Orthogonally spin-labeled rulers help to identify crosstalk signals and improve DEER signal fidelity

and improve DEER signal fidelity Markus Teucher1, Mian Qi2, Ninive Cati2, Henrik Hintz2, Adelheid Godt2, and Enrica Bordignon1 1Faculty of Chemistry and Biochemistry, Ruhr University Bochum, Universitätsstraße 150, 44801 Bochum, Germany 2Faculty of Chemistry and Center for Molecular Materials (CM2), Bielefeld University, Universitätsstraße 25, 33615 Bielefeld, Germany Correspondence: Enrica Bordignon (enrica.bordignon@rub.de)


Spectrometer
Continous wave (cw) EPR experiments for NO spin counting were performed at X band using a MiniScope MS 5000 spec-90 trometer (Magnettech by Freiberg Instruments). All pulsed EPR experiments were performed using a Bruker Biospin Q-band Elexsys E580 spectrometer equipped with a 150 W TWT amplifier from Applied Systems Engineering and a Bruker SpinJet-AWG (±400 MHz bandwidth, 1.6 GSa/s sampling rate, 14 bit amplitude resolution) in combination with a home-made Q-band resonator for 3 mm sample tubes (Tschaggelar et al., 2009;Polyhach et al., 2012).

Pulse parameters 95
All pulse experiments were performed using monochromatic pulses with a Gaussian amplitude modulation function, predefined as pulse shape 1 in Bruker Xepr 2.6b.119. In Xepr, the pulse length t p of a Gaussian pulse is defined as its time base (truncation at 2.2% of its maximum amplitude) which is related to its full width at half maximum (FWHM) by t p = 2 √ 2ln2 · FWHM ≈ 2.3548 · FWHM (Teucher and Bordignon, 2018).

DEER setup
100 DEER experiments were performed using the dead-time free 4-pulse DEER sequence (π/2) obs -(d 1 ) -(π) obs -(d 1 +T) -(π) pump -(d 2 -T) -(π) obs -(d 2 ) -(echo) (Martin et al., 1998;Pannier et al., 2000) with 16-step phase cycling (Tait and Stoll, 2016) using (0)-(π) for (π/2) obs and (π) obs , and (0)-(π/2)-(π)-(3π/2) for (π) pump . Gaussian π/2-and π-pulses at the observer frequency were created by varying the pulse amplitude at a fixed pulse length to maintain a uniform excitation bandwidth for the refocused echo (Teucher and Bordignon, 2018). pulse amplitudes within the spectral overlap of NO and Gd (at the maximum of the NO signal). Based on the different longitudinal relaxation times of the two spin probes, srt filtering allows independent addressability. Adjusting the pulse amplitudes allows matching the π-pulse lengths of NO and Gd. At an srt of 300 ms the predominant nutation signal contribution arises from the NO which has in the center of the dip at 100% AWG amplitude a 24 ns (10.2 ns FWHM) Gaussian π-pulse length. Decreasing the srt from 300 ms to 1 ms allows filtering for Gd, which matches the 24 ns π-pulse length at 22% AWG amplitude. This difference in AWG amplitude corresponds to a 12 dB difference in power between NO and Gd (Yulikov, 2015) based on their distinct transition moments.
synthesizing the frequency offset required for the pump pulse. More details about the DEER setups for the orthogonal spin probes are given in Fig. 3. The evaluation of the DEER data was performed using the Gaussian fitting routine of DeerAnaly-sis2019 (Jeschke et al., 2006). Gaussian fitting was chosen over Tikhonov regularization since it simplifies data evaluation for well-defined narrow distance distributions as it is the case for the rulers. In particular, it allows simultaneous fitting of compo-110 nents with very different distribution widths as required for the utilized samples. A side-by-side comparison of both methods is presented in Fig. S1 (SI Part B).  Figure 3. Three channel DEER setups. NO and Gd field-swept echo (FSE) spectra are shown as shaded gray areas with overlaid Gaussian pump and observer π-pulse excitation profiles simulated with EasySpin 5.2.2 (Stoll and Schweiger, 2006). In all setups, Gaussian observer pulses of 32 ns time base length (13.6 ns FWHM) (Teucher and Bordignon, 2018) were used in combination with a shot repetition time (  Preprint. Discussion started: 23 June 2020 c Author(s) 2020. CC BY 4.0 License.
3 Experimental results and discussion

Isolated rulers
The DEER characterization of the three individual rulers is shown in Fig. 4. Since both the NO-NO and the Gd-Gd rulers 115 contain only one type of label, we could probe per sample only one DEER channel, namely the NONO or GdGd channel, respectively, whereas the orthogonally labeled NO-Gd ruler gives access to all three DEER channels. Notably, the obtained dipolar frequencies and modulation depths of the isolated ruler samples with their corresponding distance distributions are characteristic sample-and setup-dependent parameters which will be used in the following to identify and quantify crosstalk signals in the ruler mixtures.

120
The NONO DEER time trace (blue) detected on the NO-NO ruler in Fig. 4(a) shows a dipolar frequency characterized by a 35% modulation depth, corresponding to a well-defined 2 nm distance. The GdGd DEER time trace (green) detected on the Gd-Gd ruler shows a dipolar frequency with a modulation depth of ≈ 3%, corresponding to a monomodal distance distribution centered at 4.7 nm (see Fig. 4(b)).
The time traces obtained on the NO-Gd ruler with the three DEER channels are shown in Fig. 4(c). The NOGd DEER time 125 trace (red) shows a defined dipolar frequency which is characterized by a 30% modulation depth and correlates with a 2.5 nm distance. Unexpectedly, the NONO DEER channel (blue) also contains a dipolar signal whose distance distribution coincides with the one obtained in the NOGd DEER channel. In contrast, the GdGd channel (green) contains a mere background function.
The absence of a dipolar modulation in the GdGd DEER channel proves that the NO-Gd ruler is monomeric in solution.
Therefore, we can conclude that the signal detected in the NONO DEER channel is a crosstalk signal. This crosstalk signal 130 originates from a residual excitation of the Gd spectrum overlapping with the nitroxide spectrum by the pump and/or observer pulses in the NONO DEER sequence. This crosstalk signal is significant, because its ≈ 4% modulation depth is in the order of 10% of the maximally achievable modulation depth for the spin-labeled NO-NO ruler (see Fig. 4(a)). We classify this signal as a NO-Gd crosstalk in the NONO DEER channel and designate it as X 1 .

135
In this section we investigate the appearance of crosstalk signals between the DEER channels in samples containing pairwise mixtures of the three rulers. The data are presented in Fig. 5.
The three DEER experiments performed on the mixture of the NO-NO ruler with the NO-Gd ruler in a 1:1 molar ratio are shown in Fig. 5(a). The NONO DEER channel contains the expected distance distribution of the isolated NO-NO ruler characterized in Fig. 4(a) with a slightly smaller modulation depth. The NOGd channel reproduces the signal obtained on the 140 isolated NO-Gd ruler previously shown in Fig. 4(b). The GdGd channel shows no dipolar modulation, in line with the absence of Gd-Gd rulers in this sample.
The NO-Gd crosstalk signal in the NONO channel (X 1 ) detected for the isolated NO-Gd ruler in Fig. 4(c) would be theoretically expected at 2.5 nm. Interestingly, this signal is not experimentally resolved in the mixture of NO-NO ruler with NO-Gd ruler.
As addressed before, this crosstalk signal is caused by a residual excitation of Gd spins via the NO-optimized pump and/or observer pulses. When considering that the pump pulse excites the Gd spins, and the observer excites the NO spins of the NO-Gd ruler, one needs to take into account that in this sample, only 1/3 of the NO observer signal in the NONO channel originates from the NO-Gd ruler, which would decrease the modulation depth of the crosstalk signal from the 4% detected on the isolated NO-Gd ruler (see Fig. 4(c)), to 1% for this sample, which is hardly detectable. When considering that the NOoptimized observer pulse excites the residual Gd spins of the NO-Gd ruler, one has to consider that the overall observer echo 150 has a predominant NO contribution and therefore an even lower modulation depth is expected. Accordingly, we suggest that the dominant signal contribution at 2 nm arising from the NO-NO ruler masks the NO-Gd crosstalk signal. The analysis of the sample containing the NO-NO ruler and the Gd-Gd ruler in a 1:2 molar ratio is presented in Fig. 5(b).
Both the NONO and the GdGd channels reproduce nicely the DEER signals obtained on the isolated NO-NO and Gd-Gd rulers.
As there is no NO-Gd ruler present in the sample, no signal would be expected in the respective DEER channel. Indeed, no https://doi.org/10.5194/mr-2020-15 Discussions Open Access Preprint. Discussion started: 23 June 2020 c Author(s) 2020. CC BY 4.0 License.
from that, the NOGd channel contains a secondary 3.5 nm distance which originates from a spectrometer-specific oscillatory signal of constant amplitude present in the time domain trace which is discussed in more detail in section 3.3. The modulation amplitude of this spectrometer artifact adds up to the modulation depth of the Gd-Gd crosstalk signal. The actual crosstalk signal has a modulation depth of about 3% which is (as in the case of the NO-Gd crosstalk in the NONO channel, defined 160 as X 1 ) in the order of 10% of the maximally expected modulation depth of 30% in this channel (see Fig. 4(c)) and therefore non-negligible.
The results of the experiments with the 1:2 mixture of the NO-Gd ruler with the Gd-Gd ruler are presented in Fig. 5(c).
The NONO DEER channel of this sample shows the NO-Gd crosstalk signal (X 1 ) as reported for the isolated NO-Gd ruler in Fig. 4(c). The identification of this crosstalk signal is facilitated by the absence of a real NO-NO distance. The GdGd channel, 165 due to the absence of any spectral overlap in our DEER setup, is intrinsically artifact-free and shows the expected pure Gd-Gd distance. In contrast, the NOGd channel contains, besides the expected NO-Gd distance, a Gd-Gd crosstalk signal defined as X 3 which is not fully resolved in the 1.7 µs time trace presented in Fig. 5(c). However, it is clearly visible in a longer time trace presented in section 3.3. X 2 in Fig. 5(b) and X 3 in Fig. 5(c) are both Gd-Gd crosstalk signals in the NOGd channel, however, we decided to keep a distinction in the names based on the absence/presence of a "real" NO-Gd distance which will 170 have an influence on the identification procedure discussed below.

NO-NO + + 2x NO-Gd
Gd-Gd ruler The DEER data obtained on the sample containing the NO-NO, NO-Gd and Gd-Gd rulers in a 1:1:2 ratio are presented in Fig. 6. Essentially, these data can be seen as a superposition of the data detected on the pairwise mixtures of rulers. The NONO DEER channel shows the distance distribution of the NO-NO ruler but lacks the X 1 crosstalk signal because it is masked by the intensity of the NO-NO DEER signal. Besides the expected NO-Gd ruler distance, the NOGd channel shows the Gd-Gd 175 crosstalk signal X 3 as in Fig. 5(c), which is clearly visible in the asymmetry of the time trace due to the underlying low frequency Gd-Gd signal. Finally, the GdGd DEER channel resolves the Gd-Gd distance free of crosstalk signals.
In conclusion, we identified three non-negligible crosstalk signals in the NONO and NOGd DEER channels and we showed that the GdGd DEER setup with the observer frequency placed on the maximum of the Gd signal and the pump frequency at the high field edge of the Gd spectrum ( Fig. 3(c)) is intrinsically crosstalk-free in all experimental conditions tested.

DEER channel crosstalk identification and suppression
The DEER channel crosstalk signals discussed in this work are named as follows: X 1 is an NO-Gd crosstalk signal in the NONO channel, while X 2 and X 3 are both Gd-Gd crosstalk signals in the NOGd channel but in the absence or presence of a "real" NO-Gd signal, respectively. An overview of all crosstalk signals that can be theoretically expected versus those that were experimentally detected using our samples and experimental setups is presented in Table S2 (SI Part B). The origin of all 185 crosstalk signals reported in this work is the spectral overlap between NO and Gd illustrated in Fig. 3(b), which does not allow a completely independent addressability.
The NO-Gd crosstalk signal in the NONO channel X 1 is unavoidable when probing the NONO DEER channel at Q band (see Fig. 3(a)) in presence of a NO-Gd distance. There are two possible origins for this crosstalk signal of the NO-Gd ruler: i) the observer pulses selectively excite the NO and the pump pulse excites the NO and sub-optimally the coupled Gd spins; ii) 190 the observer pulses excite the NO and sub-optimally the Gd while the pump pulse selectively excites the coupled NO spins. To test the effect of the pump pulse on the crosstalk signal, we decreased its power by 12 dB in order to optimize the pump π-pulse for the Gd spins. This resulted only in a small decrease in the modulation depth of the crosstalk signal (data not shown) which implies that there are contributions of Gd spins both in the pump and in the observer echo at the same time. Subsequently, both possibilities discussed above occur simultaneously. We could not find a strategy to minimize this crosstalk signal in the NONO 195 channel, however, if a real NO-NO distance is present, the contribution of this unwanted signal was found to be negligible (see Fig. 5(a)).
We focus now on the X 3 crosstalk signal (Gd-Gd crosstalk in the NOGd channel in the presence of a real NO-Gd distance) from Fig. 5(c). In Fig. 7(a) we present a long NOGd DEER time trace (red) detected on the sample from Fig. 5(c). The two distinct dipolar frequencies of the NO-Gd (high frequency) and Gd-Gd (low frequency) rulers are clearly visible in the primary 200 data. The distance analysis of this time trace using two Gaussians reveals both an NO-Gd distance peak and a Gd-Gd crosstalk distance peak. Since in the NOGd DEER channel, NO is pumped and Gd is observed (see Fig. 3(b)), the observer pulses excite only Gd spins, therefore, the Gd-Gd crosstalk signal originates from sub-optimally pumping the Gd at the NO position due to the spectral overlap. Decreasing the pump pulse power by 12 dB from optimally pumping the NO with a π-pulse to optimally pumping the Gd with a π-pulse (dark red in Fig. 7(a)) strongly decreases the modulation depth (from 15% to 2%; ≈ 1/7x) and  Figure 7. Crosstalk signal identification. Left, primary data with background fit (gray areas are excluded from data evaluation); middle, form factors with Gaussian fit (original data and modulation depth scaled data); right, obtained distance distributions. Color coding as in Fig. 4. In the NOGd DEER setup, NO is pumped and Gd is observed as illustrated in Fig. 3(b). Both X2 and X3 crosstalk signals arise from a partial excitation of the underlying Gd at the NO position in an NOGd DEER. (a) and (b) show how decreasing the pump pulse power from optimally pumping NO (red) to optimally pumping Gd (-12 dB, dark red) changes the signal-to-crosstalk ratio and thereby allowing the identification of the crosstalk signal. The relative change in modulation depth is considerably larger if a "real" NO-Gd signal is present together with the crosstalk signal (≈ 1/7x for X3 versus ≤ 1/2x for X2). The 3.5 nm distance marked with an asterisk originates from a spectrometer-specific artifact.
changes the ratio of the two distance peaks in favor of the crosstalk distance, as expected (see inset). The decrease in power of the pump pulse allows the identification of the X 3 -crosstalk signal since optimally pumping the Gd promotes the intensity of the Gd-Gd crosstalk distance (see next paragraph for additional information) while strongly decreasing the intensity of the NO-Gd distance. It is important to note that also the Gd-optimized pump pulse still partially pumps the NO and therefore the DEER trace contains a residual NO-Gd signal contribution.

210
In Fig. 7(b) the same approach is used to identify the Gd-Gd crosstalk signal in the NOGd channel in the absence of a NO-Gd signal (X 2 ) (see Fig. 5(b)). The NO-Gd signal (red) is a superposition of the Gd-Gd crosstalk signal corresponding to a distance centered at 4.7 nm and an additional 3.5 nm distance originating from a spectrometer-specific artifact. Decreasing the pump pulse power by 12 dB to optimally pumping the Gd (dark red) considerably suppresses the spectrometer-specific artifact contribution while slightly decreasing the main Gd-Gd dipolar modulation. The modulation depth contribution of the Gd-Gd 215 signal in this setup is about 1.25%, which is in line with the modulation depth obtained with the same setup on the isolated Gd-Gd ruler shown in Fig. S2 (SI Part B). Therefore, we found that by changing the pump power by 12 dB to optimize the inversion pulse for the Gd spins, the modulation depth of the Gd-Gd signal slightly decreases. We can conclude that if there is https://doi.org/10.5194/mr-2020-15 Discussions Open Access Preprint. Discussion started: 23 June 2020 c Author(s) 2020. CC BY 4.0 License.
a Gd-Gd crosstalk signal in the NOGd channel in the absence of a real NO-Gd signal, decreasing the pump power by 12 dB produces a small change in modulation depth (≤ 1/2x). This makes it possible to identify an X 2 crosstalk signal. In contrast, 220 if a real NO-Gd distance is present, as in the case of the X 3 crosstalk signal in Fig. 7(a), the overall modulation depth largely decreases (to ≈ 1/7x) and the ratio of the distance peaks in the overall distance distribution changes in favor of the crosstalk peak, which can be identified.
To actually suppress crosstalk signals in the overall distance distribution, swapping the pump and observer positions in the NOGd channel could be an option. Usually we pump the NO and observe the Gd spins (see Fig. 3(b)) to optimize the 225 modulation depth in the NOGd channel. In this setup, the observer echo is solely created by the Gd spins. Therefore, crosstalk signals are caused by the pump pulse which is optimized to selectively address NO spins but also partially excites Gd spins.
If the positions of pump and observer pulses are swapped, the observer pulses will be placed in the spectral overlap, while the pump pulse will excite only Gd spins. The advantage of the latter approach is that the observer sequence being composed of three pulses can act as a better filter for one spin species than a single pump pulse. This is illustrated in Fig. 8 where 230 field-swept echo experiments were performed at different pulse amplitudes using the refocused echo created by the DEER observer sequence (in absence of the pump pulse). At 100% pulse amplitude, the pulses are optimal π/2-and π-pulses for the NO, but strongly over-flipping the Gd spins, for which optimal π-pulses require an amplitude of approximately 14% (both determined via transient nutation experiments, data not shown). By lowering the pulse amplitude of the observer pulses to 50%, it is possible to favor even more the intensity of the NO spins in the refocused echo with respect to the Gd spins, therefore 235 increasing the selectivity of the observer sequence towards the wanted NO spins and suppressing the Gd contribution. Using these pulse amplitudes in the observer sequence should maximize the wanted NO-Gd signal, while minimizing the unwanted Gd-Gd crosstalk signal. The main disadvantage of this approach is that very long shot repetition times are required to observe NO Gd 1208 1212 1216 1220 1224 Figure 8. Crosstalk signal suppression. The absolute values of the complex FSE spectra detected using a refocused Hahn echo sequence (DEER observer sequence) on the 1:2 mixture of the NO-Gd and Gd-Gd ruler are shown. The experiments were performed at 10 K with a shot repetition time of 100 ms (filtering for the NO, see Fig. 2). With respect to the standard NOGd DEER setup shown in Fig. 3(b), the positions of pump and observer pulses were exchanged. The observer was placed at a frequency where 100% pulse amplitude corresponds to a π-pulse on the NO. Due to the distinct transition moments of NO and Gd, the relative intensities of the spectral contributions change when varying the pulse amplitudes. At 50% amplitude only the NO spectrum is refocused. Preprint. Discussion started: 23 June 2020 c Author(s) 2020. CC BY 4.0 License.
on the NO (100 ms for the NO with respect to 1 ms for Gd in the conventional setup), which makes DEER data acquisition impractically long for this combination of spin labels to achieve a satisfactory signal-to-noise. Additionally, the small fraction 240 of Gd spins excited by a Gaussian pump leads to a small modulation depth for the desired NO-Gd signal. The latter issue could be improved using phase-and amplitude-modulated broadband pump pulses, as it was previously shown as a way to improve modulation depths for Gd spin pairs (Doll et al., 2013;Spindler et al., 2013;Doll et al., 2015;Bahrenberg et al., 2017). Overall, this approach is interesting and might be of use for other pairs of orthogonal spin labels.

Conclusions and outlook 245
In this work we thoroughly investigated the appearance of crosstalk signals between the three possible DEER channels at Q-band frequencies on mixtures of NO-NO, NO-Gd and Gd-Gd rulers with non-overlapping distance distributions. Crosstalk signals in DEER experiments with two types of spin systems had been suspected in the literature before (Gmeiner et al., 2017;Teucher et al., 2019) but could never be unambiguously identified and characterized.
We experimentally detected a NO-Gd crosstalk signal X 1 in the NONO DEER channel in the absence of real NO-NO 250 distances and two Gd-Gd crosstalk signals X 2 or X 3 , in the NOGd DEER channel in the absence and presence of a real NO-Gd distance, respectively. We theoretically predicted a fourth crosstalk signal X 4 (see Table S2) which describes a Gd-Gd crosstalk in the NONO channel that was not experimentally detected. This is most likely due to: the low modulation depth that would be expected for this signal based on the non-perfect pump and observer pulses; the long Gd-Gd distance of 4.7 nm of the chosen ruler, which makes it more difficult to identify signals with very low modulation depth; the low spectral density of the Gd at 255 the position of the NO; and the large modulation depth of a real NO-NO signal if present.
Our experimental findings confirm that crosstalk signals can be expected in 4-pulse DEER experiments performed with the observer and/or pump pulses positioned in the region of spectral overlap between NO and Gd spins. Therefore, NONO and NOGd DEER experiments are prone to crosstalk signals, while the GdGd DEER channel with the setup suggested here is intrinsically crosstalk-free (see Fig. 3(c)).

260
All detected crosstalk signals are of experimental relevance when orthogonally-labeled biomolecular complexes are investigated, since they are in the order of 10% of the maximally expected modulation depth in the respective DEER channel for a doubly spin-labeled protein with 100% labeling efficiency. Signals of this strength are easily resolvable by state-of-the-art high-power Q-band spectrometers (Polyhach et al., 2012) and therefore entail the risk of data misinterpretation when unknown mixtures of orthogonally-labeled proteins are studied. Notably, we found that if a real NO-NO dipolar oscillation with a large 265 modulation depth is present in the NONO channel and the stoichiometric ratio of the different spin types is similar, the possible NO-Gd crosstalk signal is negligible, due to the dominating modulation depth of the real signal (in the order of 30 -40%) with respect to the expected 1 -2% of the crosstalk signal.
We were not able to find a spectroscopic approach to identify the NO-Gd crosstalk signal in the NONO channel (X 1 ), apart from an identification based on the comparison of the distance distributions detected with the NONO and NOGd DEER 270 channels on the same sample, which can be ambiguous. Therefore, if a dipolar oscillation is detected in the NONO DEER channel and the obtained distance distribution overlaps with the one detected in the NOGd channel, further analysis is required.
To clarify whether a crosstalk signal is detected, we propose to prepare an analogous sample with the Gd-labeled proteins exchanged with the unlabeled variants. If the NONO DEER channel is free of dipolar oscillations, the signal previously detected signal was a crosstalk signal; otherwise, if the same dipolar frequency is detected, then it was a real NO-NO distance.

275
For the Gd-Gd crosstalk signals in the NOGd DEER channel, which are the most relevant unwanted signals in the analysis of complex protein mixtures, we propose an identification strategy based on decreasing the power of the pump pulse positioned at the maximum of the nitroxide spectrum by 12 dB to optimally pump the Gd spins. This allows to unambiguously identify crosstalk signals X 2 and X 3 via relative changes in their modulation depth. If the overall modulation depth decreases only marginally (maximally to ≈50% of its initial value), the DEER signal is caused by Gd-Gd crosstalk and no real NO-Gd 280 distances are present. In contrast, if the modulation depth decreases to ≈15% of its original value and the primary time trace differs, then the signal is a mixture of a real NO-Gd signal and a Gd-Gd crosstalk signal. In this case, the crosstalk signal is of type X 3 and there must be a relative increase in intensity of the crosstalk signal with respect to the real signal contribution in the distance distribution. This change in relative intensities aids the identification of the crosstalk signal. Notably, the exact values of the relative changes of the modulation depths presented here are valid only in our experimental setup and need to be 285 calibrated for each setup using standard samples. Swapping the position of the pump and observer pulses in the NOGd DEER channel was found to be in principle promising to suppress the NO-Gd crosstalk signal, but experimentally impracticable for samples containing NO and Gd spins due to the prohibitively long shot repetition time of the experiment and the small modulation depths expected. Broadband excitation pump pulses may alleviate the modulation depth issue for spins with large zero field splittings and the long acquisition times would benefit if faster relaxing spin 1/2 labels are used.

290
It is important to note that the relative strengths of the crosstalk signals depend on the relative molar ratio of the different types of spin labels, as shown by a complete set of experiments performed on an independent set of samples with mixtures of the rulers in equimolar quantities (Table S3 and Fig. S3 to S5, SI Part B). Additionally, other experimental properties such as the relative modulation depths of the real signals and the relative widths of the distance distributions may modulate the relevance of the crosstalk signals in the overall data analysis. Therefore, in this work we identified possible problems arising 295 from crosstalk signals in three DEER channels when using NO and Gd mixtures, but the extent of the crosstalk signals and their relevance on data interpretation depends on the specific properties of the sample under investigation.
Q band currently offers the highest sensitivity to perform the three-channel DEER experiments with samples containing both NO and Gd spin labels on a commercial spectrometer. Gd spin labels would gain in sensitivity at higher frequencies thanks to the narrowing of the spectrum, however, the broadening of the NO spectrum would probably counterbalance these effects in 300 the NONO and NOGd DEER channels. In general, it would be insightful to have a multi-frequency approach and perform these types of experiments at Q-and W-band or higher frequencies to find the best-suited frequency band for each DEER channeland label-combination. The use of an AWG is advantageous for the proposed identification strategy due to the possibility to use Gaussian pulses (Teucher and Bordignon, 2018), which remove residual "2+1" pulse train signals increasing signal fidelity and to further explore additional benefits of broadband excitation pulses. We did not analyze the effects of multispin systems 305 with more than 2 spins in the mixture, but we can anticipate that appearance of ghost peaks (von Hagens et al., 2013) will measurements between nitroxides and low-spin Fe (III) ions, J. Phys. Chem. B, 119, 13 534-13 542, 2015.