Degradation of a Nonylphenol Single Isomer by Sphingomonas sp. Strain TTNP3 Leads to a Hydroxylation-Induced Migration Product

Abstract

Sphingomonas sp. strain TTNP3 degrades 4(3′,5′-dimethyl-3′-heptyl)-phenol and unidentified metabolites that were described previously. The chromatographic analyses of the synthesized reference compound and the metabolites led to their identification as 2(3′,5′-dimethyl-3′-heptyl)-1,4-benzenediol. This finding indicates that the nonylphenol metabolism of this bacterium involves unconventional degradation pathways where an NIH shift mechanism occurs.

Nonylphenol (NP) is a ubiquitous pollutant which mainly results from the biodegradation of widely used NP polyethoxylate surfactants, and it is well known for its endocrine system-disrupting potential (5, 8, 11, 13, 20, 21). NP is used as part of a technical mixture which consists of more than 20 branched isomers of the nonyl chain (23, 28). Although some studies report an easy degradation of the NP with a linear alkyl chain (4-n-NP) through oxidation of the side chain, the situation is different for branched isomers of NP (2, 25, 26). The main reason for the poor degradability of NP is a quaternary carbon atom (α-carbon) of the branched alkyl chain, which is involved in the attachment of the nonyl group to the phenol moiety. More than 85% of the isomers of the technical mixture contain such quaternary α-carbon atoms (26, 28). Some strains belonging to the genera Sphingomonas, Sphingobium, and Pseudomonas have been reported to be able to degrade the branched isomers of NP, but the catabolic pathways in these bacteria remain poorly characterized (6, 7, 18, 19, 21, 24).

Previously, it was proven that the dead-end product of the degradation of 4(3′,5′-dimethyl-3′-heptyl)-phenol (p353NP) diastereomers is 3,5-dimethyl-3-heptanol (nonanol). This metabolite corresponds to the nonyl chain of p353NP, where the aromatic ring has been replaced by a hydroxyl group (3). It remains to be elucidated how the very stable quaternary C atom structure of NP can lead to the formation of nonanol. Nonanol has been found only in cultures of Sphingomonas sp. strain TTNP3 and Sphingomonas cloacae, so these bacteria seem to possess particular degradation pathways for NP (3, 6, 21). Recently, it was reported that neither 4-n-NP nor short-chain alkylphenols were degraded by Sphingomonas sp. strain TTNP3, whereas the diastereomers of the branched single isomer p353NP (Fig. 1) led to the formation of ring-oxidized metabolites (4). Ring-oxidized metabolites were detected only in the intracellular fraction and were shown to be different from 4(3′,5′-dimethyl-3′-heptyl)-catechol (Fig. 1b) and 4(3′,5′-dimethyl-3′-heptyl)-resorcinol (Fig. 1c), hydroxylated at positions C-2 and C-3 (Fig. 1), respectively. We hypothesize the formation of metabolites which undergo either hydroxylation with an alkyl chain internal rearrangement, as is the case for bisphenol A (14), or hydroxylation-induced migration of the alkyl chain immediately after hydroxylation at position C-4. The latter process would lead to the formation of 2(3′,5′-dimethyl-3′-heptyl)-1,4-benzenediol (o353NHQ) (Fig. 1d). In order to identify definitively these metabolites of p353NP, o353NHQ has been synthesized and compared to them.

FIG. 1.

Chemical structure of 4(3′,5′-dimethyl-3′-heptyl)-phenol (p353NP) (a), 4(3′,5′-dimethyl-3′-heptyl)-catechol (b), 4(3′,5′-dimethyl-3′-heptyl)-resorcinol (c), and 2(3′,5′-dimethyl-3′-heptyl)-1,4-benzenediol (o353NHQ) (d). Also shown is the schematic presentation of the reaction pathway proposed for the formation of NIH-shifted metabolites of p353NP (a through d).

Sphingomonas sp. strain TTNP3 precultures and cultures were grown aerobically in complex medium and mineral medium supplemented with p353NP, respectively, as described previously (3). p353NP was synthesized by a Friedel-Crafts alkylation of phenol with 3,5-dimethyl-3-heptanol in the presence of BF₃ as a catalyst (3). Crude cell extracts were prepared according to the following procedure. Strain TTNP3 suspensions (1.4 liters) cultivated on p353NP were harvested after 6 days of incubation. Cells were washed with phosphate buffer (50 mM, pH 7.0). The pellets were resuspended in 25 ml of the same buffer (approximately 17 g [dry weight] of biomass/liter) and sonicated. The lysate was extracted with ethyl acetate after acidification with HCl to a pH between 2 and 3. The extracts were dried over anhydrous Na₂SO₄ and adjusted to a volume of 4 ml with an approximate metabolite concentration of 470 μg/ml. The method has been described in detail previously (4). Fractions containing the metabolites [pseudomolecule ion at m/z 235 (M−H)] were separated from the intracellular extracts of strain TTNP3 by means of semipreparative high-performance liquid chromatography (HPLC) with online UV diode array detection (HP 1100; Agilent Technologies). UV detection was performed at 210 nm with spectrum acquisition from 190 to 400 nm. The mobile phase was split after the column chromatography for parallel performance of electrospray ionization-mass spectrometry (MS) (SSQ 7000; Finnigan MAT) for identification of the metabolites. A voltage of −4.2 V was applied to the electrospray needle, the capillary temperature was set to 200°C, the sheath gas (N₂) pressure was adjusted to 60 lb/in², and the auxiliary gas (N₂) pressure was adjusted to 20 lb/in². Detection was performed in the negative mode by applying in-source collision-induced dissociation of 10 or 70 eV. Separation was achieved with a Eurospher-100 C₁₈ column (8 by 250 mm; particle size, 5 μm) (Knauer, Berlin, Germany) at 35°C and a flow rate of 3 ml/min. After 22 min of isocratic elution with 25% water-75% methanol, the organic proportion was increased linearly to 100% within 3 min and held at 100% for 5 min. One hundred microliters of the intracellular extract was injected repeatedly, and the ion that caused the peak at m/z 235 (retention time [RT], 9.5 min) was collected with a fraction collector. Methanol was evaporated under nitrogen in the combined HPLC cuts of the peak (8 to 10.5 min). The remaining aqueous phase was acidified with HCl (pH 2 to 3) and extracted two times with 2 volumes of ethyl acetate. The combined organic phases were dried over Na₂SO₄ and adjusted to a volume of 1 ml.

The o353NHQ was synthesized by Friedel-Crafts alkylation from hydroquinone with nonanol. For 60 min, 880 mg of hydroquinone (99%; Fluka), 1.4 ml of 3,5-dimethyl-3-heptanol (nonanol, 99%; Avocado), 5.0 ml of diethyl ether (Acros), and 10 ml of BF₃-ether complex (for synthesis; Merck) were incubated together at 60°C, as described previously (27). The product yield was 134.5 mg of o353NHQ with a purity of 49% (gas chromatography [GC]-MS analyses). GC-MS was used for identification and characterization of the reaction product (27). In addition, 3 mg of o353NHQ was purified by semipreparative HPLC (see above), and a ¹H solution-state nuclear magnetic resonance (¹H-NMR) with CDCl₃ was recorded as reported previously (27). The following ¹H-NMR chemical shifts are given in parts per million relative to the internal standard tetramethylsilane at 400 MHz, and the coupling constants of J are given in hertz (for position of atoms, see Fig. 1): 0.50 to 1.60 (several broad m, 19 H [C₉H₁₉; compare to reference 27]), 4.30, 4.40 (each s, 2 H—OH), 6.50 (dd, J₁ = 8.4, J₂ = 1.7, H—C₅), 6.56 (d, J = 8.4, H—C₆), 6.68 (d, J = 1.7, H—C₃). The chemical structure of the synthesized o353NHQ was verified by its MS and ¹H-NMR spectra.

Prior to GC analyses, 50 μl of each sample was dried under a nitrogen stream and derivatized with 50 μl of N-methyl-N-trimethylsilyltrifluoroacetamide (Fluka) at 60°C for 15 min. The trimethylsilylated metabolites and reference compound o353NHQ were analyzed by GC-MS as described previously (27). Both the extract and the authentic compound o353NHQ gave rise to separated diastereomeric peaks with RTs of 22.2 and 22.3 min. It is important to note that all compounds synthesized with 3,5-dimethyl-3-heptanol, such as p353NP and o353NHQ, are found as four diastereomers due to the presence of two asymmetric carbon atoms in the nonyl chain. They give rise to only two characteristic diastereomeric peaks in all GC analyses, and both peaks display the same mass spectrum (23, 27). To allow better comprehension of the GC-MS analyses, only fragment ions of the first diastereomeric peak of the metabolites and o353NHQ are presented in Table 1. Both trimethylsilylated derivatives of the metabolites and the reference compound o353NHQ showed the molecule ion at m/z 380. Other characteristic fragment masses were observed at m/z 309, 281, 267, 254, and 73. The main ion was found at m/z 73 and corresponded to a trimethylsilyl fragment. The occurrence of the (M-113)⁺ ion (at m/z 267) was explained by the formation of a ditrimethylsiloxy tropylium ion. Such C₇ ring results from the benzylic cleavage, which is one of the preferred ways to fragment NP (23). The ion (M-71)⁺ can be attributed to the loss of the pentyl group. For the (M-99)⁺ ion, fragmentation implies the loss of an ethyl and a pentyl fragment with a shift of a hydrogen atom to the positive ion. Furthermore, the GC analysis of a mixture of the metabolites and of the synthesized reference compound showed similar chromatography results (data not shown).

TABLE 1.

Chromatographic and mass spectral characteristics of the GC-MS analysis^a

m/z	(M-Xb)+	Relative abundance (%) of ions of:
		p353NPc	o353NHQc
380	0	27	26
309	71	38	40
281	99	74	73
267	113	23	24
254	126	8	8
73	307	100	100

^aAnalysis of the trimethylsilated derivatives of the metabolites of p353NP and the reference compound o353NHQ.

^bX, variable for ion designation.

^cDiastereomer retention times, 22.2 and 22.3 min.

In order to demonstrate further the identity of the metabolites and the o353NHQ, GC-tandem MS (MS/MS)-selective reaction monitoring (SRM) analyses were performed by means of a Finnigan MAT GCQ gas chromatograph described previously (17). Two-microliter split injections (1:10) were carried out. The GC separation was performed on a Valcobond VB-5 fused-silica capillary column (0.25-μm film thickness, 60 m by 0.25 mm; VICI). Temperature and gas conditions used for analyses were as described previously (17). The positive-electron-impact full-scan data were acquired under the following conditions: ion source temperature, 200°C; mass range, m/z 30 to 450; scan time, 1.76 s; mass resolution (M/ΔM), 500; solvent delay, 8 min; electron energy, 70 eV; emission current, 250 μA; and electron multiplier voltage, 1,750 V. In the MS/MS-SRM mode, the formation of product ions generated from selected parent ions was monitored. The parameters for the MS/MS-SRM studies were as follows: helium as the collision gas; isolation time, 14 ms; excitation voltage, 1.00 V; excitation time, 15 ms; excitation energy, 0.450; and product ions scan range, m/z 190 to 400. MS² analysis (parent transition 380⁺→309⁺) results show that the trapped molecule ion 380 leads to ions at m/z 351 (loss of ethyl), 309 (loss of pentyl), 281 (loss of ethyl and pentyl), 267 (disubstituted tropylium), 254, and 251 for both the metabolites and the o353NHQ (Fig. 2A and C). MS³ analyses (parent transition 309⁺→267⁺) showed the presence of a prominent ion at m/z 267 and a minor ion at m/z 251 (Fig. 2B and D). Further MS⁴ experiments (parent transition 267⁺→251⁺) showed that the ion at m/z 251 results from the disubstituted tropylium derivative (data not shown). Tropylium ions are assumed to be stable ions, and the loss of 16 atomic mass units can be explained only by a loss of a CH₄ group from one of the trimethylsiloxy groups. Another MS² analysis has been performed for the parent transition 380⁺→351⁺ in order to ensure that the nonyl chain structures were identical for both the metabolites and the o353NHQ. The ions at m/z 281 (100% relative abundance and further loss of pentyl) and at m/z 251 (2 to 6% relative abundance) were observed for the reference compound, as well as for the metabolites (data not shown). These results showed that the ion at m/z 351 contains the intact pentyl fragment and thus results from the loss of ethyl (C-1′ and C-2′ of the heptyl chain), and they also confirm the alkyl chain structure identity for the metabolites and the authentic compounds.

FIG. 2.

GC-MS/MS-SRM spectra of trimethylsilylated derivates of the reference compound 2(3′,5′-dimethyl-3′-heptyl)-1,4-benzenediol (o353NHQ) (A and B) and of metabolites of 4(3′,5′-dimethyl-3′-heptyl)-phenol (p353NP) (C and D). (A and C) transition 380⁺→309⁺, MS²; (B and D) transition 309⁺→267⁺, MS³.

GC-MS and GC-MS/MS-SRM analyses led to the conclusion that metabolites and o353NHQ give rise to the same fragmentation pattern. GC-MS/MS-SRM analyses of different parent transitions have shown that the observed fragment ions of the metabolites and of the reference compound have the same origin. This fact is of the utmost importance, because it indicates that the alkyl chain has not undergone an internal rearrangement. Additional information on the aromatic structure of the metabolites and of o353NHQ was obtained by acquisition of their respective UV spectra during HPLC-diode array detection analyses. The UV spectra completely matched each other, and both showed a maximum of absorption at 294 nm (data not shown). On this basis, we identified the isolated metabolites of p353NP definitively as o353NHQ. This implies that the parent compound has been hydroxylated at a quite unexpected position (C-4) and that the side chain has migrated to the adjacent carbon atom of the ring. The most probable mechanism is the so-called NIH shift, which would correspond to a hydroxylation-induced migration of the nonyl chain (Fig. 1). Such mechanisms were reported in higher organisms and bacteria for halogenated as well as alkyl substituents (9, 10, 12, 15, 22). Incubations of Burkholderia cepacia with 3,5-dimethyl-4-hydroxybenzaldehyde, syringaldehyde, or 5-bromovanillin led to the formation of para hydroxylation of the phenol with a simultaneous shift of the aldehyde group to the vicinal position (15). Interestingly, these authors also did not observe the formation of catechols and the production of hydroquinone derivatives by chlorophenol 4-monooxygenase. Hydroquinone derivatives were also reported in Sphingomonas chlorophenolica, a Mycobacterium sp., and a Rhodococcus sp. grown on pentachlorophenol, but hydroquinones resulted from the loss of chloride in the para position of the phenol (1, 16). An NIH shift of a Cl atom would not have been apparent in these cases, due to the hexa-substitution of the pentachlorophenol ring. Further studies are currently being performed in order to elucidate whether the o353NHQ is further metabolized and if it does lead to the production of nonanol.