A N-propionylsulfonamide prodrug of K-5a2 provided improved aqueous solubility and hERG inhibition

Background Having the potential disadvantages and safety risk of the use of anti-HIV-1 drug candidate K5a2 in the longterm treatment of HIV patients in mind, we set out with the goal of nding a second-generation backup compound of K-5a2 with the appropriate anti-HIV potency, signicantly reduced hERG activity, decreased induction of the CYP enzyme, and improved aqueous solubility. Herein, using a N-propionylsulfonamide prodrug strategy, we report the discovery of compound HM-1 Methods In vitro assay of anti-HIV activities in TZM-bl and MT-4 cells, metabolic stability in HLM and human plasma, measurements of water solubility and Log P, assay procedures for hERG activity, acute and subacute toxicity experiment and cytochrome P450 inhibition assay were carried out for HM-1. Results HM-1 can be rapidly hydrolyzed to parent drug K-5a2 and exhibited high potency against HIV-1 NL4 − 3 strain (EC 50 = 7.99 nM) in TZM-bl cells, HIV-1 IIIB strain (EC 50 = 2.9 nM) and HIV-1 Y181C strain (EC 50 = 5.5 nM) in MT-4 cells. And it also showed a > 70-fold improvement in aqueous solubility and presented a low acute toxicity in mice (LD 50 > 2 g•kg − 1 ); no obvious organ damage was detected in the assessment of subacute toxicity. Meanwhile, HM-1 also showed 50 times lower hERG inhibition (IC 50 = 6.39 µM) than K5a2 (IC 50 = 0.13 µM). was HM-1 appeared to be free of most of the drawbacks associated with K-5a2 and has been selected for further as an oral anti-HIV-infection agent. virus type 1; HLM, human liver microsomes. NNRTIs, non-nucleoside reverse transcriptase inhibitors; RPV, rilpivirine; SI, selection index; WT, wild-type.


Background
Non-nucleoside reverse transcriptase inhibitors (NNRTIs), which target the binding pocket of reverse transcriptase (RT), are a vital component of highly active antiretroviral therapy (HAART) regimens for the treatment of human immunode ciency virus type 1 (HIV-1) infection, [1][2][3] due to their potent antiviral activity, high selectivity, modest toxicity, and favorable pharmacokinetics. [4] K-5a2, a promising anti-HIV-1 drug candidate, was recently discovered by structure-based design in our lab. Compared to etravirine (ETR), it is gratifying that K-5a2 was 2.8 fold more potent against HIV-1 wildtype (WT) strain, and 5-fold more potent against single mutants Y188L, E138K and double mutant F227L + V106A, and almost equipotent against L100I, K103N, and Y181C, which justi es its further development as a potential clinical candidate for the treatment of HIV-1 infection (Fig. 1). [5] Despite these encouraging results, further development of K-5a2 faces many challenges for its low aqueous solubility (< 1 µg/mL), which is one of the important parameters to achieve a desired drug concentration in systemic circulation to achieve the intended pharmacological response. [5][6][7] In addition, high hERG liability (IC 50 = 0.13 µM) is an important problem encountered with K-5a2, which may engender the potential risk for cardiotoxicity. Hence, challenges associated with poor aqueous solubility and high hERG inhibition to K-5a2, highlight the need for continuous research and development of rational chemical modi cation. The application of the prodrug approach proved to be a successful strategy for improving aqueous solubility and other drug-like pro les of the parental drugs result in a positive impact on their biological e cacy. [8] Recent studies demonstrated that prodrug modi cation to improve water solubility of some parent drug containing sulfonamide group has been successfully achieved using N-propionyl moiety. [9][10][11] For example, as a potent NNRTI, GW678248 was limited into clinical trials due to its poor water solubility (0.18 µg/ml). With the aim to improve solubility, N-propionyl sulfonamide prodrug of GW678248 was discovered. GW695634 obviously enhanced aqueous solubility (92 mg/ml). And the oral bioavailability of GW695634 in rats and dogs (> 40%) was also increased. [12] Besides, the marketed parecoxib was also the N-propionyl sulfonamide prodrug of valdecoxib which is relatively insoluble in water at about 10 µg/mL. [13] Elsulfavirine, a new generation NNRTI, is the prodrug of the active compound VM 1500A that obtained its rst market authorization in Russia (Fig. 2). [14] Herein, inspired by these recent successful examples, with the aim to enhance aqueous solubility and drug-like pro les of our drug candidate K-5a2, we looked for a second-generation backup compound for K-5a2, with the appropriate anti-HIV potency, signi cantly reduced hERG activity, decreased induction of the CYP enzyme, and improved aqueous solubility. Consequently, the N-propionylsulfonamide prodrug strategy was employed in a structural modi cation. The synthesis and characterization of prodrug HM-1 ( Fig. 3), including antiviral activity, acute toxicity, subacute toxicity, hERG inhibition, and CYP enzymatic inhibitory activity, are presented in this paper.

In vitro anti-HIV assay in MT-4 cells
This kind of anti-HIV activity and cytotoxicity assay were conducted in MT-4 cells named MTT method. 10 different concentrations of K-5a2, HM-1 and reference drugs were added in the corresponding volumes to formulate two series of triplicate wells to evaluate anti-HIV activity simultaneously on mock-and HIVinfected cells. Serial 5-fold dilutions of relevant compounds were added in at-bottomed 96-well microtiter trays. HIV stock at 100-300 CCID 50 or culture medium was added to the infected or mockinfected wells, and untreated HIV-and mock-infected cell samples were as controls. After the MT-4 cells being centrifuged, the supernatant was deserted. MT-4 cells were resuspended and transfer 50 µL volumes for infecting. Five days later, MTT assay was conducted on mock-and HIV-infected cells. All data were calculated by the median absorbance value. EC 50 was concentration achieving 50% protection against the cytopathic effect of the virus in infected cells, CC 50 was the concentration reducing the absorbance (OD540) of the mock-infected control sample by 50%.

Metabolic stability in HLM and human plasma
To validate our hypothesis, prodrug HM-1 was further assessed for its ability to undergo conversion to the parent drug K-5a2. HLM stability assay was primarily carried out. HM-1 incubated with microsome solution for 10 minutes, and then 2 mM MgCl 2 was added. Each bioanalysis plate was sealed after adding into stop solution for LC-MS/MS analysis.
We also tested the stability of HM-1 in human plasma. Blank plasma spiked with HM-1 (100 µM) was incubated at 37 °C. Stop solution was added at 0, 10, 30, 60 and 120 min. The supernatant was transferred from each well and mixed with ultra pure water for LC-MS/MS analysis.

Measurements of water solubility and Log P
Water solubility was measured in phosphate buffer at pH 7.4 and 2.0 by using a HPLC-UV method. [16] HM-1 or K-5a2 was dissolved in DMSO, and then spiked into 1 mL puri ed water. After ultrasonic for 2 h, the saturated solution was ltrated through a lter membrane and prepared two samples. LC-20AT HPLC-UV (SHIMADZU) system was used to compare the peak area of the saturated solution with a standard curve plotted peak area. Stock solution of HM-1 or K-5a2 were prepared for HPLC analysis. The P (partition coe cient) value was obtained by the peak area ratios in n-octanol and in H 2 O.

Assay procedures for hERG activity
The hERG channels were stably expressed in Chinese Hamster Ovary (CHO) cells that had no endogenous IKr, so CHO-hERG cells obtained from Sophion Biosciences were used in this assay. [17] CHO-hERG cells were maintained at 37 °C (± 2 °C) under 5% CO 2 (4%~8%) in a high humidity incubator. The whole-cell patch-clamp technique was used to record hERG current controlled with Patchmaster Pro software at room temperature and output signals from the patch-clamp ampli er was digitized and low-pass ltered at 2.9 KHz. Electrode was recorded by a micropipette lled with ICS, which was drawn out from borosilicate glass to give a tip resistance between 3 ~ 5 MΩ. The initial holding potential was − 80 mV, and then the voltage was increased to + 60 mV for 850 ms to open the hERG channels. After that, the voltage was decreased to -50 mV for 1275 ms to cause a "rebound" or tail current, before the voltage was decreased to the holding potential (-80 mV), the peak tail current was measured and collected for data analysis. This command voltage protocol was repeated every 15 s continuously.
Acute and subacute toxicity experiment All animal treatments were performed strictly in accordance with the institutional guidelines of Animal Care and Use Committee at Shandong University, after gaining approval from the Animal Ethical and Welfare Committee (AEWC). Healthy male Kunming mice (17 ~ 22 g) were randomly divided into three groups: blank group, K-5a2 group, and HM-1 group. Each group consisted of 5 mice. K-5a2 and HM-1 were suspended in dimethyl sulfoxide (DMSO): normal saline: polyethylene glycol (PEG) 400 = 5%: 25%: 70% at concentrations of 200 mg·mL − 1 , respectively, and all mice had been fasted for 12 h, and then the mice were given by gavage a dose of 2000 mg/kg of blank solvent, K-5a2 and HM-1 respectively for the following 7 days. The death, body weight and behavioral abnormalities were monitored.
Another group of healthy male Kunming mice were also randomly divided into three groups: blank group, K-5a2 group, and HM-1 group. Each group consisted of 5 mice. Suspension of K-5a2 and HM-1 were prepared (DMSO: sodium carboxymethylcellulose: H 2 O = 3%: 77%: 20%), respectively, with concentrations of 6 mg·mL − 1 , and administered intragastrically after the mice had been fasted for 12 h. Before dosages of 60 mg·kg − 1 of blank solvent, K-5a2 and HM-1 were administered respectively to three groups, and the body weight of the mice was measured. All of the mice were dissected on the last day, and the heart, liver, spleen, lung, and kidney were extracted. These organs were then examined by HE to see if any damage had occurred.
Kunming mice were purchased from the animal experimental center of Shandong University. Animals were maintained during the experiment in standard conditions of 12 hours light-dark cycle, food and water ad libitum, temperature 24 °C, humidity above 55%. When the experiments nished, all animals were euthanized using the method of spinal dislocation. The details were as follow: grasp the mouse with the right hand and pull it backward hard, while pressing the mouse's head with the thumb and index nger of the left hand to pull the spinal cord and brain marrow to break, and the mouse died immediately.
The target compound HM-1 was completely characterized by the spectral means including 1 H NMR, 13 C NMR and MS. The purity was > 95% as determined by analytical HPLC.

Anti-HIV activity assay in TZM-bl cells
To examine the antiviral activity of HM-1, along with K-5a2, reduction in luciferase gene expression was detected in TZM-bl cells. As exhibited in Table 1, the new compound HM-1 still revealed high potency suppressing HIV-1 NL4 − 3 strain at a nanomolar concentration (EC 50 = 7.99 nM), although it was inferior to that of the parent drug K-5a2 (EC 50 = 1.17 nM). The CC 50 values of HM-1 and K-5a2 were greater than 210 nM, which illustrated that both compounds had low cytotoxicity in TZM-bl cells. Anti-HIV activity assay in MT-4 cells Furthermore, we also evaluate the anti-HIV-1 potency against wild type (WT) and common mutant strains in MT-4 cells. The results in Table 2 showed that HM-1 still exhibited high antiviral activity against HIV-1 IIIB with EC 50 values of 2.9 nM, which was better than the reference drug ETR (EC 50 = 3.0 nM). Besides, the cytotoxicity of HM-1 (CC 50 = 75.7 µM, SI = 26161) in MT-4 cells was far lower than ETR (CC 50 > 4.6 µM, SI > 1546). And HM-1 didn't show inhibitory against HIV-2 strain. As showed in Table 3, HM-1 exhibited higher anti-HIV-1 activities towards mutant strains than ETR except RES056 strain. Notably, the potency against Y181C strain of HM-1 (EC 50 = 5.5 nM) was superior to the parent K-5a2 (EC 50 = 5.9 nM). Metabolic stability assays Since the prodrug Parecoxib with the N-propionyl moiety was rapidly hydrolyzed by the liver to valdecoxib, [18] the stability of N-propionylsulfonamide prodrug HM-1 in the presence of HLM was determined. The results demonstrated that HM-1 was rapidly converted to the parent drug K-5a2 in 10 min. In addition, HM-1 was highly stabilized in human plasma via plasma stability assay (Table 4).

Solubility and lipophilic e ciency
As well known, poor water solubility and hydrophilicity/lipophilicity will in uence drug metabolism and distribution, K-5a2 displayed poor solubility which can limit its further development as a drug. HPLC was used to test the solubility and logP of K-5a2 and HM-1, and other drug-like properties such as MW (molecular weight) and nON (no. of hydrogen bond acceptors) were predicted (Table 5).Surprisingly, HM-1 showed a > 70-fold improvement in aqueous solubility compared to K-5a2, which supported that the strategy of N-propionylsulfonamide can improve water solubility effectively.  [19]. c See Ref [20]. d Measured in phosphate buffer by HPLC in duplicate Assessment of hERG activity Acquired long QT syndrome caused by inhibition of the voltage-dependent ion channel, which is encoded by human ether-a-go-go-related gene (hERG), is a critical adverse effect of noncardiovasular therapeutic drugs. [21] As we all know, decreasing alkalinity is an important strategy for structure optimization of lead compounds to reduce the inhibitory activity of hERG. [22] Through predicting by ChemDraw, the alkalinity of HM-1 was decreased, which revealed that HM-1 may have lower hERG activity. Indeed, in vitro manual patch-clamp electrophysiology was used to evaluate the activity against the hERG potassium channel to estimate the potential risk for cardiotoxicity, [23] and the result veri ed our prediction exactly.
As shown in Fig. 4, HM-1 showed an IC 50 value of 6.39 µM against the potassium channel, which was nearly 50 times higher than that of K-5a2 (IC 50 = 0.13 µM). Although the hERG IC 50 value of HM-1was not greater than 10 µM, it was also 12 times better than rilpivirine (RPV, IC 50 = 0.5 µM).
[24] Therefore, the hERG inhibitory activity of HM-1 was much lower than that of K-5a2, suggesting that the potential cardiotoxicity was signi cantly diminished.
Assessment of acute and subacute toxicity.
Regarding the relatively low cytotoxicity of K-5a2, a single-dose acute toxicity trial of HM-1 in healthy Kunming mice was carried out. One mouse died after intragastric administration of K5a2 and HM-1 at a dose of 2000 mg/kg. However, no signi cant behavior abnormalities (lethargy, clonic convulsion, anorexia, and ruffed fur) were observed and the body weights changes was the same as in the blank group (Fig. 5A). Therefore, HM-1 can be tolerated at a dose of 2000 mg/kg with low toxicity.
To further evaluate the in vivo safety of HM-1, the subacute toxicity was investigated in healthy mice. No mice died, and no behavioral abnormalities occurred during the treatment period of mice treated po with 60 mg/kg·day − 1 of HM-1 for 2 weeks. The body weight of mice in the HM-1 group increased gradually and the mice grew normally as in the K-5a2 group (Fig. 5B). Additionally, organs toxicity experiments were carried out by hematoxylin − eosin (HE) staining after dissecting all the mice. As depicted in Fig. 5C, the heart, liver, and spleen of the mice were not damaged with either HM-1 or K-5a2, but after treatment with K-5a2, it was remarkable that there were alveolar hemorrhage and alveolar interstitial thickening (red arrows) in the lung and glomerular edema (blue circles) in the kidney, which were obviously improved with HM-1. The subacute toxicity assessment results indicated that HM-1 possessed favorable in vivo safety pro les.
In vitro effects of HM-1 on CYP enzymatic inhibitory activity.
In the liver, the majority of drugs are metabolized by enzymes like cytochrome P450 enzymes, whose activity can be inhibited or induced by drugs. Therefore, when co-administering different drugs, therapeutic effect may be decreased or increased, this is called metabolism-mediated drug − drug interactions (DDI). Previous studies disclosed that ETR is an inducer of CYP3A4 and inhibitor of CYP2C9 and CYP2C19, and DDI will occur when ETR is co-administrated with other anti-HIV-1drugs, which limits the selection of concomitant antiretrovirals. Hence, HM-1 was evaluated for its ability to inhibit CYP drugmetabolizing enzymes in vitro.
As shown in Table 6 and Table 7, the CYP3A4 inhibition of HM-1 (IC 50 > 50 µM) was far superior to K-5a2 (IC 50 = 18.9 µM), which revealed that HM-1 might decrease toxicity and DDI caused by CYP3A4 inhibition in anti-HIV infection. In addition, HM-1 displayed no CYP enzymatic inhibition of CYP1A2 and CYP2D6.
Although the CYP2C9 and CYP2C19 inhibitions (IC 50 < 3 µM) of HM-1 were inferior to K-5a2, [25] which may induce side effects in the liver and probably result in DDI when co-administered with other different drugs, the perfect result of CYP3A4 inhibition was still worthy to emphasized.

Discussion
A high-quality drug should emerge balanced e cacy against its therapeutic targets, safety, physicochemical properties and ADME properties. [26] In order to improve key drug-like properties, namely the solubility, an effort to identify prodrugs was undertaken. This has led to the discovery of a N-propionylsulfonamide prodrug HM-1. HM-1 could be rapidly converted to the parent drug K-5a2 in the liver, and exhibit remarkable potency against HIV-1 NL4 − 3 strain with an EC 50 value of 7.99 nM.
Furthermore, HM-1 also showed excellent potency against HIV-1 IIIB (EC 50 = 2.9 nM) and HIV-1 Y181C strain (EC 50 = 5.5 nM).The physicochemical evaluation showed that HM-1 exhibited a 70-fold increase in aqueous solubility (69.66 µg/mL) when compared to that of K-5a2 ( < < 1 µg/mL). The hERG liability (IC 50 = 6.39 µM) of HM-1 was signi cantly reduced by 50 times, compared to K-5a2 (IC 50 = 0.13 µM), which would allow higher doses to be used in the clinic thereby realizing full potential of this compound. In addition, HM-1 showed a low acute toxicity in mice (LD 50 > 2 g•kg − 1 ) as well as no obvious organ damage detected in the assessment of subacute toxicity.

Conclusion
In summary, the N-propionylsulfonamide prodrug approach has wide applicability in sulfonamide-bearing compounds and provides an effective means of augmenting the therapeutic potential of our anti-HIV-1 drug candidate K-5a2. What can not be ignored is that HM-1 may induce some side effects in the liver and DDI probably occur when co-administrated with other drugs with an IC 50 value less than 3 μM against CYP2C9 and CYP2C19. Thus, further studies should be carried out to discover potential prodrugs with improved physiochemical and PK properties, as well as decreased induction of CYP enzymes.

Declarations Declarations
Author declares that there is no con ict of interests.

Ethics approval and consent to participate
We performed this study with an authorized standard method to detect acute and subacute toxicity using mice and accredited by Laboratory Animal Ethical and Welfare Committee of Shandong University Cheeloo College of Medicine.

Consent to publish
Not applicable.
wild-type and E138K mutant virus with significantly improved water solubility and favorable safety profiles, Journal of medicinal chemistry, 62 (2019) 2083-2098.

Figure 2
Structures of successful N-propionyl sulfonamide prodrugs Figure 3 The design concept of K-5a2 prodrug. Activity of HM-1 and K-5a2 against hERG potassium channel in CHO-hERG cells.

Figure 5
A. Body weight of all mice in three groups (g)-time (day) in acute toxicity experiment. B. Body weight of all mice in three groups (g)-time (day) in subacute toxicity experiment. C. HE of different organs of normal mice with K-5a2 (60 mg/kg) and HM-1 (60 mg/kg). Scale bars represent 50 μm.