Non-target screening, suspected-target screening and target screening – of technologies and philosophies, databases and crafts

Detective work with molecular accuracy

von PD Dr Thomas Letzel

The three terms in the title, as well as “Known Unknowns” and “Unknown Unknowns” are new keywords that are currently con­fusing the analytical water scene. The procedure in use of just these ­technologies, however, often is not consistent yet. This article will now try to put the many different terms and approaches into some ­semblance of order. At the same time, it will report on pragmatic appli­cations.

Non-target screening, suspected-target screening and target screening as well as “Known Unknowns” and “Unknown Unknowns” are new keywords that are currently increasing in interest in water analysis. The search for unknown or expected molecules in the matrix water brought about new instrumental technologies and analytical strategies. A great share is based on liquid-chromatography separation (LC) with atmospheric pressure ionisation (API)-coupled mass spectrometric detection (MS) and is technologically very mature.

Analytical screening (by LC-API-MS): What is this?

Let us first have an entirely un-scientific (but properly scientifically quoted [1]) look at what the public platform Wikipedia contributes to the term of “sieving”. The word screening is mainly used in the medical context here, and the results of screening studies are medical in origin as well. However, there also is a more general definition: “Screening (screening, grid application, ­selection, raying) is a systematic test method that is used for identifying certain properties of the test objects within a defined test area – usually comprising of a large number of samples or persons. Screening therefore is an orienting screen test aligned with ­certain criteria.” [1]

Referring this definition to the currently highly popular screening area of analytical chemistry in water research, it may mean: ­“A LC-API-MS screening is a systematic analytical test method that is used to recognise the organic molecules within the matrix water, characterising them physico-chemically and determining their quantity. This screen test is thus aligned with the criterion of organic water contents determination.”

The fact that screening is a clearly aligned but also limits application is important here. This “screening for organic water contents” says nothing about any other components such as micro-organisms or inorganic substances, nor are there any indications of toxicology or biological function. For this, further analytical screening technologies are necessary. In the best case, they can be ­coupled with this one [2].

Now, for the first time we stumble across the term of “non-target screening”, which must be a “targeted analytic” in spite of its name.

Therefore, there cannot be any “non-target screening” in the actual sense.
Has the article arrived at its end now? As the following text shows: It has not!
Let us first have a look at figure 1, which tries to illustrate classification of the terms. Even if entirely untargeted analytic is not possible, these tests still can be performed without a prior objective. The results are two types of untargetedly detected target molecules, the “Unknown Unknowns” and the “Known Unknowns”.

The class of “Unknown Unknowns” cannot be uniquely identified by any currently available evaluation method and ­also cannot be assigned to any known ­molecules. There remains only the option to determine structure by similarity comparisons with known molecules, such as hydrophobicity, molecular mass or mass-mass spectrometric fragmentation patterns.

The class of “Known Unknowns” has ­recently been described very nicely by J.L. Little et al. as “Known Unknowns – that is, species known in the chemical literature or MS reference databases, but unknown to the investigator.” [3]. This definition conceals another dilemma of non-target screening. “not targeted” does not always mean “unknown”. Non-target screening thus often is quick to use the same path as “suspected-target screening”, which implements identification of expected molecules using analytical and chemical databases. The sus­pected-target screening, however, will start out with a list of expected substances and can be specifically aligned with this.

This approach and the “Known Unknowns” share the fact that there initially are no real reference substances but that substances can be identified via substance databases, chemical databases, analytical or mass spectrometric databases and/or reconciliation with in-silico predictions.

If this search or interpretation leads to success, the corresponding real reference substance can be synthesised and used. This clear identification and the further use are consistently also referred to as “Target Screening”, which contains the quantitative analysis using isotopic labelled reference materials.

Non-target screening: Possibilities of an ­analytical technology

Now using the non-target screening approach in the text consistently as a technology ­approach and less strictly according to the definition, we can continue in good spirits:

This approach typically is a water ­analytic with a) sampling, b) sample preparation, c) liquid-chromatographic separation, d) ionisation and ion transfer of the contained analytes and e) mass spectrometric detection of the same (poss. incl. structure determination by tandem-MS).

Now, clearly targeted and analyte-­focused methods are used (even though these are preferably kept as universal as possible to do more justice to the name):

a) Sampling

Sampling is typically performed using ­different sampling devices (e.g. trowel or hose); the sample is then transferred to sampling vessels and stabilised with chemicals or by storage in the frozen ­condition. All steps may falsify our sample, e.g. by loss, enrichment and modification of the organic molecules. Practical sampling that is correct as well as impeccable under statistics aspects must be ensured here. In the latter case, the sampling area as well as the sampling duration (e.g. short sample intervals [4] or 24 h mixed sample [5]) plays a very important role.

b) Sample preparation

The best sample preparation is, of course, no sample preparation at all!
Since this is frequently not possible, an attempt is made to separate the desired analytes (by enrichment) from the undesired matrix molecules (by reduction) by many enrichment and reduction methods. The most common ones are solid phase extractions (SPE); e.g. by online-SPE [5,6] and high polarity value SPE [6]), where – as in sample separation - mainly chromatography material is used.

c) Sample separation

Classically, the reversed phase – high ­performance liquid chromatography (also RP-HPLC) with C18-phases or other unpolar to medium-polar phases is used for separation of organic analytes in water. This has increasingly become established in the recent past, since this separation can be very well connected with mass spectrometric detection. We also refer to the following sections for this. A currently completed round robin test on standardisation of the retention times achieved in turn facilitates transferability of the results interlaboratory results [7], with analytical data becoming easier to reconcile.

For some time, the numbers of users of the hydrophilic interaction liquid chromatography (HILIC) have been increasing. This is due to the also-excellent possibilities of coupling to the MS as well as due to its ability of separating polar molecules. A very current application with RPLC-HILIC-MS-coupling [8] even promises increased use of chromatography in the expanded polarity range (e.g. logD -5 to logD +5). This value reflects, among others, the hydrophobicity that was characterised highly variably over the last decades [9].

Thus, the separation properties of organic molecules can be used as specific physico-chemical indices for characterisation of the individual molecules.

d) Ionisation and ion transfer

Molecules must ionically be transferred to the gas phase after separation so that they can enter the mass spectrometer. Particularly observe the nature of molecules (e.g. their functional groups) as such, but also the characteristics of the mobile phase (e.g. signal-influencing matrix effects or pH-­values). Finally, ion sources with different properties (e.g. API-techniques, such as ESI, APCI and APPI) in positive and/or negative ionisation are mainly used [9,10].

The technology of ion mobility increases in importance. It permits another independent molecule separation (after entry into the mass spectrometer) and is driven by the geometry of the molecules [11].

e) Mass spectrometric detection (incl. structural analytic by tandem-MS)

The mass spectrometric detection can be used highly variably, both in the type of application and, specifically, in the type of instruments. We refer to the present literature for detailed treatment of the mass spectrometric possibilities (e.g. [9,10]).

Some questions are of special importance here, however: Which detection mode, which detection range, accurate, internal standard, neutralisation effects, isotopic shift [12], adduct formation, which sensitivity, in-source fragmentation, tandem-MS, significant fragments, which dissociation technologies, which analytical evaluation software, etc.?

By combining these analysis methods, more or less “non-targeted” measurement is now possible. However, the data gained can often be processed further well – as shown below.

Suspected-target screening: Possibilities of different databases

The suspected target screening typically starts with a list of analytes to be ­measured (since expected). The results of these measurements are subsequently subjected to the “molecule search” (see fig. 1). As with the “Known Unknowns”, most ­molecules are stored in the databases [13] (but not specifically present in the lab as reference substances). The “suspected targets” and the “Known Unknowns” can be further characterised using substance databases (such as STOFF-IDENT), chemical databases (such as Chemspider or Chemicalize), analysis or mass-spectrometry databases (such as DAIOS, MassBank, local databases [5,6,14] or commercial MS spectrum databases [15]) as well as in reconciliation with in-silico predictions (such as the UM-BDD, EPI SuiteTM, filter for MS-similarity trees [16] or MetFrag). If these databases lead to clear results, the user can procure (or synthesize) the corresponding substances and then use them in target screening (see * in fig. 1).

Target screening: Possibilities of the LC-API-MS-“craft”

Identification from the non-target and suspected target screening lead to use of reference substances (see above and * in fig. 1) and their use in Target Screening. By use of isotopic labelled reference substances, concurrent quantity analytic of many molecules is possible as well (examples include 72 analytes by “MRM” [5] and 88 analytes by “SRM” [6]).

Application of different screening strategies in water

a) Use of conventionally gained non-target screening data

A pragmatic path of using results from the non-target screening is direct comparison of analytical RT-MW-plots (retention time (RT) against molecular weight MW) at different times (also see fig. 2a –upper left), received using LC-API-MS.[4] This strives not solely for the identification of individual analytes but also compares the analytical results by “imaging” and takes their differences into consideration.

b) Use of the suspected-target screening path for “Known Unknowns” ­(with data of accurately measuring MS)

Use of LC-QqToF [14] and Orbitrap-MS [17] (this applies to FTICR-MS and IT-ToF-MS accordingly) in the non-target screening leads to the above RT-MW-plots with accurate masses (i.?e. elemental formula) for the respective molecules. Thus, they can be subjected to the molecule search and characterised as “Known Unknowns” via databases (see fig. 2a).

Molecules that cannot be identified or assigned via this path thus correspond to the current “Unknown Unknowns” and thus must be supplied to the “Similarity search” (see fig. 1 –left and [11]).

c) Use of the suspected-target screening path for “Known Unknowns” (with data of inaccurately measuring MS)

Using the non-target strategy with inaccurately measuring mass spectrometric as in the LC-QqQ (this also applies to IT and linearIT), the databases can be used as well – with limitations (fig. 2b). Due to the missing mass spectrometric accuracy, no elemental formulas can be generated in this approach and equipment sensitivity will be very low at the large mass spectrometric detection range of the quadrupole devices. Considering these disadvantages, these ­devices are still justified in these screening approaches. This is particularly important because these mass spectrometers are much easier and cheaper to operate and thus also are very common in analytical labs. Thus, it is possible that labs, which typically operate target screening, use their devices for non-target examinations as well.

d) Use of target screening

The known analytes are sensitively detected and quantified with e.g. LC-QqQ-MS and the respective MRM-methods [5,6].

Summary

The time in which analytical chemistry was considered an auxiliary science is now actually past. Due to the current diverse use of the described LC-API-MS screening techniques in many different disciplines (e.g. in proteomics [9], metabolomics [16], human samples [15], in foods [18] as well as in wine [8]) analytic even enters the foreground. Now, however, great care must be taken to ensure that the earlier problem is not turned around, pushing the applicative disciplines into the background. Consider: Only a strongly linked analysis is a good analysis system. Therefore, “the analysts” are only excellent when they also look over the rims of their teacups, get comfortable in the other disciplines and closely cooperate with their representatives.

The screening techniques as such are currently subject to great interest. This up-to-datedness is also the reason for most quotes from this publication from 2012/2013. It is certain that the subject will remain in the focus and that further papers will deal with screening in water [10]. Now we are ready for this and for the screening terms.

Literature
[1] http://de.wikipedia.org/w/index.php?title=Screening&oldid=116082118 (English translation from the German version published 28.03.2013) last called 20th March 2014
[2] Türk, J. (2012), Labor&More, 8.12, 46–49
[3] Little, J.L., et al. (2013), LCGC Europe 26 (3), 163–168
[4] Müller, A. et al. (2011), Chemosphere 85, 1211–1219
[5] Wode, F. et al. (2012), J. Chromatogr. A 1270, 118–126
[6] Huntscha, S. et al. (2012), J. Chromatogr. A 1268, 74-83
[7] http://www.lw-online.de/fileadmin/downloads/aktu_fachbeitraege/Vortrag2_RiskIdent1211_RTI.pdf ; last called 20th March 2014
[8] Greco, G. et al. (2013), J. Sep. Sci. 36 (8), 1279-1388
[9] Berkemeyer, C. & Letzel, T (2007), LC-GC AdS 2(4), 36–45
[10] Farré, M. et al. (2012), J. Chromatogr. A 1259, 86-99
[11] Menikarachchi, L. C. (2012), Anal. Chem. 84, 9388-9394
[12] Jobst, K. J. et al. (2013), Anal. Bioanal. Chem. 405, 3289-3297
[13] Zedda, M. & Zwiener, C. (2012), Anal. Bioanal. Chem. 403, 2493-2502
[14] Masiá, A. et al. (2013), Anal. Chim. Acta 761, 117–127
[15] Oberacher H., et al. (2013), Anal. Chim. Acta, 770, 121–131
[16] Jin, Y.et al. (2013), Anal. Chim. Acta 768, 111-117
[17] Bijlsma, L. et al. (2013), Anal. Chim. Acta 768, 102-110
[18] Gomez-Ramos, M.M. et al. (2013), J. Chromatogr. A, 1287, 24–37

Photo: ©Photocase.de|Fiebke