Characterisation of selected drugs with nitrogen-containing saturated ring structures by use of electrospray ionisation with ion-trap mass spectrometry
The electrospray ionisation–ion-trap mass spectrometry (ESI–MSn) of selected drugs with nitrogen-containing saturated ring structures has been investigated. Sequential product-ion fragmentation experiments (MSn) have been performed to elucidate degradation pathways for the [M+H]+ ions and their predominant fragment ions. These MSn experiments result in characteristic fragmentations in which functional groups are generally cleaved from the ring systems as neutral molecules such as H2O, amines, alkenes, esters, carboxylic acids, etc. When such a nitrogen-containing drug molecule also contains a functional group, such as an ester, that on liberation as a neutral molecule has a significantly lower −ΔHf° value than that of the corresponding amine then the former is preferentially liberated. Furthermore, when an aromatic entity is present in these drug molecules together with the nitrogen-containing saturated ring structure fragmentation of the latter ring occurs with the former, predictably, being resistant to fragmentation. The structures of fragment ions proposed for ESI–MSn can be supported by electrospray ionisation–quadrupole time-of-flight mass spectrometry (ESI–QTOFMS). The data presented in this paper therefore provide useful information on the structure of these heterocyclic compounds which could be used to characterise unknown drug compounds isolated from natural sources, for example.
KeywordsElectrospray ionisation Ion trap mass spectrometry Time-of-flight mass spectrometry Drugs
Joyce et al.  have investigated the characterisation of selected drugs with amine-containing side-chains using electrospray ionisation and ion trap mass spectrometry (ESI–MSn). From this study certain rules can be formulated with respect to the ESI–MSn behaviour of drugs with amine-containing side-chains. For example, drugs such as trimpramine and methadone with a carbon chain ending in a tertiary nitrogen atom with at least two methylene or substituted methylene groups separating this nitrogen atom from the other end of the carbon chain will lose the end nitrogen atom as the corresponding secondary amine followed by loss of the corresponding alkene formed from these two methylene or substituted methylene groups. If there is only one methylene group adjoining the end nitrogen atom as is the case with lignocaine then the end of this carbon chain becomes the detectable charged species as in CH2=N+(C2H5)2 and no neutral amine and alkene is formed. Drug molecules such as clenbuterol and salbutamol which have a secondary amine-containing carbon chain –CH(OH)–CH2–NH–C(CH3)3 lose H2O followed by the end substituted carbon group as alkenes (CH3)2C=CH2.
These MSn experiments, supported by QTOFMS–MS data, therefore show certain characteristic fragmentations with respect to the amine-containing side-chains. The data therefore provides useful information on the structure of these compounds with amine-containing side-chains and can be used in the characterisation of such drugs and their structurally related metabolites. The ESI–MSn data of such compounds can be held in a database and neutral mass losses/low molecular mass ions cross-referenced with such data obtained from analytes of unknown structure which can then be of value in their structural characterisation with respect to those molecules with amine-containing side-chains.
This paper follows on from this study by presenting a similar study of the ESI–MSn behaviour of some twelve selected drugs with nitrogen-containing saturated ring structures again with a view to establishing rules of fragmentation that can be used for characterisation purposes and to ultimately aid in the identification of unknown analytes isolated from natural sources, for example. The proposed MSn fragmentations are supported by QTOFMS–MS data.
Fragmentation of these drugs with nitrogen-containing saturated ring structures using ESI–MSn, supported by ESI–QTOFMS–MS data commonly involves loss of certain functional groups such as H2O, amines, alkenes, esters, carboxylic acids etc. The acquisition of such information and the resulting rules of fragmentation can therefore be of value in structural characterisation of these molecules and can therefore help in the characterisation of unknown drugs, isolated from natural products, for example.
Instrumentation and Chemicals
MSn characterisation of the drugs and natural products was achieved using the Classic LCQ quadrupole ion-trap mass spectrometer (Finnigan MAT, San Jose, California, USA) utilising electrospray ionisation (ESI). 1.0×10−5 mol L−1 solutions of the drugs in methanol, also containing 1% acetic acid, were infused into the ESI probe at a rate of 10 μL min−1. The collision energy was kept at the instrument default value of 25% (arbitrary unit set by the software). The sheath gas flow was set to 50 (arbitrary unit defined by the software) and the auxiliary gas to 5. Nitrogen gas for the LCQ was delivered from a Whatman nitrogen generator (Whatman, Haverhill, MA, USA), while the helium damping gas present in the ion-trap was obtained from BOC Medical Gases (Guildford, Surrey, UK)
The capillary temperature was set to 250 °C and the spray voltage to 3.5 kV for the MSn studies. For MSn characterisation of the compounds the most intense peak in the mass spectrum, i.e. [M+H]+, was chosen for MS2 analysis, providing first generation product ions in this mode. The process was repeated for MS3 and MS4. Generally, the most intense peak from the previous analysis was chosen for further characterisation at each MSn stage. Other signals of at least 10% abundance that are reproducible are reported in this paper. An isolation width of 1 u was used for the various MSn stages.
QTOFMS was carried out with a Micromass Q-Tof Ultima API mass spectrometer. Solutions (10−6 mol L−1) of the drugs in methanol, also containing 1% formic acid, were directly infused into the ESI probe at 2 μL min−1. The spray voltage was set to 2 kV with the source temperature at 80 °C and the desolvation temperature at 150 °C. Nitrogen was delivered from a Peak Scientific Nitrogen Generator set at 100 psig and resulted in a nebulising gas flow of 50 L h−1 and a desolvating gas flow of 300 L h−1. Argon was used as the collision gas with 10 eV energy used in the MS mode and 20–40 eV energy used in the MS–MS mode.
The spectrometer was calibrated with sodium formate and adenosine, with lock mass of 268.1040, being used as an internal standard for each sample. The resolution of the spectrometer was ca 10,000 for the [M+H]+ ions of the drug compounds.
Methanol, acetic acid and formic acid were obtained from BDH (Poole, Dorset, UK). All drug compounds were obtained from Sigma (Gillingham, England) and The Forensic Science Association of Northern Ireland (Carrickfergus, NI). Standard solutions of the drugs were prepared by dissolving an appropriate mass in 25 mL of methanol to provide a concentration of 1.0×10−3 mol L−1.
Results and discussion
It should be noted that the LCQ does not give high accuracy mass measurements necessary for unequivocal identification of the products of the fragmentation processes so the QTOFMS was used to support the fragmentation structures proposed using the LCQ. It should also be noted that generally in each case only the major signal (100% abundance) observed using the LCQ has been selected for further characterisation at each MSn stage and it is principally these results that are used in the formulation of rules of fragmentation. There are additional peaks at lower intensities in a particular MSn scan that can give rise to a fingerprint of fragmentation, assisting further in the structural characterisation of the molecule in question. The signals are particularly examined in the MS and MS2 modes when, in the latter case, the [M+H]+ ion is fragmented. Signals in later MSn modes are generally of lesser value since the ion that is being fragmented may be substantially different than [M+H]+ and have involved ring closure, ring contraction etc., although losses of marker species such as CO in later MSn modes can be of diagnostic value.
In these studies it is incorrect to consider only the stability of the resulting ionic product formed by fragmentation; the heat of formation of the ejected species/neutral molecule must also be taken into consideration. The elimination of small stable molecules such as H2O (ΔHf°=−241.8 kJ mol−1 for gaseous H2O) and CO (ΔHf°=−110.5 kJ mol−1 for gaseous CO), are commonly observed in ESI–MS and MSn ion trap studies. Furthermore, the bond energies of the bond(s) broken and formed during fragmentation must also be considered. Finally, it appears that even-electron ions can generate odd-electron ions/radicals, e.g. CH3, even though such processes are unlikely under the low energy conditions of ion-trap collisionally induced dissociation (CID) with helium atoms. If the proposed formation of radicals takes place at higher values of n in MSn experiments then it is possible to attribute some of these signals to accidental resonances at higher harmonics.
Pethidine gives its [M+H]+ ion at m/z 248.5 with no in-source fragmentation. On application of MS2 a major signal is given at m/z 220.2 and lesser signals at m/z 202.2 and 174.3. The proposed LCQ fragmentation pattern involving the ethyl ester substituent losing C2H4, C2H5OH and the ester itself is supported by QTOFMS–MS data. Song et al.  have also observed this product ion at m/z 220 when using liquid secondary ion and tandem mass spectrometry for quantitative analysis of pethidine in urine. MS3 of m/z 220.2 gives a single signal at 174.1 which at MS4 results in a signal at m/z 70.0. Similar signals are given for EI of ethyl esters in that M−28, M−46 and M−73 signals are observed as with the determination of pethidine in body fluids by gas chromatography–EI–MS . The 73u that is lost in EI is given as COOC2H5. It should be noted that in this case the amine, CH3NH2, is not liberated in the fragmentation process but rather it is the ethyl ester substituent. This should be contrasted with nicotines, discussed later in this section, which can lose such amines from such nitrogen-containing rings in the absence of an ethyl ester substituent, for example. When such a nitrogen-containing drug molecule also contains a functional group such as an ester that on liberation as a neutral molecule has a significantly lower −ΔHf° value than that of the corresponding amine then the former is preferentially liberated.
These observations show that the unsaturated heterocyclic entities, (i.e. those of benzisoxazole and pyrimidine) are resistant to fragmentation and it is the relatively weak piperidine C–N bond (bond energy 293 kJ mol−1)  that is fragmented using MSn rather than double bonds in the heterocycles such as C=N (bond energy 615 kJ mol−1), C=C (bond energy 614 kJ mol−1) and C=O (bond energy 799 kJ mol−1) .
The [M+H]+ ion at m/z 304.1 undergoes in-source fragmentation to give a signal at m/z 182.1 (relative abundance 40%), a signal that is also observed at MS2. This corresponds to loss of the benzoic acid substituent as a neutral molecule, leaving a secondary carbonium ion. This is supported by QTOFMS–MS data with elemental analysis data of C10H16NO2 for a signal at 182.1195. MS3 and MS4 give major signals at m/z 150.0 and 122.2, presumably corresponding to successive loss of CH3OH and CO from the methyl ester substituent. These substituent losses are also observed in EI, together with a variety of low mass signals and a base peak at m/z 82.
The behaviour of cocaine is somewhat similar to pethidine in that the methyl substituted piperidine is not involved in fragmentation with successive losses of the benzoic acid and methyl ester substituents occurring instead. This is understandable in that the −ΔHf° values  for liberated neutral molecules such as CH3OH(g) and CO(g) are respectively −201.0 and −110.4 kJ mol−1 compared to amines such as CH3NH2(g) which has a value of −28.0 kJ mol−1.
Fuh et al.  have recently determined the free form of cocaine in rat brain by HPLC–ESI–MS with in vivo microdialysis. The [M+H]+ ion was observed at m/z 304 with a fragment ion at m/z 182, corresponding to loss of the benzoic acid entity as stated above. In addition, the benzoylecgonine metabolite was also investigated and it gave rise to a similar fragmentation pattern with a product ion being observed at m/z 168.
The protonated molecular ion at m/z 414.2 gives a signal at m/z 220.1 on application of MS2, suggesting that narcotine could fragment at the C–C bond that joins the two cyclic isoquinoline and benzofuranone structures together, leaving an isoquinolinium ion for the next MSn stage. Aggarwal et al.  have also suggested C–C bond cleavage in their recent ESI–MS–MS study of narcotine-type molecules. This postulation is supported by QTOFMS–MS data with an elemental formula of C12H14NO3 at m/z 220.0955. MS3 gives a signal at m/z 205.1 presumably corresponding to the loss of a methyl group from the methoxy group. Even though this is unlikely in this low-energy technique, such 15-u losses are observed for other methoxy-substituted aromatics such as coumarins [9, 10]. MS4 and MS5 give rise to major signals at m/z 188.1 and 160.1, not observed using QTOFMS–MS.
Reserpine contains a piperidine ring with the N atom at a ring junction in an overall five fused ring system containing certain substituent groups. The protonated molecular ion at m/z 609.4 gives a major signal at m/z 397.3 on application of MS2 with lesser signals at 436.1, 448.1 and 577.2 using the MS2 mode. The former signal corresponds to loss of the substituted benzoic acid entity as supported by QTOFMS–MS data with an elemental formula of C23H29N2O4 and the latter signal would appear to be due to loss of CH3OH from the methyl ester substituent as was observed for cocaine with the fused ring system remaining resistant to fragmentation. However, the 436.1 signal would appear to be due to fragmentation occurring in the piperidine ring system (elemental formula C22H30NO8 using QTOFMS–MS data) with the substituted indole being resistant to fragmentation and being lost as a neutral molecule. Finally, MS3 of the 397.3 signal gives a major signal at m/z 365.2 corresponding to loss of CH3OH as observed above and MS3 of the 577.2 signal, which is the pseudomolecular ion having lost CH3OH, gives a signal at m/z 365.1 which corresponds to loss of the substituted benzoic acid entity. These MSn experiments illustrate the technique’s utility in helping to characterise such a complex low molecular mass molecule as reserpine in that functional groups are cleaved from the ring system as neutral molecules such as CH3OH and a substituted benzoic acid moiety, quite apart from fragmentation in the piperidine ring system. EI mass spectrometry gives similar signals but many additional signals in the range m/z 45–265 as would be expected.
The [M+H]+ ion at m/z 475.3 on application of MS2 gives a base peak at m/z 311.0 with another major signal at m/z 377.1 (85% relative abundance) which correspond to loss of the SO2–piperazine and the piperazine entities respectively. This is supported by QTOFMS–MS elemental analyses of C17H19N4O2 and C17H21N4O4S for signals at m/z 311.1554 and 377.1315 respectively. As with prazosin, the aromatic heterocyclic ring system is resistant to fragmentation processes. MS3 of the m/z 311.0 signal gives one at m/z 283.1 which appears to be due to loss of C2H4 from the ethyl ether group as also occurs for EI of such compounds with M−28 losses. To support this postulation, QTOFMS–MS gives a base peak at m/z 283.1228 with elemental analysis of C15H15N4O2. Eerkes et al.  have also observed this m/z 283 ion after fission of the C–S bond and loss of C2H4 for both sildenafil and its demethyl metabolite using ESI and tandem mass spectrometry.
Olanzapine, like prazosin and sildenafil, contains a piperazine ring and an unsaturated heterocyclic ring system. Again it is the saturated heterocyclic entity that is subjected to fragmentation. The protonated molecular ion at m/z 313.2 gives a signal at 256.2 on application of MS2 which in turn gives an m/z signal at 213.0 using MS3 which gives a signal at 198.2 at MS4. This fragmentation process which primarily involves the piperazine side-chain is supported by QTOFMS–MS data. This latter data also gives an m/z signal at 282.1059 corresponding to loss of CH3NH2 which is observed with the LCQ only at 5% relative abundance. Bogusz et al.  have observed in source CID with m/z ions at 282, 256 and 213 for olanzapine using APCI–MS with a single quadrupole mass spectrometer and an octapole offset value of 20 V.
This case of olanzapine and those of the other drugs in this paper illustrate that, in contrast to EI–MS, many factors influence the mass spectra of organic compounds when analysed by techniques such as ESI–MS and APCI–MS. In-source fragmentation or CID using such ionisation techniques have demonstrated the need for the standardisation of experimental conditions when building a library of reference spectra. Furthermore, magnetic sector, quadrupole, ion trap or time-of-flight filters placed on single or multiple mass spectrometry instruments give rise to unique mass intensities patterns from the same molecule that are most often not readily comparable with each other. This has therefore made it very difficult or even impossible to rely on any collections of reference data whatever comparison algorithm is applied for library searching or manual comparison when full confirmation of identity is required for a court’s scrutiny, for example. However, it is still possible in these cases to obtain sufficient mass spectral information about the unknown molecule and the reference substance providing a particular instrument/technique such as the LCQ is employed with standardised experimental conditions. Neutral mass losses calculated in the MSn modes, for example, can help characterise the structure of an unknown molecule such as a natural product pharmaceutical for which there is no reference substance available.
Morphine, (X) and codeine, (XI)
LCQ data on the protonated molecule of morphine at m/z 286.4 shows a base peak at m/z 201.2 and signals of relative abundance greater than 50% at m/z 211.1, 229.1 and 268.1 using the MS2 mode. The last signal is due to the loss of the aliphatic OH group as H2O and the two signals at m/z 211.1 and 229.1 appear to be due to loss of the N-containing ring leaving the structure shown in structure (Xa), with and without the loss of H2O. This is supported by QTOFMS–MS. MS3 of m/z 201.2 gives a signal at m/z 183.1 corresponding to the further loss of H2O from (Xa) and MS3 of the 268.1 signal, the pseudomolecular ion less H2O, gives a signal at m/z 211.1, again corresponding to loss of the N-containing ring. The related opiate 6-monoacetylmorphine also gives fragment ions at m/z 268 and 211 as would be expected from the above postulation . These MSn experiments again illustrate the technique’s utility in helping to characterise such a complex low molecular mass molecule as morphine in that functional groups are cleaved as neutral molecules such as H2O and the N-containing ring as the liberated molecule CH3CH=NCH3 (57 u).
Codeine has a very similar structure to morphine with the phenolic group being replaced by an aromatic methoxy group. As a result its LCQ and QTOFMS–MS behaviour is very similar.
The [M+H]+ ion at m/z 163.5 gives a main MS2 signal at m/z 132.1 corresponding to loss of CH3NH2 and also a signal at m/z 106.1. QTOFMS–MS assigns elemental formulae of C9H10N and C7H8N to these signals supporting loss of the amine and C2H2.
These MSn experiments, supported by QTOFMS–MS data, show certain characteristic fragmentations in that functional groups are generally cleaved from the ring systems as neutral molecules such as H2O, amines, alkenes, esters, carboxylic acids, etc. When such a nitrogen-containing drug molecule also contains a functional group such as an ester that on liberation as a neutral molecule has a significantly lower −ΔHf° value than that of the corresponding amine then the former is preferentially liberated. When an aromatic entity is present in a drug molecule together with a nitrogen-containing saturated ring structure fragmentation occurs to the latter ring with the former being predictably resistant to fragmentation.
The data therefore provide useful information on the structure of these compounds and can be used in the characterisation of such drugs and their structurally related metabolites. The ESI–MSn data of such compounds can be held in a database and neutral mass losses/low molecular mass ions cross-referenced with such data obtained from unknown analytes which could then be of value in their structural characterisation with respect to those molecules with nitrogen-containing saturated ring structures.
Observed neutral mass losses using MSn, and their structural inferences
Loss of CH3 from methoxy-substituted aromatics
Coumarins [9, 10] and quinolines 
Loss of CH3 substituent from aromatic ring
Loss of NH3 from end of chain
NH2 group with at least two adjacent CH2 groups
Loss of NH3 from pyrrolidine ring
Nornicotine (unpublished work)
Loss of OH from N-oxide
Loss of H2O from aliphatic OH
Clenbuterol, salbutamol 
Chloramphenicol (unpublished work)
Loss of HF
Loss of CH2=CH2 after loss of adjacent end-of-chain amine
Loss of CH2=CH2 from ethyl ester substituent
Loss of CH2=CH2 from aromatic ethyl ether
Loss of CO/NCH2 with ring contraction
Benzodiazepines , zopiclone and its N-desmethyl metabolite 
Coumarins [9, 10], quinolines 
Loss of COH/CH2=NH with ring contraction
Benzodiazepines , e.g. Flunitrazepam 
Loss of CH3NH2 from end of chain
N′-Methyl 5-HT 
Loss of CH3NH2 from CH3-substituted pyrrolidine ring
Nicotine (unpublished work)
Loss of OCH3
Coumarins [9, 10]
Loss of S from cyclic structure
Loss of CH3OH from methyl ester substituent
Loss of aromatic Cl/HCl
Clenbuterol , benzodiazepines 
Loss of CH3CH=CH2 after loss of end-of-chain amine
Loss of C3H6
Coumarins [9, 10]
Loss of CH3CH=CH2 from –NH–CH(CH3)2 end-of-chain group
Loss of ketene, CH2=C=O
Coumarins [9, 10], 7-acetamidonitrazepam 
Loss of CH2=N–CH3
LSD and derivatives (unpublished work)
Loss of CO2 from ring
Lactones, e.g. coumarins [9, 10]
Loss of CO2 from COOH substituent
Penicillins (unpublished work)
Loss of CO2 in molecule
Loss of (CH3)2NH from end of chain
Loss of NO2
7-NO2-1,4-benzodiazepines , metronidazole (unpublished work)
Loss of C2H5OH from ethyl ester substituent
Loss of (CH3)2C=CH2 from chain end
Clenbuterol, salbutamol , quinolines 
Loss of CH3CH=NCH3
Loss of CH3CONH
Loss of N(CH3)3 from end of chain
Loss of CH3CONH2
Loss of CH3COOH as substituent
Coumarins [9, 10], quinolines 
Loss of HCOOCH3 from –COOCH3 substituent
Loss of SO2
Loss of C5H8 as side chain
Coumarins [9, 10] and quinolines 
Loss of NH(C2H5)2 from end of chain
Flurazepam , procaine 
Loss of HCOOC2H5 from –COOC2H5 substituent
Loss of Br as aromatic substituent
Br-Substituted quinolines 
Loss of SO3
Loss of OHC–N(C2H5)2
LSD (unpublished work)
Loss of C6H5COOH
Loss of C9H16 from side chain
Loss of I
I-Substituted coumarins [9, 10]
The authors would like to thank The Forensic Science Agency of Northern Ireland, Carrickfergus, Northern Ireland for kind provision of drug samples.
- 1.Joyce C, Smyth WF, Ramachandran VN, O’Kane E, Coulter D (2004) Anal Bioanal Chem, submitted for publicationGoogle Scholar
- 2.Long PW, Risperidone Drug Monograph, Internet Mental Healthhttp://www.mentalhealth.com/drug/p30-r05.html
- 5.Remmerie BMM, Sips LAA, de Vries R, de Jong J, Schothius AM, Hooijschuur EWJ, van de Merbel NC (2003) J Chromatogr B 783:461Google Scholar
- 6.Brown TL, LeMay HE (1985) Chemistry, The Central Science. Prentice–Hall, New Jersey, USA, p 201Google Scholar
- 7.Fuh M-R, Tai Y-L, Pan WHT (2001) J Chromatogr B 752:107Google Scholar
- 12.Eerkes A, Addison T, Naidong W (2002) J Chromatogr B 768:277Google Scholar
- 15.Patton E, O’Donnell F (2002 and 2003) MRes theses, University of UlsterGoogle Scholar
- 16.Smyth WF, Joyce C, Ramachandran VN, O’Kane E, Coulter D (2004) Anal Chim Acta, in printGoogle Scholar
- 19.Conneely A, McClean S, Smyth WF, McMullan G (2001) Rapid Commun Mass Spectrom, 15:2076Google Scholar