Regions of Interest Multivariate Curve Resolution Liquid Chromatography with Data-Independent Acquisition Tandem Mass Spectrometry

New data-independent acquisition (DIA) modes coupled to chromatographic separations are opening new perspectives in the processing of massive mass spectrometric (MS) data using chemometric methods. In this work, the application of the regions of interest multivariate curve resolution (ROIMCR) method is shown for the simultaneous analysis of MS1 and MS2 DIA raw data obtained by liquid chromatography coupled to quadrupole-time-of-flight MS analysis. The ROIMCR method proposed in this work relies on the intrinsic bilinear structure of the MS1 and MS2 experimental data which allows us for the fast direct resolution of the elution and spectral profiles of all sample constituents giving measurable MS signals, without needing any further data pretreatment such as peak matching, alignment, or modeling. Compound annotation and identification can be achieved directly by the comparison of the ROIMCR-resolved MS1 and MS2 spectra with those from standards or from mass spectral libraries. ROIMCR elution profiles of the resolved components can be used to build calibration curves for the prediction of their concentrations in complex unknown samples. The application of the proposed procedure is shown for the analysis of mixtures of per- and polyfluoroalkyl substances in standard mixtures, spiked hen eggs, and gull egg samples, where these compounds tend to accumulate.

https://doi.org/10. 1016/j.envpol.2021.118555). The protocol used for the analysis of PFAS is specific for these compounds which encompasses a large family of Fcontaining substances. After an extraction with acetonitrile (not methanol which would extract a much larger amounts of lipids), a cleanup was performed with activated carbon which eliminates the lipids while PFAS are recovered in a theoretically fat-free extract. However, the clean-up is never 100% efficient and some lipids may remain in the final tissue, as visually observed as the extracts are yellow-colored. The analysis by HRMS detects "everything" with high sensitivity, so part of these co-extracted lipids elutes at the very end of the chromatogram (in agreement with their somewhat apolar nature). Therefore, the identification of these additional compounds (see Supporting   Table 1 and Supporting Figure 3 below) is evidence that ROIMCR could identify the "other" components present in the sample. We expected to detect additional Fcontaining compounds in the unknown gull-egg samples, but unfortunately this was not the case. Most studies reporting PFAS in biological samples report similar compounds as the ones we have herein identified.

UHPLC-qTOF -MS1 and MS2 analysis
Analysis was carried out by ultra-high-performance liquid chromatography (UPLC) coupled to a Bruker Impact II Q-TOF mass spectrometer. Parameters from the cold Apollo ion source were set as follow: negative electrospray ionization with a capillary at 2500 V, dry gas temperature at 200°C, drying gas flow at 8 L/min, nebulizer at 2 bars, and plate offset at 500 V. Q-TOF was tuned and calibrated with sodium formate using 14 m/z selected ions for mass error calculation. An accuracy of 0.1 ppm was achieved.
This calibration was carried out at every injection to monitor and detect any change in the signal. Acquisition was done in full scan mode at a mass range from 30 to 1000 m/z. Resolving power was 60.000 at full width at half maximum (FWHM) at m/z 200.
DIA was obtained using 6 eV energy for MS1 and 30 eV energy for MS2 using bbCID from Bruker technology.
A chromatographic column Phenomenex C18 Luna Omega (100 mm length × 2.1 mm inner diameter, 100 Å pore size, 1.6 µm particle size) was employed at 40ºC. The mobile phase consisted of (A) MeOH:ACN (80:20, v/v) buffered with 10 mM NH 4 Ac (aq) and (B) 10 mM NH 4 Ac (aq) aqueous solution. Initial conditions were 50% A and 50% B kept for 1 min, increased to 90% A in 9 min (2 min hold time) and reaching initial conditions in 6 min. Flow rate was set at 0.3 mL/min. The injection volume was 2 µL.
An XBridge C18 column (50 mm × 4.6 mm, 3.5 μm particle size) was used as trap column to remove background PFAS contribution from the mobile phase and LC tubing.

The Multivariate Curve Resolution Alternating Least Squares (MCR-ALS) method
This MCR-ALS optimization method (Tauler, R. Multivariate curve resolution applied to second order data, Chemom Intel Lab Syst., 1995, 30, 133-146 Variance explained (R 2 , %) Equation 3 R 2 = 100 are also given (in the last column) and further used for chemical confirmation of the different resolved components. Underlined m/z MS2 signals were coincident with the ones found using the target analysis of the Brucker instrument software. In addition, additional signals encountered by ROIMCR are given and compared to those from the target approach. In general, several additional MS2 signal ions were resolved by ROIMCR. The threshold value selected to consider these signals significant was 5% in and 20% of the maximum signal, depending on the case, and they were normalized to a maximum intensity value of 1 during MCR-ALS analysis. In general, there was a high level of coincidence, and the results obtained confirm that the proposed approach is adequate for the direct identification of the chemical compounds associated with the ROIMCR resolved components, using simultaneously the MS1 and MS2 ion signals from the data independent acquisition (DIA) system.

Explanation of Supporting
At the bottom of the Table, Table values), but they could not be identified in the currently available PFAS databases.

S6
Supporting  Exact m/z values for the precursor MS1 ions of all the PFAS present in the standard mixtures were first calculated. These m/z values were then searched in the experimental low energy MS1 chromatograms using the Compass Data Analysis (Bruker Scientific LLC. 2019, GmbH. Bremen, Germany) instrument software. During the m/z ions searching, an instrumental deviation of 0.5 Da was allowed. Differences between the theoretical and the experimental m/z values of the detected PFAS were always below 0.01 Da units. MS2 DIA fragment ions of every previously identified precursor ion were then searched by the instrument software at the same retention time using also previous knowledge on the fragmentation patterns of PFAS. In the upper part of the Figure 3, the elution profiles obtained by the data analysis software of the precursor ions in MS1 (top), MS 2 (middle) and both MS1 and MS2 ions (bottom) are given. Results shown in this Supporting Figure 3 for the target analysis approach are coincident with those obtained by the non-target ROIMCR analysis approach proposed in this work shown in Figure 3 of the main manuscript.