetic and Statistical Analysis

Area under the curve (AUC) was calculated by log-linear trapezoidal rule to the end of sample collection by noncompartmental analysis module in WinNonlin (version 5.3, Certara, St. Louis, MO). cLogPs were calculated using ChemDraw Professional (PerkinElmer, Waltham, MA). The half-life (t_1/2) of prodrug metabolic stability from brain homogenates was estimated using the first-order equation t_1/2 = 0.693/K_el, where K_el (elimination rate constant) is the slope of linear regression from natural log percentage substrate remaining versus incubation time.⁷⁵ For all studies, group means and standard errors were calculated and used for statistical comparisons where appropriate. Differences in radial arm water maze errors and latencies between groups were assessed by two-way ANOVA (treatment × trial) with posthoc Tukey test. Viral load in each tissue was compared between treatment groups by t test. Prodrug half-lives were compared by one-way ANOVA. For all tests, significance was defined as p < 0.05.

ABBREVIATIONS

DON

6-diazo-5-oxo-L-norleucine

HAND

HIV-associated neurocognitive disorders

cART

combined antiretroviral therapy

HAD

HIV-associated dementia

CNS

central nervous system

MRS

magnetic resonance spectroscopy

CSF

cerebrospinal fluid

GI

gastrointestinal

cLogP

calculated partition coefficient

DCM

dichloromethane

DMF

dimethylformamide

RAWM

radial arm water maze

LT

learning trial

RT

retention trial

AUC

area under the curve

POM

pivaloyloxymethyl

ELISA

enzyme-linked immunosorbent assay

pi

post-inoculation

iv

intravenous

ip

intraperitoneal

qPCR

quantitative polymerase chain reaction

CVC

central venous catheter

LC

liquid chromotagraphy

MS

mass spectrometry

ESI

electrospray ionization