Folate-conjugated organic CO prodrugs: Synthesis and CO release kinetic studies

Carbon monoxide (CO) is an endogenous produced molecule and has shown efficacy in animal models of inflammation, organ injury, colitis and cancer metastasis. Because of its gaseous nature, there is a need for developing efficient CO delivery approaches, especially those capable of targeted delivery. In this study, we aim to take advantage of a previously reported approach of enrichment-triggered prodrug activation to achieve targeted delivery by targeting the folate receptor. The general idea is to exploit folate receptor-mediated enrichment as a way to accelerate a biomolecular Diels-Alder reaction for prodrug activation. In doing so, we first need to find ways to tune the reaction kinetics in order to ensure minimal rection without enrichment and optimal activation upon enrichment. In this feasibility study, we synthesized two diene-dienophile pairs and studied their reaction kinetics and ability to target the folate receptor. We found that folate conjugation significantly affects the reaction kinetics of the original diene-dienophile pairs. Such information will be very useful in future designs of similar targeted approaches of CO delivery.


Introduction
Success of conventional therapeutic approach is often limited by the off-target drug delivery and associated side-effects.[1,2] To a great extent, such kind of off-target effects can be minimized by targeted delivery of drugs to the intended site, sparing the healthy cells.[3,4,2] In general, pathophysiology of most disease is often associated with over expression of certain biomarkers, [5,6] providing an opportunity for targeted delivery of therapeutics.Folic acid, also known as vitamin-B 9 , is essential for cell growth and proliferation by helping DNA and RNA synthesis.[7,8] Because mammals do not synthesize folate, it has to come from nutrients.Further, cellular uptake is essential for all types of cells for their routine metabolic process, and the demand is generally high in rapidly metabolizing tumor cells.Generally, folate uptake in mammalian cells happens through three different mechanisms, a proton coupled folate transporter, reduced folate receptor carrier (RFC), and folate receptors (FRs).[9] FRs are mainly enriched in tumor cells, macrophages, and proximal tubule cells, whereas the former two pathways are ubiquitous in nature.[10] FR is a glycosylphosphatidyl inositol modi ed cell-surface receptor and uptakes the folate via receptor mediated endocytosis.[11] Literature reports indicate high a nity binding by FRs (K d ~ 10 − 9 M); [12] thus, conjugation with folic acid provides an e cient approach for the targeted delivery of payloads to FR expressing to proximal tubule cells and tumor cells as well as in amed tissues such as in colitis.[13] Folate receptors (FRs) are of four different isoforms: FRα, FRβ, FRγ and FRδ.Among these, FRα, FRβ, and FRδ are anchored by cell-surface GPI proteins, whereas FRδ lacks the GPI proteins and consequently functions as a secretory protein.[9] We are interested in ways for intracellular delivery of carbon monoxide (CO) via folate receptor-mediated uptake for treating organ injury such as acute kidney injury [14] and in ammatory conditions such as colitis.[15] Carbon monoxide (CO) is endogenously produced as part of red blood cell turn over with signi cance on par with nitric oxide and hydrogen sul de.[16] CO has shown pharmacological effects in animal models of systemic in ammation, [17,18] lung injury, [19] liver injury, [20,21] kidney injury, [22,23] colitis, [15] and cancer metastasis.[24] Aimed at searching for easy, safe, and controllable delivery of CO as alternatives to inhaled CO gas, many labs have worked on CO donors of various types including metal-carbonyl complexes as CO-releasing molecules (CORMs) [25][26][27][28] and CORMs for triggered CO release.[29][30][31][32][33] Beyond metal-carbonyl complexes, there are organic CO donors that are photo-sensitive, ROS-sensitive, mechanical force-sensitive, or chemoexcitation-sensitive. [34][35][36][37][38][39][40] There are also innovative ways of trapped CO, [41] CO in micelles, [29,42] and CO solution.[43,41] In 2014, we reported the rst organic CO prodrugs by taking advantage of a cheletropic reaction for CO release from a norbornadienone scaffold (3, Fig. 1), which can be generated by using an intra-molecular Diels-Alder reaction (Fig. 1A), [15,44] an inter-molecular Diels-Alder reaction (Fig. 1B), [45,46] or an elimination reaction (Fig. 1C) [47][48][49] to form the norbornadiene-7-ones intermediate for subsequent CO release (Fig. 1).Independently, the Larsen lab also developed a similar approach of using the norbornadienone scaffold, generated through an elimination reaction.[50][51][52] Even with all the advances in developing new delivery forms of CO, there is a need for targeted delivery to minimize side effects and improve the safety margin.Along this line, we have worked on CO prodrugs capable of triggered release using stimuli such as esterase, [54,18] ROS,[48] and pH.[47] Especially worth noting is our efforts on enrichment-triggered release, which allows for targeted delivery of CO with improved potency.[45] The basic design principle relies on the use of a bimolecular approach (Fig. 1B) to generate the norbornadienone scaffold and the concentration-dependent nature of such a biomolecular reaction rate for substantially improved CO release rate upon enrichment.However, past enrichment efforts relied on triphenylphosphonium (TPP)-based mitochondria targeting (Fig. 2B).Though the enrichment part worked well, the use of a bulky hydrophobic TPP group brings in problems of drug-like properties.In this study, we examine the feasibility of using folate-mediated endocytosis as a way to achieve enrichment of the prodrug components and then selective targeting of CO (Fig. 2B).A critical requirement for success in such an enrichment-triggered release approach is the ability to tune the reaction rate in such a way that the bimolecular reaction is slow at dosing concentration and is fast after enrichment.Therefore, it is important to understand how conjugation with folate would affect the reaction rate of the intended bimolecular Diels-Alder reaction (Fig. 1B).In this study, we describe the synthesis of two diene-dienophile pairs that allowed us to achieve a basic understanding of how such conjugation affects reaction rate.Such information will be important for future optimization work.

Results and Discussion
Design.Again, the basic design for the compounds in this study is based on a bimolecular Diels-Alder reaction, leading to the formation of an unstable norbornadiene-7-one intermediate with subsequent spontaneous CO release through a cheletropic extrusion reaction (Fig. 1B).
Speci cally, norbornadiene-7-ones (Fig. 1, 3) are highly unstable under physiological conditions and undergo a spontaneous bond-breaking cheletropic reaction to release CO.By using an inter-molecular Diels-Alder reaction, our lab has demonstrated enrichment-triggered delivery of CO inside the mitochondria.[45] In this bimolecular reaction, both reaction partners, cyclopentadienone (CPD) and strained cycloalkyne (BCN), are conjugated with TPP for enrichment in mitochondria (Fig. 2B).A key aspect of the research design was to tune the reaction rate in such a way that at the concentration of systemic dosing, the reaction has a long half-life.However, reaction half-life would shorten proportionally upon enrichment in mitochondria, leading to quick CO release.Because TPP-mediated enrichment is known to reach several hundred-fold, the reaction rate constant can be tuned accordingly.Speci cally, the reaction rate constant between TPP-BCN and TPP-CPD (Fig. 2C) was tuned to about 0.20 M − 1 s − 1 , which gives a calculated t 1/2 of 139 h at 10 µM.Because mitochondria-enrichment is known to reach hundreds of folds, after enrichment, the reaction rate can also accelerate by hundreds of folds to allow CO release with a t 1/2 of a few hours or less.We were indeed able to demonstrate proof of principle using cell culture and animal models.[45] Then, we are interested in moving away from using the hydrophobic TPP moiety for enrichment.Thus, we turn to folate-receptor-mediated enrichment for triggered release and targeted release in cells that over-express such receptor.
First, we are interested in searching for treatment options for AKI and colitis.Folate receptor is known to be enriched in proximal tubule cells, offering a site for targeted delivery.[10] Second, macrophages are known to over express the folate receptor, [10] allowing for targeted delivery of a payload to in amed tissues including colitis.Therefore, we sought to synthesize conjugates of folate with the two respective reaction components (Fig. 2C) and assess their reactivity (reaction kinetics).

Synthesis of folate conjugated cyclopentadienone and cyclooctyne
In the designed bimolecular approach, both the click partners, dienone and dienophile (Fig. 2C), have to be conjugated with the targeting moiety, folic acid, for enrichment.We rst selected BCN as the dienophile and compound 18a as the diene.This is based on the fact that the same pair worked well after conjugation with TPP.Literature reports indicate that the α-carboxylate group of folic acid is critical for receptor binding.Therefore, the γ-carboxylate group is the preferred handle for modi cation.[11] To begin with, γ-carboxylate conjugated ethylene diamine folate (9) is synthesized by using a literature procedure [55] by reaction with Boc-protected ethylenediamine followed by TFA treatment to remove the Boc group.Then cyclooctyne (BCN) was conjugated through carbamylation using an activated carbonate group to provide compound 11 (Scheme 1).The γ-conjugation regiochemistry of compound 8, 9, and 11 is con rmed by comparing the 1 H-NMR spectra of these compounds with the corresponding literature reports.Speci cally, the 1 H-NMR γ-folate conjugates are characterized by a singlet peak of 4.28 ppm, which is exactly matching with the literature reported NMR spectra of compound 8, 9 and 11. [55,56] For the synthesis of the folate conjugate with the cyclopentadienone partner, an ethylene diamine linker was also used (19a-b).To begin with, phenylacetyl chloride was reacted with Meldrum's acid to give compound 13a-b.Subsequent decarboxylative reaction with t-butyl alcohol gave compound 14.Aldol condensation of compound 14a-b with acenaphthylene-1,2-dione followed by dehydration in acidic medium gave compound 15a-b in 72-79% yield.After removal of the t-Bu protective group from the ester, the carboxyl group was reacted with Boc-protected ethylene diamine using EDC, DMAP to give 17a-b.Subsequent Boc deprotection with TFA gave cyclopentadienone tagged with an ethylene diamine linker (18a-b), which was conjugated with NHS activated folic acid to get compound 19a-b (Scheme 2).

CO release Kinetic Studies
As discussed earlier, success of an enrichment triggered delivery approach largely depends on the reaction kinetics.As a bimolecular reaction, CO-release half-life relies on the reactivity and concentration of two individual click partners.[57,45] In a previous study, we have already shown that a second orderrate constant of 0.2 M − 1 s − 1 allows stability when dosed at single digit micromolar concentrations.However, when enriched by hundreds of folds, the reaction is proportionally accelerated to allow CO release with a half-life less than a few hours.[45] Another nding from the previous study was the improved reaction rate of the Diels-Alder reaction after conjugation.For example, the rate constant changed from 0.14 M − 1 s − 1 to 0.20 M − 1 s − 1 when the reaction partners were each conjugated with a TPP moiety.[45] Therefore, we rst need to study the reaction kinetics after folate conjugation.
Our CO prodrugs are designed in such a way that the cyclization product after CO release is uorescent and serves as way to study the kinetics for the CO release reaction.[58]Thus, the CO release kinetics by the prodrug system is studied by uorescence spectroscopy.Brie y, BCN compound at three different concentrations are allowed to react with 50 µM of a cyclopentadienone.CO release kinetics are studied by monitoring the uorescence at 464 nm (λ ex = 375 nm), which corresponds to the formation of CO released product.A plot of uorescence intensity against time provided the pseudo-rst order rate constant for each concentration of the BCN compound.Further, plotting the pseudo-rst order rate constant against the time provided the second order rate constant for the reaction (Fig. 4).
In determining the reaction rate constant, we rst examined some "control" reactions using 20 and 21 (Fig. 4A-D).This click pair yielded a rate constants of 0.097 M -1 s -1 , which is generally on the same scale as the results from earlier studies of 0.14 M -1 s -1 for a similar reaction.[45] Such results also gave us con dence to move forward with the synthesis of the folate conjugates.However, much to our disappointment, the reaction with 1.8 mM of BCN conjugate 11 and 50 µM of 19a was very slow, to the extent that the uorescence did not plateau off even after 36 h.We did not quantify the second order rate constant for the reaction between 19a and 11 because a reaction this slow would have no practical value in our designed enrichment-triggered release approach.The slow reaction was very surprising even with the understanding of the possibly signi cant effects of conjugation on reaction rate.There have been literature reports of the acceleration of Diels-Alder reaction in aqueous solution due to hydrophobic effects.[59,53] With such understanding, it is possible that the introduction of a folate moiety has substantially decreased the hydrophobicity-driven acceleration of the intended Diels-Alder reaction.With the aim of improving reaction kinetics, we designed an analog (19b) with a tri uoromethyl substituent to decrease the electron density of the cyclopentadienone moiety with the hope of increasing the reaction rate of this inverse-electron demand Diels-Alder reaction.When the reaction kinetics of pair of 19b and 11 was studied, a second order rate constant of 0.033 M -1 s -1 was obtained (Fig. 4E-H).Though this represents a signi cant improvement over 19a and 11, the reaction rate is still much slower than needed for the designed approach.It is also slower than the "parent" reaction without folate conjugation (0.097 M -1 s -1 between 20 and 21).With this rate constant, the t 1/2 is calculated to be 8418 h and 842 h when incubated at 1 and 10 mM, respectively.There would need to be close to 1000-fold increase in concentration through receptor-mediated enrichment in order for the reaction rate to be in the range of meaningful CO delivery.Otherwise, such a reaction rate is not expected to be able to overcome diffusion to make a difference.Indeed, in a preliminary imaging study using cells that are speci cally designed to probe folate receptor-mediated events,[60] we were unable to see any meaningful enrichment (Fig. S1,2).
Such results indicate the need for further optimization of the reaction rate.The signi cant difference in reaction kinetics between the two analogues underscores the signi cance of electronic factors of cyclopentadienone and its impact on CO release kinetics.The reaction rate difference between the folate conjugates and TPP conjugates also highlight the signi cant effects of the targeting moiety on reactivity.Such information combined will be helpful to future optimization work.

Conclusion
In summary, we synthesized two folate conjugated pairs for bimolecular CO prodrugs.Fluorescence studies indicate the dependence of the reaction kinetics on the electronic properties of the cyclopentadienone moiety and conjugation chemistry.Speci cally, the reaction rate for p-CF 3 substituted cyclopentadienone (19b) is signi cantly faster than the corresponding non-substituted analogue (19a).
However, folate conjugation led to a decrease in reaction rate so much so that the presumed receptormediated enrichment is not su cient to overcome the sluggish reaction rate in imaging studies.Although the design did not work out the way we envisioned, there are important lessons to learn.First, different from TPP conjugation, folate conjugation led to a decrease in reaction rate.Such ndings will be very useful in guiding future designs of targeted delivery of similar reactions in terms of the effect of hydrophilic ligands on the rate of Diel-Alder reactions.Second, the enrichment magnitude of the reaction partners used is probably nowhere near the needed 1000-fold through folate receptor-mediated endocytosis.Third, it is possible to tune the reaction rate by using various electron-withdrawing groups on the dienone to speed up the reaction.Further, the search for a new strained alkyne may also lead to a solution to the current sluggish reaction issue.Overall, the results indicate the signi cance of reaction rate difference between targeted compound and control compound for the successful application of enrichment triggered drug delivery.

Materials and Method
All solvents were of reagent grade and were purchased from Fisher Scienti c.All chemicals are of reagent grade and purchased from Sigma-Aldrich (Massachusetts, USA) or Ambeed, Inc. (Illinois, USA) or Oakwood Products, Inc. (South Carolina, USA), Column chromatography was carried out using silica gel (Sorbent Technologies (Georgia, USA) 230−400 mesh).TLC analyses were conducted on silica gel plates (Sorbent Technologies (Georgia, USA) Silica XHL TLC plates w/UV254). 1H NMR (400 MHz) and 13 C NMR (100 MHz) spectra were recorded on a Bruker Avance 400 MHz NMR spectrometer in deuterated solvent from Oakwood Products, Inc. (South Carolina, USA).Chemical shifts were reported as δ values (ppm).TMS (δ = 0.00 ppm) or residual peaks of the deuterated solvent were used as the internal reference.Mass spectrometric analyses were conducted by the Georgia State University Mass Spectrometry Facilities.The milligram scale quantities were weighed on C-33 microbalance (CAHN instruments Inc., California, USA).for the control compound dienone 20 (50 µM), but a higher concentration range of cycloalkyne 21 (1.8 mM, 2.2 mM, 2.6 mM) (Fig. 4. E-H).

Cell-Imaging Experiment
The intracellular CO release from the folate conjugated CO prodrugs were studied by using uorescent cell imaging.Literature reports indicate that the cell-culture in folate depleted medium helps in the FR over-expression (FR + cells) and facilitates increased uptake of folate conjugates.
[60]Thus, KB cells were cultured in two different media, one of the plates is cultured in normal RPMI-1540 (Gibco)) medium and other plate is cultured in folic acid depleted RPMI-1540 medium (Gibco).Both the medium is supplemented with 10% fetal bovine serum (FBS, Gibco) and 0.1% 100 x antibiotic-antimycotic solution (Gibco).To begin with, folate conjugated CO prodrug partners 11 and 19b were co-incubated with normal KB and FR + -KB cells.Fluorescence images were recorded on 4 h and 16 h time point after incubation.
Similarly, a control experiment is performed by incubating 4 h KB and FR + -KB cells with non-folate conjugated dienone 20 and cycloalkyne 21.Synthesized by a reported literature procedure.[55] Folic acid was (320 mg, 0.73 mmol) was added to 8 ml of DMSO and heated at 55 o C for 30 min to get a clear solution.The solution was cooled to room temperature and added N-hydroxy succinimide (167 mg, 1.46 mmol) and DCC (301 mg, 1.46 mmol) under nitrogen atmosphere and protected from light, stirred at room temperature for 20 h.After 20 h of stirring, the urea precipitate was ltered off and the obtained solution is used for the next step by adding triethylamine (203 µL, 1.46 mmol) followed by N-Bocethylenediamine (234 mg, 1.46 mmol) and stirred for 16 h.Once the reaction is completed, 20% acetone in diethyl ether solution is added to the reaction mixture and the resulting light-yellow precipitate is collected with the aid of centrifugation.The HPLC analysis of the reaction mixture indicates a product formation with a new peak with almost 20% of folic acid remain unreacted which was further puri ed by preparative HPLC (Shimadzu HPLC-0.1% TFA in ACN/ H20; Method 0-18 min 30%-45%, owrate 16 ml/min).The purity of the obtained product was con rmed by recording the 1   Synthesized by using a previous literature procedure.[55] Compound 8 (200 mg, 0,34 mmol) was dissolved in 5 ml of TFA and stirred at room temperature for 3 h, the reaction progress was monitored by HPLC and once the reaction is completed, the excess TFA is evaporated with the aid of 5 ml DCM and the resulting content of the ask is dissolved in 2 ml of DMF, and the product is precipitated by the slow addition of triethylamine (~ 700 µL).The light-yellow solid precipitate is collected by centrifugation to obtain compound 9 with quantitative yield.This product is used for the subsequent conjugation steps with CO prodrug partners.Synthesized by using a previous literature procedure.[55] To a solution of compound 9 (25 mg, 0.052 mmol) in 1.
Synthesized by a previously reported literature procedure.[45] To a solution of Meldrum's acid (1.0 g, 6.9 mmol) and pyridine (1.1 mL, 13.8 mmol) in dichloromethane (15 ml) under ice-cold temperature was added a solution of phenylacetyl chloride (1.278 g ,8.3 mmol.) drop by drop.After the completion of addition, the reaction was allowed to stir at room temperature, and stirred for an additional 3 h.The reaction mixture was then washed successively with 5% HCl solution and brine.The organic layer was dried with anhydrous Na 2 SO 4 , ltered and concentrated.The obtained residue was puri ed by silica gel column chromatography (Hexane: ethyl acetate = 7:3) to afford compound 13a as white solid,1030 mg, yield: 57 %.
tert-butyl 3-oxo-4-phenylbutanoate(14a) Compound 13a (500 mg, 1.90 mmol) was dissolved in 8 ml of tert-butanol and re uxed at 85 o C for 3 h.Upon completion of the reaction, the content of the ask was evaporated to dryness to obtain yellow viscous liquid with quantitative yield (422 mg).tert-butyl 8-oxo-9-phenyl-8H-cyclopenta[a]acenaphthylene-7-carboxylate (15a) A solution of compound 14a (400 mg 1.71 mmol), and acenaphthylene-1, 2-dione (373 mg, 2.05 mmol.) in THF/MeOH (3/1, v/v, 12 ml) was treated with Et 3 N (362 µL, 2.6 mmol).Then the mixture was stirred at room temperature for 3 h, after which the mixture was concentrated under vacuum, and the resulting residue was dissolved in acetic anhydride (4 mL).The resulting solution was cooled to 0 o C, and 2 drops of concentrated sulfuric acid was added.The reaction mixture was stirred for an additional 10 min at 0 o C and the reaction mixture was diluted with ethyl acetate (30 ml) and washed with NaHCO 3 solution.The organic layer was dried over anhydrous Na 2 SO 4 .After concentration, the residue was puri ed over silica gel column (Hexane: Ethyl acetate 99: 1) to yield compound 15a as purple solid, 467 mg, yield: 72%. 1  A solution of compound 15a (450 mg, 1.18 mmol) in 3 ml dichloromethane was cooled to 0 o C, to which was added tri uoroacetic acid (12 ml) as dropwise and allowed to stir for 2 h.The reaction mixture was concentrated in vacuo and used for the next step without further puri cation (363 mg).