Strategies to identify and suppress crosstalk signals in double electron–electron resonance (DEER) experiments with gadoliniumIII and nitroxide spin-labeled compounds

Abstract Double electron–electron resonance (DEER) spectroscopy applied to orthogonally spin-labeled biomolecular complexes simplifies the assignment of intra- and intermolecular distances, thereby increasing the information content per sample. In fact, various spin labels can be addressed independently in DEER experiments due to spectroscopically nonoverlapping central transitions, distinct relaxation times, and/or transition moments; hence, they are referred to as spectroscopically orthogonal. Molecular complexes which are, for example, orthogonally spin-labeled with nitroxide (NO) and gadolinium (Gd) labels give access to three distinct DEER channels that are optimized to selectively probe NO–NO, NO–Gd, and Gd–Gd distances. Nevertheless, it has been previously recognized that crosstalk signals between individual DEER channels can occur, for example, when a Gd–Gd distance appears in a DEER channel optimized to detect NO–Gd distances. This is caused by residual spectral overlap between NO and Gd spins which, therefore, cannot be considered as perfectly orthogonal. Here, we present a systematic study on how to identify and suppress crosstalk signals that can appear in DEER experiments using mixtures of NO–NO, NO–Gd, and Gd–Gd molecular rulers characterized by distinct, nonoverlapping distance distributions. This study will help to correctly assign the distance peaks in homo- and heterocomplexes of biomolecules carrying not perfectly orthogonal spin labels.

Column chromatography was carried out on silica gel 60 M (Macherey Nagel) applying slight pressure. In the procedures reported below the size of the column is given as diameter × length. The material was loaded onto the column dissolved in a small quantity of the eluent. EI mass spectra were recorded using an Autospec X magnetic sector mass spectrometer with EBE geometry (Vacuum Generators, Manchester, UK) equipped with a standard EI source. ESI mass spectra were recorded using an Esquire 3000 ion trap mass spectrometer (Bruker Daltonik GmbH, Bremen, Germany) equipped with a nano-ESI source. ESI accurate mass measurements were acquired using an Agilent 6220 time-of-flight mass spectrometer (Agilent Technologies, Santa Clara, CA, USA) in extended dynamic range mode equipped with a Dual-ESI source or using a Q-IMS-TOF mass spectrometer Synapt G2Si (Waters GmbH, Manchester, UK) in resolution mode interfaced to a nano-ESI ion source.
The ratio of the components in a mixture was determined by 1 H NMR spectroscopy and is given in a molar ratio. Syntheses 2-Iodo-5-acetaminophenol (2). The published procedure [4] was followed making small changes. A solution of KI (9.7 g, 58.6 mmol) in H2O (250 mL) and a solution of KIO3 (6.1 g, 28.6 mmol) in H2SO4 (95%, 9.0 g) and H2O (250 mL) were separately but simultaneously added to a solution of 3-acetaminophenol (1) (13.0 g, 86.0 mmol) in aqueous H2SO4 (0.05 mol·L -1 , 1.5 L) within 2.5 h at such a rate that the addition rate of the solution of KIO3 was slightly faster than the addition rate of the solution of KI. Pale yellow crystals precipitated during the addition. The suspension was stirred at room temperature for 1.5 h. The precipitate was collected by filtration, dried in a desiccator over P4O10 at reduced pressure for 2 d and recrystallized in acetone. 2-Iodo-5-acetaminophenol (2) (20.0 g, 84%) was obtained as colorless crystals. 1  PEG-tosylate 3. Our procedure deviates slightly from the published one. [5] Under cooling with an ice-water-bath, tosylchloride (14.45 g, 75.8 mmol) was added to a solution of CH3(OCH2CH2)3OH (10.8 g, 65.8 mmol) in pyridine (10.8 mL, 131 mmol). The pale-yellow solution was stirred at 0 °C for 3.5 h and afterwards aqueous HCl (10%, 120 mL) was added.
The solution was extracted with toluene (3 x 75 mL), dried over MgSO4 and the solvent was removed under reduced pressure. Column chromatography (6.5 cm x 43 cm CH2Cl2/Et2O, 1:1) of the residual colorless oil (21.2 g) gave two fractions. The first fraction contained only PEG-tosylate 3 (0.53 g; Rf = 0.40). The second fraction (18.0 g; Rf = 0.40 and 0.42) was a mixture of PEG-tosylate 3 (17.68 g) and tosyl chloride (0.32 g). The tosyl chloride in this fraction was removed easily applying the reported procedure: [6] The second fraction was dissolved in CH2Cl2 (80 mL) and pyridine (2.80 mL, 34.5 mmol). Chopped filter paper (2.80 g) was added to the solution and the suspension was treated in an ultrasound-bath for 1 h.
Then the suspension was filtered through filter paper, the filter cake was washed with CH2Cl2 (100 mL) and the combined filtrates were washed with aqueous HCl (1 M, 50 mL). The layers were separated and the aqueous layer was extracted with CH2Cl2 (25 mL). The combined organic layers were dried over magnesium sulfate. After removal of the solvents the residue was combined with the above mentioned first chromatographic fraction. PEGtosylate 3 (18.2 g, 83 %) was obtained as a colourless oil. 1  ). The shifts of the aromatic protons is significantly different from those reported for this compound. [5] However, the shifts that we determined fit well to the shifts reported in the same reference for PEGtosylates with other lengths of the PEG chain.
Pegylated amide 4. This reaction was performed in dried glassware under argon using the Schlenk technique. NaH (60 wt% dispersion in mineral oil, 735 mg, 18.4 mmol) was suspended in dry THF (20 mL). 2-Iod-5-acetaminophenol (2) (2.00 g, 7.23 mmol) was added slowly in portions (caution: strong gas development and foaming). The milky viscous suspension was stirred at room temperature for 5 min. PEG-tosylate 3 (4.96 g, 15.57 mmol) was added, whereupon the color of the suspension turned from white to slight yellow. Dry THF (10 mL) was added to reduce the viscosity of the suspension, whereupon the suspension took on a beige color. The suspension was stirred at 60 °C for 70 h. Under cooling with an ice-water bath, aqueous HCl (1 M, 1 mL) and CH2Cl2 (20 mL) were added.
Additional aqueous HCl (1 M, 2.5 mL) was added to dissolve the precipitate. Water (15 mL) was added, the organic phase was separated, and the aqueous phase was extracted with CH2Cl2 (3 x 20 mL). The combined organic phases were washed with saturated aqueous Na2CO3 solution (20 mL) and then with water (20 mL), dried over Na2SO4, and filtered. The solvents of the filtrate were removed. Chromatography (5 cm x 52 cm, CH2Cl2/EtOH, 20:1) of the residual brown oil gave pegylated amide 4 (2.42 g, 59%; Rf = 0.30) as a pale yellow oil and pegylated amine 5a (0.65 g, 17%; Rf = 0.42) as a green oil. Analytical data of pegylated amine 5a: 1  was cooled to room temperature and extracted with CH2Cl2 (3 x 15 mL), the organic phase was dried over Na2SO4, and the suspension was filtered. The solvents of the filtrate were removed. Chromatography (3.5 cm x 53 cm, CH2Cl2/EtOH, 20:1) of the residual orangebrown oil (2.08 g) gave a mixture (1.88 g) of pegylated amine 5a (1.80 g, yield: 81%) and the corresponding deiodinated amine 5b (83 mg, yield: 6%) as a green oil. This material was used for the next reaction without further treatment. Analytical data of amine 5a are given above.
The aqueous phase was extracted with CH2Cl2 (3 x 10 mL). The combined organic phases were washed with water (2 x 10 mL), dried over Na2SO4 and filtered. The solvents were removed. Chromatography (3 cm x 17 cm, CH2Cl2/EtOH, 15:1, a few drops of Et3N were added to the eluent at the beginning of the elution) of the residual yellow oil gave TIPSprotected alkyne 6 (163 mg, 76%; Rf = 0.41) as a brown oil. 1 Table S1: Sample preparation of the NO-NO, NO-Gd and Gd-Gd rulers in a 1:1:2 ratio. Stock solutions of NONO, NOGd and GdGd rulers were prepared in water using spin counting information obtained from X-band continous wave (cw) EPR for nitroxide (NO) and Q-band field-swept echo (FSE) detected spectra for gadolinium (Gd) (data not shown). 50% v/v (20 µl) fully deuterated glycerol was added to all samples as cryoprotectant.
The DEER data obtained on these samples are presented in Fig. 4 -8 in the main text.       Table S4 are comparable across the samples since recorded using the same parameters and evaluated using the same method.    Figure S4: DEER data evaluation methods (related to Fig. 4 in the main text). Comparison of Gaussian fitting and Tikhonov regularization methods for DEER data evaluation. Left, primary data with background fit (gray areas are excluded from data evaluation); middle, form factors with fit; right, obtained distance distributions. The small distance peak at 3.5 nm in the GdGd DEER (green) is a residual artifact (see Fig. S8). The small peak at 2 nm in the NOGd DEER (red) is possibly due to residual orientation selection effects. Table S5 gives an overview of all DEER data presented in the main text. DEER data evaluation was performed using the Gaussian fitting routine in DeerAnalysis2019. For each sample and DEER channel, we present the temperature in K at which the experiment was performed, the length of the time trace in µs, the number of Gaussians used for the fit, the obtained modulation depth and root-mean-square deviation (rmsd) of the fit, the population of the Gaussians relative to the NO-NO (2 nm), NO-Gd (2.5 nm), Gd-Gd (4.7 nm) and artifact (3.5 nm).  Table S5: Overview of all DEER data evaluated using the Gaussian fitting routine presented in the main text.

Quantification
SI part B page 7

Equimolar ruler mixtures
The data shown in this section are independent repetions of the samples presented in the main text prepared with equimolar misxtures of the rulers. The mixtures were prepared using independently prepared stock solutions from the same batch of rulers.

Sample preparation
[µM]  Table S6: Sample preparation of the NO-NO, NO-Gd and Gd-Gd rulers in a 1:1:1 ratio. The DEER data obtained on these samples are presented in Fig. S5 to S7.   Traces are obtained with different srt, written in the legend, and in one case the power of the pump pulse was set to zero. The artifact is mostly visible in the Imaginary part of the signal, it is a sinusoidal oscillation with a period of 0.9 µs, which is residually present in some DEER time traces (e.g. in the NOGd and GdGd channel in Fig. 6). The artifact has no dipolar origin, being visible in the isolated Gd labels, it is independent on the spin system and has no hyperfine origin (being visible also in the Mn system). It can be strongly reduced if the pump pulse amplitude is set to zero (red trace). We could not find the origin of such artifact, therefore, we speculate that it is a spectrometer-or AWG-related oscillation.  Figure S9: Comparison of different DeerNets trained to fit specific types of distances (related to Fig. 6 and 7 in the main text). First column, primary data; second to fourth column, DeerNet distance distributions. In the main figures, only the Generic network is shown since this should provide the best overall performance and does not require any previous knowledge about the system. Here, we compare the outputs of networks specifically optimized to fit broad and narrow distance peaks. (a) Sample containing the NO-NO and the Gd-Gd rulers mixed in a 1:2 molar ratio (related to the NOGd DEER channel in Fig. 6). All DeerNets fit the crosstalk signal with high significance. The intensity of the spectrometer artifact signal (asterisk) depends on the chosen network. (b) Sample containing the NO-Gd and the Gd-Gd rulers mixed in a 1:2 molar ratio (related to the NOGd DEER channel in Fig. 7). All networks show a significant crosstalk distance peak at the Gd-Gd distance position.  Table S7: This table summarizes crosstalk signals between NO and Gd that are possible based on the spectral overlap in the observer and/or pump position ("poss.") versus those that were actually experimentally detected ("exp.") in all samples using the conventional DEER setups given in Fig. 3a-c. The white cells left represent signals which cannot be expected in the setup/samples used. The other cells are colored as follows: desired dipolar frequency + / crosstalk X i / not detected . The crosstalk signals are: X 1 is an NO-Gd crosstalk signal in the NONO channel; X 2 (X 3 ) is a Gd-Gd crosstalk signal in the NOGd channel in absence (presence) of NO-Gd distance; X 4 is a Gd-Gd crosstalk signal in the NONO channel. The table shows data obtained on the 1:1:2 mixtures, which were shown to promote the Gd-Gd crosstalk signals.    Figure S11: Comparison of GdGd DEER setups (related to Fig. 7 in the main text). Left, primary data with background fit (gray area is excluded from data evaluation); middle, form factors with Gaussian fit; right, obtained distance distributions. The data plotted as a green solid line were obtained using the standard GdGd DEER setup shown in Fig. 3(c) with the pump pulse placed on the maximum of the Gd spectrum and the observer 100 MHz lower in frequency. The second setup (green dotted line) is basically the same as the NOGd DEER setup shown in Fig. 3(b) with the observer on the spectral maximum and the pump placed 280 MHz higher in frequency but in this case with a pump pulse optimized in power to optimally excite the Gd. The data obtained with the standard GdGd DEER setup are characterized by a 3.5% modulation depth while the second setup shows a 1.25% modulation depth due to the lower spectral density at the pump position.