**1. Introduction**

The exploration of marine secondary metabolites only began in the early 1950s [1] with the landmark identification of the two nucleosides spongothymidine and spongouridine from the Caribbean marine sponge *Cryptotethia crypta* [2,3]. Since then, marine natural product discovery has increased annually and has accelerated over the last two decades [4]. By 2010 more than 15,000 new marine natural products were discovered with 8368 new compounds recorded in a decade between 2001 and 2010 [5]. Approximately 75% of marine natural products were isolated from marine invertebrates [4] from which sponges, tunicates, bryozoans or molluscs have provided the majority of the compounds involved in clinical or preclinical trials [5]. It is expected that the discovery of marine natural products will provide new and improved therapeutics for human illnesses, along with other innovative products for other industrial activities (e.g., nutraceutics and biotechnology) [6].

To maximize the chance to discover new marine natural products, a library of fractions from rare or restricted marine invertebrates was generated from the Nature Bank database [7]. Qualitative analysis of lead-like enhanced fractions of samples in the selected library by LC-MS dual electrospray ionization coupled with PDA and ELSD detectors [8–10] indicated that the marine tunicate *Cnemidocarpa stolonifera* contained two different compound classes in two different fractions. The first class showed strong MS signals in a positive mode and strong UV absorptions at 380 nm while the second one displayed strong MS signals in a negative mode and strong UV absorptions at 280 nm. Mass-guided isolation of the marine tunicate *C. stolonifera* extract led to the isolation of two known pyridoacridine alkaloids, 11-hydroxyascididemin (**4**) and cnemidine A (**5**) [11], accounting for the first compound class. For the second class of compounds, three new taurine amide derivatives, stolonines A–C (**1**–**3**), were isolated.

Taurine is found at high concentration in mammalian plasma and tissues [12]. This compound has been known to exhibit a large spectrum of physiological functions in the liver, kidney, heart, pancreas, retina, and brain [13,14]. Its depletion is associated with several disease conditions such as diabetes, Parkinson's, Alzheimer's, cardiovascular diseases, and neuronal damages in the retina [13]. So far, 81 taurine derivatives from natural sources have been reported of which 47 compounds are amides produced between taurine with fatty acids, steroid acids or other aromatic acid residues [15]. The occurrence of the conjugates of taurine with 3-indoleglyoxylic acid, quinoline-2-carboxylic acid and β-carboline-3-carboxylic acid present in stolonines A–C (**1**–**3**), respectively, has not been previously reported. This paper describes the isolation, structure elucidation and total synthesis of stolonines A–C (**1**–**3**) from the marine tunicate *C. stolonifera*. Cytotoxicity of **1**–**3** against human prostate cancer PC3 cell line was evaluated and the result indicated that they were not active up to 20 μM. However, an immunofluorescence assay on PC3 cells revealed that compounds **1** and **3** increased cell size, induced mitochondrial texture elongation and caused apoptosis in PC3 cells.

### **2. Results and Discussion**

The freeze-dried *C. stolonifera* tunicate was sequentially extracted with *n*-hexane, dichloromethane (DCM), and methanol (MeOH). The DCM/MeOH extracts were then combined and chromatographed using C18 bonded silica HPLC MeOH/H2O/0.1% trifluoroacetic acid (TFA) to yield three new alkaloids, stolonines A–C (**1**–**3**) together with two known alkaloids, 11-hydroxyascididemin (**4**) and cnemidine A (**5**) (Figure 1).

**Figure 1.** Structures of stolonines A-C (**1**–**3**), 11-hydroxyascididemin (**4**) and cnemidine A (**5**) isolated from the tunicate *C. stolonifera*.

Stolonine A (**1**) was obtained as a white amorphous solid. The (−)-HRESIMS spectrum displayed a molecular ion [M−H]<sup>−</sup> at *m*/*z* 295.0390, which was consistent with the molecular formula C12H12N2O5S. The IR spectrum indicated the presence of an S=O stretching band at 1205 cm−<sup>1</sup> [16]. A 1H NMR spectrum of **1** showed two exchangeable protons (δ<sup>H</sup> 12.20 and 8.78 ppm), five aromatic protons (δ<sup>H</sup> 8.82, 8.22, 7.52, 7.26 and 7.24 ppm) and two methylenes (δ<sup>H</sup> 3.50 and 2.66 ppm). Further analysis of the 13C NMR and gHSQCAD spectra indicated that the molecule contained two carbonyls (δ<sup>C</sup> 181.6 and 162.7 ppm), eight aromatic carbons (δ<sup>C</sup> 138.5, 136.2, 126.2, 123.3, 122.4, 121.3, 112.5 and 112.1 ppm) and two methylenes (δ<sup>C</sup> 49.8 and 35.5 ppm) (Table 1). *J* coupling constants of aromatic

protons H-4 (δ<sup>H</sup> 8.22, dd, 1.8, 7.2 Hz), H-5 (δ<sup>H</sup> 7.24, dt, 1.8, 7.2 Hz), H-6 (δ<sup>H</sup> 7.26, dt, 1.8, 7.2 Hz) and H-7 (δ<sup>H</sup> 7.52, dd, 1.8, 7.2 Hz) and their COSY correlations were characteristic of a 1,2-disubstituted benzene ring (**a**, Figure 2). A gCOSY spectrum displayed correlations from the exchangeable proton H-1 (δ<sup>H</sup> 12.20, br.s) to H-2 (δ<sup>H</sup> 8.82, d, 3.0 Hz) and also from H-11 (δ<sup>H</sup> 3.50, dd, 6.0, 6.6 Hz) to the triplet exchangeable proton H-10 (δ<sup>H</sup> 8.78, t, 5.4 Hz) and H-12 (δ<sup>H</sup> 2.66, t, 6.6 Hz) facilitating the establishment of two other spin systems, –NH–CH= (**b**, Figure 2) and –NH–CH2–CH2– (**c**, Figure 2) respectively. The aromatic carbon C-3 (δ<sup>C</sup> 112.1 ppm) was attached to C-2 (δ<sup>C</sup> 138.5 ppm) in the moiety **b** determined by a HMBC correlation from H-2 to C-3. A HMBC correlation from H-2 to C-7a (δ<sup>C</sup> 136.2 ppm) supported the linkage from **a** to **b** at C-7a (Figure 2). Both protons H-10 and H-11 in the moiety **c** showed HMBC correlations with a carbonyl carbon at δ<sup>C</sup> 162.7 ppm suggesting the connection of this carbonyl to N-10 to form an amide bond (**c**, Figure 2). Two methylene signals at δ<sup>H</sup> 3.50 and 2.66 ppm corresponding to carbons C-11 (δ<sup>C</sup> 35.5 ppm) and C-12 (δ<sup>C</sup> 49.8 ppm) as well as the relatively downfield resonance of the methylene C-12 were diagnostic of methylenes in a taurine moiety (**c**, Figure 2) [17–19].



a 1H NMR at 600 MHz referenced to residual DMSO solvent (δ<sup>H</sup> 2.50 ppm) and 13C NMR at 150 MHz referenced to residual DMSO solvent (δ<sup>C</sup> 39.52 ppm).

**Figure 2.** Partial structures (**a**, **b** and **c**) of **1** and their key HMBC correlations.

No HMBC correlation from any proton to the carbonyl C-8 (δ<sup>C</sup> 181.6 ppm) was observed when HMBC experiments were performed and optimized with different *J*HC couplings. Therefore, two different isomers **1-I** and **1-II** were conceivable from these data (Figure 3). Detailed HMBC analysis showed that H-2 had a HMBC correlation with C-3a (δ<sup>C</sup> 126.2 ppm). This suggested that **1** was favorable to **1-I** since the H-2 to C-3a correlation in **1-I** was a three-bond coupling while the H-2 to C-3a correlation in **1-II** was a four-bond coupling (Figure 3).

**Figure 3.** Two possible structures **1-I** and **1-II** of **1** (**1-I** and **1-II** are possible structural isomers of **1**).

Density functional theory (DFT) NMR calculations together with the DP4 probability analysis, which have recently emerged as powerful tools for the determination of structures that lack sufficient NMR characterization or contain unusual constituents or connectivities [20–25], were employed to verify the assigned structure for **1**. Theoretical 1H and 13C chemical shifts were calculated at the mPW1PW91/6-31G(d)//B3lyp/6-31G(d) levels. Calculated NMR data of **1-I** and **1-II** were then compared with the experimental NMR data based on the corrected mean absolute error (CMAE) and the DP4 probability (Computational Details, Experimental Section) to determine which of the isomers fit with the experimental data. The results (Table 2) indicated that both 13C and 1H chemical shifts of the isomer **1-I** showed lower CMAE compared to those of the isomer **1-II**. Significant differences were observed in DP4 probabilities. In particular, **1-I** had 100% probability for 13C chemical shifts and 85.3% probability for 1H chemical shifts while **1-II** only occupied 0% for 13C chemical shifts and 14.7% for 1H chemical shifts (Table 2). These probabilities indicated that the assignment of the isomer **1-I** for **1** was at a high level of confidence [20]. Therefore, the structure of **1** was suggested as the isomer **1-I**.


**Table 2.** Comparison of experimental and calculated 13C and 1H chemical shifts for **1** in DMSO-*d6*.

Compound **1-I** was synthesized by coupling 3-indoleglyoxylic acid (**6**) with taurine (Scheme 1) in the presence of *N*-(3-dimethylaminopropyl)-*N* -ethylcarbodiimide (EDCI), *N*-hydroxybenzotriazole (HOBt) and dimethylformamide (DMF) at room temperature (rt) for 21 h (64% yield) [26]. Both 1H and 13C NMR data of the synthetic **1-I** completely matched those of the natural product **1** (Table 3) confirming that **1-I** was the structure of **1**. Therefore, compound **1** was defined as a new alkaloid with a trivial name stolonine A.

**Scheme 1.** Synthesis of **1**. (**a**) Taurine, EDCI, HOBt, DMF, rt, 21 h, 64%.

**Table 3.** 1H and 13C NMR chemical shifts of natural and synthetic products of **1**–**3** a.


a 1H NMR at 600 MHz at 30 ◦C referenced to residual DMSO solvent (δ<sup>H</sup> 2.50 ppm) and 13C NMR at 150 MHz at 30 ◦C referenced to residual DMSO solvent (δ<sup>C</sup> 39.52 ppm); b 1H NMR at 500 MHz at 30 ◦C referenced to residual DMSO solvent (δ<sup>H</sup> 2.50 ppm) and 13C NMR at 125 MHz at 30 ◦C referenced to residual DMSO solvent (δ<sup>C</sup> 39.52 ppm).

Stolonine B (**2**) was purified as a white amorphous solid. A molecular ion [M − H]<sup>−</sup> at *m*/*z* 279.0441 in the (−)-HRESIMS spectrum indicated that **<sup>2</sup>** had a molecular formula C12H12N2O4S. A 1H NMR spectrum of **2** displayed one exchangeable proton (δ<sup>H</sup> 9.29 ppm), six aromatic protons (δ<sup>H</sup> 8.55, 8.15, 8.08, 8.06, 7.87 and 7.71 ppm) and two methylenes (δ<sup>H</sup> 3.65 and 2.72 ppm). 13C NMR combined with gHSQCAD spectra indicated that **2** contained nine aromatic carbons including six tertiary carbons (δ<sup>C</sup> 137.8, 130.4, 128.7, 128.1, 128.0 and 118.6 ppm) and three quaternary carbons (δ<sup>C</sup> 150.1, 146.0 and 129.2 ppm), one carbonyl (δ<sup>C</sup> 163.5 ppm) and two methylenes (δ<sup>C</sup> 50.1 and 35.9 ppm) (Table 4). Compared to **1**, compound **2** displayed similar COSY and HMBC correlations from the two methylene protons resulting in the assignment of a taurine moiety. A 1,2-disubstituted benzene ring was deduced on the basis of *J* coupling constants and COSY correlations of H-5 (δ<sup>H</sup> 8.08, d, 7.8 Hz), H-6 (δ<sup>H</sup> 7.71, t, 7.2, 7.8 Hz), H-7 (δ<sup>H</sup> 7.87, t, 7.2, 7.8 Hz) and H-8 (δ<sup>H</sup> 8.06, d, 7.8 Hz). Two doublet aromatic protons H-3 (δ<sup>H</sup> 8.15 ppm) and H-4 (δ<sup>H</sup> 8.55 ppm) coupling together with a *J* coupling value of 8.4 Hz showed HMBC correlations from H-3 to C-4a (δ<sup>C</sup> 129.2 ppm) and from H-4 to C-5 (δ<sup>C</sup> 128.7 ppm) and C-8a (δ<sup>C</sup> 146.0 ppm) facilitating a connection from **c** to **b** at C-4. Two quaternary carbons C-2 (δ<sup>C</sup> 150.1 ppm) and C-8a (δ<sup>C</sup> 146.0 ppm) were linked through a nitrogen atom N-1 based on their downfield resonances, which are characteristic for the imine carbons [27]. According to a HMBC correlation from H-4 to the imine C-2, C-2 was attached to C-3 leading to the establishment of a quinoline ring system.


**Table 4.** NMR data for **2** in DMSO-*d6* a.

a 1H NMR at 600 MHz referenced to residual DMSO solvent (δ<sup>H</sup> 2.50 ppm) and 13C NMR at 150 MHz referenced to residual DMSO solvent (δ<sup>C</sup> 39.52 ppm).

Due to lack of HMBC correlations from H-3 (δ<sup>H</sup> 8.15 ppm) to C-9 (δ<sup>C</sup> 163.5 ppm) or from H-10 (δ<sup>H</sup> 9.29 ppm) to C-2 to obtain unequivocal evidence of the final structure, total synthesis of **2** was undertaken to verify the structure assignment (Scheme 2). *o*-Nitrobenzaldehyde (**7**) was reduced with 4.5 equivalents (eq.) of iron powder in the presence of 0.05 mol of HCl (aqueous, aq.) in ethanol (EtOH) under reflux in 40 min. Methyl pyruvate and potassium hydroxide powder were then added and the condensation reaction was under reflux in additional 1.5 h to obtain quinoline-2-carboxylic acid (**8**) (55% yield) [28]. This compound was coupled with taurine using EDCI/HOBt in DMF at rt for 48 h to produce **2** with a yield of 75%. The NMR data of the synthetic compound **2** (Table 3) was identical to that of the natural product **2** confirming the structure assignment of **2** as a new taurine amide. Thus, the structure of **2**, stolonine B, was established.

**Scheme 2.** Synthesis of **2**. (**a**) (i) Fe, HCl, EtOH, reflux, 40 min; (ii) methyl pyruvate, KOH, reflux, 1.5 h, 55% (**b**) Taurine, EDCI, HOBt, DMF, rt, 48 h, 75%.

Stolonine C (**3**) was isolated as a white amorphous solid. Its (−)-HRESIMS spectrum showed a signal for [M <sup>−</sup> H]<sup>−</sup> at *<sup>m</sup>*/*<sup>z</sup>* 318.0550 indicating a molecular formula C14H13N3O4S to be assigned to **<sup>3</sup>**. A 13C NMR spectrum combined with 2D NMR data indicated that compound **<sup>3</sup>** had 11 aromatic carbons including six tertiary carbons (δ<sup>C</sup> 132.2, 128.2, 122.3, 120.0, 113.9 and 112.3 ppm), five quaternary carbons (δ<sup>C</sup> 141.1, 139.6, 137.1, 128.7 and 121.0 ppm), one carbonyl (δ<sup>C</sup> 164.3 ppm) and two methylenes (δ<sup>C</sup> 50.4 and 35.7 ppm) (Table 5). COSY and HMBC correlations confirmed **3** had a taurine moiety (**a**, Figure 4) and a 1,2-disubstituted benzene ring (**b**, Figure 4), which were similar to those in **1** and **2**. HMBC correlations from a singlet proton H-1 (δ<sup>H</sup> 8.89 ppm) to C-3 (δ<sup>C</sup> 139.6 ppm), C-4a (δ<sup>C</sup> 128.7 ppm) and C-9a (δ<sup>C</sup> 137.1 ppm) and from a singlet proton H-4 (δ<sup>H</sup> 8.84 ppm) to C-9a indicated the presence of a 2,4,5-trisubstituted pyridine ring (**c**, Figure 4). HMBC correlations from an exchangeable proton H-9 (δ<sup>H</sup> 11.96 ppm) to C-4a, C-9a, C-4b (δ<sup>C</sup> 121.0 ppm) and C-8a (δ<sup>C</sup> 141.1 ppm) supported the connection of **b** to **c** forming a β-carboline ring system. A HMBC key correlation from H-4 to C-10 (δ<sup>C</sup> 164.3 ppm) allowed the connection of **c** to **a** at C-3 (δ<sup>C</sup> 139.6 ppm). Therefore, the structure of stolonine C (**3**) was determined as shown in Figure 4.


**Table 5.** NMR data for **3** in DMSO-*d6*.

a 1H NMR at 900 MHz at 25 ◦C referenced to residual DMSO solvent (δ<sup>H</sup> 2.50 ppm) and 13C NMR at 225 MHz at 25 ◦C referenced to residual DMSO solvent (δ<sup>C</sup> 39.52 ppm); b 1H NMR at 600 MHz at 30 ◦C referenced to residual DMSO solvent (δ<sup>H</sup> 2.50 ppm) and 13C NMR at 150 MHz at 30 ◦C referenced to residual DMSO solvent (δ<sup>C</sup> 39.52 ppm); c 13C chemical shifts obtained from correlations observed in gHSQCAD and gHMBCAD spectra.

**Figure 4.** Partial structures (**a**, **b** and **c**) of **3** and their key HMBC correlations.

Total synthesis of **3** was performed using L-tryptophan methyl ester (**9**) as a starting material (Scheme 3). The L-tryptophan methyl ester was treated with formaldehyde (37% aqueous) in a mixture of MeOH and HCl 0.1 N (ratio 10:1) at rt for 16 h to give methyl (3*S*)-1,2,3,4-tetrahydro-β-carboline-3-carboxylate [29]. The crude methyl (3*S*)-1,2,3,4-tetrahydro-β-carboline-3-carboxylate was further oxidized by activated manganese (IV) oxide (MnO2) in benzene (C6H6) under reflux for 5 h yielding a crude methyl β-carboline-3-carboxylate [30]. This ester was then hydrolysed in a mixture of aq. NaOH 20% and MeOH (ratio 1:4) to provide β-carboline-3-carboxylic acid (**10**) (10% yield in three steps) [30]. Amide coupling of **10** and taurine was performed using EDCI/HOBt in DMF at rt in 16 h to produce **3** with a yield of 40%. The synthetic product **3** had superimposable 1H and 13C NMR data with stolonine C (**3**) confirming its structure assignment (Table 3).

**Scheme 3.** Synthesis of **3**. (**a**) (i) HCHO, MeOH, HCl 0.1 N, rt, 16 h; (ii) MnO2, C6H6, reflux, 5 h; (iii) NaOH 20%, MeOH, reflux, 45 min (10% in three steps); (**b**) Taurine, EDCI, HOBt, DMF, rt, 16 h, 40%.

Biosynthesis of stolonines A–C (**1**–**3**) is shown in Scheme 4. Tryptophan (**11**) has been known as a precursor of 3-indoleglyoxylic acid (**6**), quinoline-2-carboxylic acid (**8**) and β-carboline-3-carboxylic acid (**10**) [31]. The acids **6**, **8** and **10** undergo amide bond formation with taurine either by non-ribosomal peptide synthase [32,33] or by acyl-CoA:amino acid *N*-acyltransferase 1 [34,35] to produce stolonines A–C (**1**–**3**).

**Scheme 4.** Biosynthesis of stolonines A–C (**1**–**3**). **a**: transaminase; **b**: decarboxylase; **c**: aldehyde dehydrogenase; **d**: hydroxylase; **e**: dehydrogenase; **f**: taurine + non-ribosomal peptide synthase or acyl-CoA:amino acid *N*-acyltransferase 1; **g**: tryptophan 2,3-dioxygenase or indoleamine 2,3-dioxygenase; **h**: kynurenine formamidase or arylformamidase; **i**: kynurenine aminotransferases; **j**: reductase; **k**: glyoxylic acid.

The two known compounds, 11-hydroxyascididemin (**4**) and cnemidine A (**5**), were identified by NMR comparisons with those in the literature [11].

Cytotoxic evaluation of compounds **1**–**3** against PC3 cells using the Alamar blue assay indicated that these compounds inhibited the growth of PC3 cells at only 19%, 14%, and 26%, respectively, at their maximum tested concentration of 20 μM. In order to explore whether these compounds have any effect on cellular organelles in PC3 cells, an immunofluorescence assay with three markers for cell membrane, nuclei, and mitochondria was performed. A high-content imaging system was used to image and analyze the data. Compared to vehicle, compounds **1** and **3** had effects on cell morphology, nuclei and mitochondria while compound **2** showed no or very weak effects (Figure 5A). In general, the influence of **1** and **3** on PC3 cells was similar and clearly observed in cell morphology, nuclear, and mitochondrial intensities as well as mitochondrial texture in a dose-dependent manner. In particular, cells treated with **1** and **3** increased cell size and induced cells to become longer shown by the increase in cell, mitochondrial and nuclear area, and the decrease in cell roundness. Compounds **1** and **3** also caused mitochondrial texture to become larger and elongated compared to **2** and DMSO. The increase in nuclear and mitochondrial intensities, which displayed brighter blue-whitish and brighter red-whitish fluorescent appearance respectively in the images (Figure 5B), indicated that the nuclei and mitochondria had more packed masses [36]. The brighter blue-whitish also demonstrated the condensation of chromatins [37], which has been considered as a hallmark feature of apoptotic pathway of programmed cell death [38,39]. The phenotype of nuclei suggested that compounds **1** and **3** affected the PC3 cell death via apoptosis. Moreover, at the non-toxic doses, compounds **1** and **3** also increased the cell size and induced mitochondrial texture elongation. How these compounds influence the PC3 cell morphology and mitochondria is something that warrants further investigation.

**Figure 5.** (**A**) Phenotypic profiles for cell proliferation study of **1**–**3** (results were normalized by a vehicle DMSO); (**B**) Representative images of the PC3 cells treated with 20 μM of **1**–**3** and DMSO in three channels: Hoechst 33,342 (Blue), MitoTracker (Red) and CellMask (Yellow) (scale bar: 50 μm); chromatin condensation (white arrow), mitochondrial texture elongation (green arrow) and larger cell size (cyan arrow) compared with vehicle DMSO.

### **3. Experimental Section**
