Identification of transformation products during Doxylamine chloramination for NDMA mitigation

ABSTRACT N-nitrosodimethylamine (NDMA) is a disinfection byproduct that forms at the presence of an organic nitrogen precursor. Doxylamine, an antihistaminic pharmaceutical, is a precursor of NDMA and has been shown to form NDMA in the presence of chloramine. In this study, the effect of Doxylamine as an NDMA precursor has been further studied during chloramination. The end product and byproducts during chloramination were investigated using a high-resolution mass spectrometer by taking samples at different time intervals. Results suggest that NDMA is not the only end product forming during chloramination of Doxylamine and several transformation products that do not end up as NDMA may form. A group of these transformation products have been selected based on their relative amounts during chloramination with time and notated as Focus Tentative Transformation Products (FTTPn). The identification of these byproducts will make it easier to study the conditions during chloramination that may favour these ‘known’ transformation products with the use of less sophisticated analytical instruments. Then, it might lead to the establishment of chloramination protocols that will minimise the formation of NDMA from its precursors. GRAPHICAL ABSTRACT


Introduction
Nitrosamines are chemical compounds known as one of the most carcinogenic and toxic compounds and their negative effects can be observed even at low (ng/L) concentrations [1,2].As their precursors such as amines and nitrosating agents exist widely in industrial products and in food industry, nitrosamines are likely to form rather easily [3].In addition, one of the nitrosamines, N-nitrosodimethylamine (NDMA) is observed as a byproduct during disinfection processes in water treatment plants [4][5][6] and wastewater treatment plants [7][8][9][10][11].
NDMA is not as widely regulated as other disinfection byproducts (DBPs) even though the concentration of NDMA causing cancer with one chance in a million is 0.7 ng/L for drinking water and 0.42 ng/L for tap water [12].Nevertheless, there are recent additions to the regulations worldwide to control NDMA concentration in drinking water.Drinking water guidelines or standards include limits for NDMA concentration in Japan, Australia, Canada, Alabama, Alaska, California, Delaware, Florida, Indiana, Massachusetts, Mississippi, New Jersey, North Carolina, Pennsylvania, Washington and West Virginia, ( [13][14][15][16]).
In addition, the need for direct or indirect reuse of treated wastewater has also led to research conducted on NDMA and its precursors in wastewater.After NDMA presence has been confirmed in drinking water and wastewater by several researchers since the beginning of 2000s with concentrations up to 100 ng/L [7,[9][10][11][17][18][19][20][21][22], research has also been conducted on its formation during disinfection processes and its precursors.
Among different NDMA precursors, the ones that stand out as compounds that exist widely in wastewater and surface waters, are pharmaceuticals with concentrations ranging from ng/L to μg/L [36].In some cases, the concentration of pharmaceuticals can even reach mg/L levels [35].Pharmaceuticals with amine moities have more potential to act as NDMA precursor [20].Doxylamine (Figure 1) is an antihistamine and nausea preventer [38] whose concentration of in wastewater can reach up to 300 ng/L [39] and it has been shown to act as NDMA precursor upon chloramination [33], oxidation with ClO 2 [40] and during UV and UV/H 2 O 2 treatments [29].Therefore, its presence in wastewater may increase NDMA concentration due to disinfection.Moreover, in case of discharge without disinfection, Doxylamine may be present as an NDMA precursor in the surface water to be used directly or indirectly for different beneficial purposes, including drinking water.
Several pathways for NDMA formation from several NDMA precursors have been studied [41][42][43].These probable pathways could provide information on how to decrease or to prevent the formation of NDMA by changing conditions that have been determined to lead to NDMA formation, but no study has attempted to look at possible reactions of NDMA precusors that will not lead to NDMA formation under certain conditions as a means of controlling or reducing NDMA concentration during disinfection.This other possibility to reduce NDMA formation, namely, defining conditions that do not lead to NDMA will first require to study the reactions that lead to formation of products other than NDMA.It could be possible to decrease or to prevent NDMA formation by focusing on transformation products that form during disinfection but then do not end up as NDMA.To be able to do such an analysis, one has to perform a measurement of 'unknown unknowns'; defined as non-targeted analysis.Although quantification of compounds of interest that have been defined and for which we have reference standards is needed for general water quality monitoring as well as to check compliance with regulations, in non-targeted analysis, there is no pre-defined or known compounds for which the concentration has to be measured.These two different analyses also require different equipment and user expertise.Even though liquid or gas chromatography coupled with tandem triple quadrupole mass spectrometer (LC-MS/MS or GC-MS/MS) is enough to detect and quantify 'known' compounds at relatively low concentrations in difficult matrices (through the use of isotopic dilution), more costly equipment and more expertise are needed for non-target analyses.Therefore, suggestions or identification of transformation products with non-target analysis can act as a preliminary but necessary step for further experiments where one can measure these newly-identified 'known' compounds.
In non-targeted analysis, the mass/charge (m/z) of compounds occurring during reactions are scanned by highly sensitive instruments such as high-resolution mass spectrometry.Compounds of which the formula or the structure is not known can be identified at different confidence levels based on the data collected from experiment.During identification of compounds obtained with non-targeted analysis, 5 different confidence levels are defined [45].Level 1 includes compounds which have a confirmed structure through the use of a reference standard.Level 2 contains compounds which have probable structures which are confirmed by library or diagnostic evidence.Level 3 of confidence levels is a result of suspected substructures of compounds and tentative classification.The Level 4 is identified when the structural evidences for the compounds are insufficient and results in unequivocal molecular formula.Level 5 identification label is used to identify a 'mass of interest' without any proof.
Since Doxylamine is already known to act as an NDMA precursor, the aim of this study was to focus on identification of transformation products of Doxylamine that do not lead to formation of NDMA during chloramination.To be able to understand whether it would be possible to change chloramination conditions in a water/wastewater treatment plant so that less NDMA is formed from Doxylamine, first, transformation products that do not form NDMA upon chloramination have to be explored and identified.Once the transformation products have been identified or at least suggested, further studies can focus on these compounds by the use of less expensive equipment to elaborate on chloramination conditions.Hence, in this study, we employed a non-targeted mass spectroscopic analysis to investigate the transformation products and their tentative structure during chloramination of Doxylamine to identify the compounds that do not lead to formation of NDMA.

Materials and methods
Analytical grade acetonitrile (Merck), High Performance Liquid Chromatograph (HPLC) grade water (Merck) were used during the experimental study.The NDMA and deuterated NDMA (d 6 -NDMA) standards were purchased from Absolute Standards, USA.All chemicals except Doxylamine (SIGMA-ALDRICH Co., USA) were purchased from MERCK KGaA, Germany.Only glassware and Teflon were used to prevent any contamination with plastics.

Chloramination of Doxylamine
Experiments were conducted on synthetic sample which was prepared by addition of Doxylamine at a concentration of 50 µg/L into distilled water.Chloramination of Doxylamine were conducted under the conditions of NDMA Formation Potential (FP) test [6] except the duration time.The principle of the formation potential test is to provide enough time (7 days in our experiments instead of 10 days) for NDMA to form from precursors in the presence of excess chloramine dosage (2 mM) in a pH-controlled environment (pH:8 with phosphate buffer).
The concentration of Doxylamine was selected as 50 µg/L corresponding to an expected NDMA concentration of 5 µg/L in 7 days.This concentration is higher than the concentration of Doxylamine found in water and wastewater samples (0.1-1 µg/L, [38]) since the aim was not to measure NDMA but rather to observe and suggest transformation products that do not end up as NDMA upon chloramination.The lack of information on transformation products of Doxylamine upon chloramination and more specifically the transformation products that do not form NDMA also required this study to be conducted in a rather clean matrix since it is the first study that aims to study the transformation products of Doxylamine during chloramination.
In addition of measuring the NDMA FP at the end of the total duration, additional samples were taken during chloramination under NDMA FP conditions to evaluate the formation of NDMA over time.The experiments were stopped with addition of ascorbic acid (end concentration: 11 mM) after selected time intervals of t = 0, 30 min, 12 h and 168 h.The sample denoted as t = 0 min is taken approximately 1 min after the start of the chloramination process.The test duration is arranged to represent the conditions both in the treatment plant and distribution system and samples are taken at different time intervals in order to be able to identify the change related to the contact time.

Sample preparation and NDMA measurement
Solid phase extraction (SPE) is a tool to enrich the analytes and decrease matrix effect in samples.SPE was applied to all samples before measurement with LC-MS/MS.The SPE method was described in detail by Topuz et al. [46].SPE extraction was done with Bakerbond Carbon (1000 mg; Mallinckrodt-Baker, Philippsburg, NJ).The Bakerbond Carbon cartridges were conditioned using 2 × 5 mL of acetonitrile, 2 × 5 mL of HPLC grade water.The 1000-mL aliquots of samples were spiked with internal standard (100 ng of deuterated NDMA).Samples passed at a flow rate of ∼3 mL/ min from cartridge under vacuum.Cartridges were rinsed with 5 mL of HPLC grade water, and dried under a vacuum for 1 h.The cartridges were eluted with 2 × 2 mL of acetonitrile and 2 × 2 mL of acetone into a conical flask, respectively.Using Turbovap II instrument (Caliper Life Sciences) under 1 bar pressure of Nitrogen gas at 40°C, eluted solvent was evaporated until 0.75 mL extract was left.The extracts were transferred to 2-mL auto sampler vials after the volumes adjusted volumetrically to 1 mL using HPLC grade water.Before transferring of the extracts to vials were filtered using a 0.22-μm syringe filter.
NDMA was measured with Thermo Accela model Ultra High Performance Liquid Chromatograph (UPLC) pump and Accela model automatic sampler connected with TSQ Quantum Access model MS/MS.For liquid chromatographic separation, Thermo Hypersil Gold Column (100× 2.1 mm; 1.9 µm Thermo Scientific, Germany) was used.For the measurement of NDMA, the method developed by Topuz, Aydin, and Pehlivanoglu-Mantas [46] was used.During measurement of samples duplicate of samples were prepared and each of them were measured three times in the analysis.

Non-targeted analysis
The main aim of this study is the identification of compounds that do not lead to formation of NDMA when Doxylamine is chloraminated.For this purpose, transformation products of Doxylamine were identified for samples obtained at specific time intervals of t = 0 min, t = 30 min, t = 12 h and t = 168 h with a high-resolution mass spectrometer (LC-Q Exactive Hibrid Quadropole-Orbitrap MS, Thermo Scientific).During measurements, Thermo Hypersil Gold aQ (100× 2.1 mm, 1.9 µm) column was used.The flowrate of mobile phase was 400 µL/min and volume of injection was 10 µL.The mobile phases were water containing 10 mM ammonium formate and 0.1% formic acid and acetonitrile containing 0.1% formic acid.
Non-targeted analysis was based on evaluation of MS/MS spectra of Doxylamine transformation products during chloramination.Data obtained from MS/MS in non-targeted analysis need to be evaluated through mass filtering conducted by special software applications to identify common fragments or neutral loses [46][47][48][49][50][51].The data of the measurements with MS/MS spectra (t = 0, 30 min, 12, and 168 h) were analyzed by using the Compound Discoverer Programme version 2.1.The transformation product investigations were focused on determining and defining compounds that form during chloramination of Doxylamine but do not end up as NDMA.Therefore, first, compounds formed from Doxylamine are filtered (Figure 2) as compounds having a m/z of 182.09644 and 167.077925.Totally, 1441 compounds are listed and among them transformation products that have an m/z of 58.0651 are excluded.The basis for this exclusion is the literature information suggesting that compounds with an m/z of 58.0651 form NDMA through disinfection hence are NDMA precursors [51].Although it might be possible that there might be other fragmentation patterns, the use of 14 different precursors with different structures in Hanigan's study suggest that the use of m/z of 58.0651 is a reliable tool to observe NDMA formation.In this study, by filtering out and excluding the transformation products that have an m/z of 58.0651,only the transformation products that do not form NDMA are listed.Possible candidates were either proposed by the Compound Viewer or drawn by Marvin Sketch (to match the exact mass and the isotopic pattern) and then evaluated based on their Fragment ion coverage (FISh) score.FISh is eventually based on three different annotation sources available in Compound Discoverer programme: predicted compounds, m/z cloud and ChemSpider.Among these 'tentative' compounds [50], those with high FISh scores and expected reactions during chloramination were selected.

Results and discussions
NDMA formation with time as C t /C f (where C f denotes the final NDMA concentration at the end of NDMA FP test at 168 h) is provided in Figure 3.Samples taken at different time intervals suggested that chloramination reaction is rather fast and NDMA formation seems to follow a logarithmic increase.Approximately 15% of the total NDMA was formed within 30 min and C t /C f was 0.85 ± 0.25 after 12 h.These results suggest that it might be more important to focus on conditions within the water treatment plant where chlorination time is usually around 60 min or less to decrease NDMA formation.However, one needs to keep in mind that the experiment was conducted in a clean matrix and real-life conditions might prolong or postpone the formation of NDMA.
During chloramination of Doxylamine, the decrease in the concentration of Doxylamine (as a decrease in peak intensity) can be observed with time (Figure 4) while the concentration of transformation products starts to increase with different trends for different compounds.For the non-target analysis, compounds have been analyzed for transformation products that form  from Doxylamine and do not end up as NDMA by selecting the appropriate fragmentation ions as m/z for two factors (i.e. containing m/z of 182.09644 or 167.077925 and not containing m/z of 58.0651 [53], respectively).In our non-targeted analysis, the compounds are can be confirmed by library evidence or by MS/MS fragmentation pattern [45], suggesting Level 2. Nevertheless, due to the lack of a match of retention times associated with a specific method [50], they are labelled as Level 3, or tentative compounds to err on the side of caution.
Among the compounds which met both m/z criteria, there are more than 40 compounds that have a FISh coverage higher than 50% (Table 1).FISh score shows the match of the fragmentation structures of precursor ions among the expected fragments in a list.It is an algorithm which uses all of the fragmentation scans in the spectrum tree for a compound and a higher FISh score indicates a better matching of the fragmentations.Hence, FISh score indicates a higher possibility of any unknown compound (in our case, transformation products of Doxylamine that to not yield NDMA during chloramination) to be identified.Among the transformation products with high FISh scores, 12 compounds that have yielded high peak area intensities at different sampling times during chloramination were selected as 'focus tentative transformation products (FTTP)' (Table 2) based on their potential importance for decreasing NDMA concentration under right circumstances.These compounds can be divided in 4 groups depending on how fast they form during chloramination and whether they further take part in the chloramination reaction that leads to a decrease in their concentration (Figures 5-8).Group 1 compounds form rather fast as indicated by their high concentration at t = 0 min.Since the first sample denoted as t = 0 min is taken at approximately 1 min after chloramination started, the presence of a high area at t = 0 min demonstrates transformation products that form upon a relatively fast reaction.This reaction could be the hydrolysis of Doxylamine and the subsequent chloramination of the transformation products as suggested by the structures of the transformation products (Table 2).The concentration of these compounds does not decrease right away as indicated with a high peak area observed for samples obtained at t = 30 min.
Transformation products classified as Group 2 include compounds that form at t = 30 min and then start to disappear upon further chloramination.The structure of     these compounds indicates formation of double or triple bonds within 30 min of chloramination.Group 3 compounds have usually smaller molecular weight compared with Group 1 and Group 2 and their concentration is highest at t = 12 h suggesting a slower reaction for their formation compared with the first two groups.When the compounds in group 3 were investigated it is seen that the increasing contact time with chloramine solution provides formation of more branched molecules.
The last cluster of transformation products includes those that form only after a much longer time, i.e. 168 h.This reaction time is higher than the retention time of the chlorination unit in water treatment plants or even higher than the retention times in some water distribution systems.Nevertheless, these compounds seem to be stable once they are formed.
Although currently there are no studies on the pathways for the chloramination of Doxylamine, several studies focused on the investigation of NDMA formation pathways from model compounds such as ranitidine, dimethylamine, trimethylamine, chlorpheniramine, chlortetracycline hydrochloride, benzylamine, dimethylaniline, choline, poly (diallydimethylammonium chloride), polyquaternium-7, polyacrylamide and cocoamidopropyl betaine [40][41][42][43].As suggested for Group 1 compounds of Doxylamine chloramination, hydrolysis was shown to be an important reaction for other NDMA precursors [54,53].Hydrolysis of monochloramine provides decomposition of ranitidine and increases the formation of NDMA in water media [32].Also, hydrolysis of amines might be an important step during formation of NDMA [56].The presence of non-protonated amines from NDMA precursors are important for NDMA formation and hence lowering the pH may decrease NDMA concentration during chloramination of Ranitidine and Sumatriptan [34].However, it is important to note that the pH also might shift the speciation of mono vs. dichloramines.
In addition to hydrolysis, studies about the NDMA formation pathway revealed mainly five different pathways: nucleophilic substitution, aminyl radical generation, oxidation of radical by O 2 , N-nitroso-compound cation formation and release of NDMA [10,11] but it is important to consider that the oxidation processes used are not necessarily the same among these studies.While nucleophilic substitution between amines and dichloramine starts the reaction chain leading to NDMA formation of NDMA during chloramination of amines [42], aminyl radical generation is proposed for NDMA formation during chloramination of ranitidine [57] suggesting that formation of NDMA is not a result of only one type of reaction.

Conclusions
One possibility to decrease the concentration of NDMA during disinfection is to decrease the concentration of precursors.However, it also might be possible to change the disinfection conditions to lower NDMA concentration through the maximisation of compounds that do not end up as NDMA.
During chloramination, several transformation products form and some of them lead to NDMA whereas some transformation products either disappear through further reactions or turn into compounds other than NDMA.
However, compounds that form from Doxylamine and do not end up as NDMA are so far not identified since non-targeted analysis or the study of 'unknown unknowns' requires special equipment as well as high level of expertise.Once the transformation products are identified, much simpler instrumental methods can be used to monitor their concentrations in the samples.NDMA formation from precursors depends on the conditions of water media.Parameters such as monochloramine dose, temperature, contact time and pH (41) as well as the ratio of mono and dichloramines [42] are factors that are affecting the formation of NDMA not only from Doxylamine but also any other NDMA precursor.Controlling these factors to promote target transformation compounds specified in this study could be useful to decrease NDMA formation and its potential harmful effects on public health.The time required for their formation suggest that Group 1 and 2 compounds could be good candidates for target compounds within the water treatment plant to focus on in the future studies to increase their concentrations.On the other hand, Group 3 and Group 4 compounds can be screened in further experiments to represent conditions encountered in water distribution systems.As this is the first study to use a different and a rather unorthodox approach for decreasing NDMA concentrations with Doxylamine as the only precursor, the results are limited.However, based on this manuscript, similar and more thorough studies on decreasing the concentration of NDMA or other disinfection byproducts could be studied in the future.Furthermore, investigation of detailed pathways could also be useful in determining the conditions to eliminate or decrease NDMA formation.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Figure 4 .
Figure 4. Change of concentration of Doxylamine during chloramination.

Table 1 .
Tentative transformation products formed during chloramination of Doxylamine based on FISh score >50%.

Table 2 .
Focus tentative transformation products to focus on to decrease the formation of NDMA from Doxylamine.