Preliminary investigation of deoxyoligonucleotide binding to ribonuclease A using mass spectrometry: An attempt to develop a lab experience for undergraduates

Deoxyoligonucleotide binding to bovine pancreatic ribonuclease A (RNase A) was investigated using electrospray ionization ion-trap mass spectrometry (ESI-IT-MS). Deoxyoligonucleotides included CCCCC (dC 5) and CCACC (dC 2AC 2). This work was an attempt to develop a biochemistry lab experience that would introduce undergraduates to the use of mass spectrometry for the analysis of protein-ligand interactions. Titration experiments were performed using a fixed RNase A concentration and variable deoxyoligonucleotide concentrations. Samples at equilibrium were infused directly into the mass spectrometer under native conditions. For each deoxyoligonucleotide, mass spectra showed one-to-one binding stoichiometry, with marked increases in the total ion abundance of ligand-bound RNase A complexes as a function of concentration, but the accurate determination of dC 5 and dC 2AC 2 dissociation constants was problematic.

RNase A+dC 2 AC 2 ligand-bound form of RNase A (with one dC 2 AC 2 ligand)

Introduction
Bovine pancreatic ribonuclease A (RNase A) is an endoribonuclease (EC 3.1.27.5) that hydrolyzes RNA. It is a small single chain polypeptide (124 amino acids) containing four disulfide bridges and is known for its significant stability 1 . RNase A has been called "the most studied enzyme of the 20 th century" and it has seen wide use as a model protein in biochemical and biophysical experiments 1 . Undergraduate life-science majors often learn of RNase A as part of a biochemistry course in the context of the Nobel Prize winning protein folding experiments performed by Christian Anfinsen 2 . Students may also be familiar with the need to inhibit ribonucleases when working with RNA in the lab, often accomplished with diethyl pyrocarbonate, or will have learned about the role of ribonucleases in microRNA biology 3 . Still others may recognize RNase A as an example of an enzyme that employs general acid-base catalysis as part of its chemical mechanism 4 . Thus, RNase A is an excellent model for undergraduate lab experiments, not only because it has been extensively studied, but also because its use presents an opportunity to reemphasize important concepts in biochemistry and biology.
The application of mass spectrometry to the analysis of biomolecules has made an enormous impact in the life sciences. Protein identification, the characterization of protein modifications, and the quantification of biomolecules using mass spectrometry are commonplace. Of these, protein identification is the most established in an undergraduate teaching lab 5-10 . Numerous other biological applications of mass spectrometry have existed for many years, but some of these are arguably, less broadly appreciated, and this is especially true for undergraduates. Native mass spectrometry is an approach based on electrospray ionization, where biomolecules are sprayed from a non-denaturing solvent 11  This work was an attempt to develop a biochemistry lab experience that would introduce undergraduate life-science majors to the use of mass spectrometry for the analysis of proteinligand interactions. Two deoxyoligonucleotides, CCCCC (dC 5 ) and CCACC (dC 2 AC 2 ), were investigated for their ability to bind RNase A. Titration experiments were performed using a fixed RNase A concentration and variable deoxyoligonucleotide concentrations. Samples at equilibrium were infused directly into our ESI-IT-MS under native conditions. The relative simplicity of the sample preparation and instrument operation (by direct infusion) were viewed as desirable features for an undergraduate teaching lab. Data analysis was also straightforward. Herein is described the results of this preliminary investigation. This work differentiates itself from the abovementioned RNase A ligand binding studies (using mass spectrometry) by the experimental conditions employed, which includes the identity of the investigated ligands and the type of mass spectrometer used 12,15,16 .

REVISED
in LC-MS grade water (Thermo-Fisher Scientific, Waltham, MA, USA). Ammonium acetate (NH 4 OAc) was LC-MS grade (#73594, Sigma-Aldrich). HPLC-purified deoxyoligonucleotides with the sequence "CCCCC" (dC 5 ) and "CCACC" (dC 2 AC 2 ) were obtained from ThermoFisher and the stock solutions (200 μM) were prepared in LC-MS grade water. Samples were prepared in 1.5 mL microcentrifuge tubes as indicated in Table 1. Six replicates were prepared and analyzed for "Sample 1" whereas "Samples 2-5" were prepared and analyzed in triplicate. Each sample was mixed by micropipetting, and incubated at room temperature for ten minutes, prior to analysis.

Mass spectrometry
Samples were analyzed with a Thermo LCQ Advantage iontrap mass spectrometer equipped with an electrospray ionization source. The instrument was operated in positive ion mode using a 4.5 kV spray voltage, 60°C capillary temperature, 200 ms inject time, 10 microscans, and nitrogen sheath and aux gas settings of 30 and 15, respectively. The instrument was tuned on the +8 charge state of free RNase A at m/z 1723.7 (Table 2). Each sample was subjected to direct-infusion at 2.5 μL/min using the LCQ syringe pump and full-scan mass spectra (m/z 1500-1950) were collected for two minutes. The upper m/z range was capped at 1950 to exclude the +7 charge state of free RNase A, which in its various adduct forms, began at m/z 1955.5 ( Table 2). The rationale was that the +7 charge state of the ligand-bound forms of RNase A were above m/z 2000, which made +7 data incomplete and unusable ( Table 3).

Determination of total ion abundance
To facilitate determination of total ion abundance, tables of predicted m/z values for free RNase A (Table 2) and the ligandbound forms of RNase A (RNase A+dC 5 and RNase A+dC 2 AC 2 ) ( Table 3) were constructed. A series of 98 Da adducts were included in Table 2 and Table 3 due to their presence in the mass spectra of this work, and that of earlier studies 12,15 . These  Where X=0 (no phosphate adduct), X=1 P i (+98), X=2 P i (+196), X=3 P i (+294), X=4 P i (+392), X=5 P i (+490). . Other RNase A studies have assigned these adducts as phosphate, and so each 98 Da adduct (X) in this work was designated as "P i " ( Table 2 and Table 3) 12,15 . Although mass spectra showed that free RNase A had up to 8 P i adducts ( Figure 1A and 1F), only the 0-5 P i adduct forms of free RNase A and its ligand bound forms were used. This restraint was necessitated by the predicted m/z overlap of the ligand-bound forms of RNase A (with P i adducts >5) with the m/z of free RNase at the +7 charge state. The "Qual Browser" feature of Xcalibur 1.4 SR1 software (Thermo) was used for analysis of each *.raw file. For each sample, mass spectra comprising the two-minute data collection were averaged. The "spectrum list view" was used to obtain intensity data for all of the ions in the ranges comprising the +8 charge state (with 0-5 P i adducts) for free RNase A (m/z 1710.7-1772.9), RNase A+dC 5 (m/z 1883.7-1945.9), and RNase A+dC 2 AC 2 (m/z 1886.7-1948.9). The intensity data for all ions in each m/z range were added to give the "total ion abundance" of the free (Ab (P) ) and ligand-bound forms (Ab (PL) ) of RNase A. The total ion abundance for the ligand-bound forms (RNase A+dC 5 and RNase A+dC 2 AC 2 ) were plotted as a function of [deoxyoligonucleotide] using GraphPad Prism 7.
Calculation of total ion abundance ratio and K d The total ion abundance ratio was determined at each [deoxyoligonucleotide] using the method described by Kitova et al. 13 , where for a 1:1 protein-ligand complex, the total ion abundance ratio (R) is calculated using the total abundance of all ligand-bound ions (Ab (PL) ) and the total abundance of all free protein ions (Ab (P) ) as shown in Equation 1: R= Ab (PL) /Ab (P) = [PL] eq /[P] eq [1] The total ion abundance ratio (R) is used with the initial ligand concentration ([L] 0 ) and initial protein concentration ([P] 0 ) to calculate the association constant (K a ) using Equation 2 13 : The K d can then be calculated as the reciprocal of the K a value. Table 1 indicates that samples contained an overall [RNase A] of 40.9 μM. Relatively low signal intensities observed for the +8 charge state of free and ligand-bound forms of RNase A necessitated this concentration, which was higher than the 5-20 μM RNase A used by others in nESI-Q-TOF-MS experiments 12,15,16 . Table 2 and Table 3 present predicted m/z values for free RNase A and the ligand-bound forms of RNase A (RNase A+dC 5 and RNase A+dC 2 AC 2 ) with multiple P i adducts, which correlated well with observed m/z values ( Figure 1). Upon increasing the concentration of dC 5 , the total ion abundance of free RNase A was found to decrease in intensity while the total ion abundance of RNase+dC 5 was found to increase in intensity, which suggested 1:1 stoichiometry for the dC 5 :RNase A interaction ( Figure 1A-E). Similar results were seen for the titration using dC 2 AC 2 ( Figure 1F-J). Table 4 presents total ion abundance data for free RNase A in samples that contained no added deoxyoligonucleotide. Total ion abundance data for free RNase A across six replicates gave a RSD of 16.4% (Table 4).   approximately 20% or less (Table 5). A plot of the total ion abundance for free RNase A, RNase A+dC 5 , and RNase A+dC 2 AC 2 as a function of [deoxyoligonucleotide] is shown in Figure 2. The total ion abundance for RNase A+dC 5 and RNase A+dC 2 AC 2 increased until 20 μM deoxyoligonucleotide, but decreased at 40 μM ( Figure 2).

Conclusions
This preliminary work demonstrates the potential and pitfalls of a LCQ ESI-IT-MS instrument to investigate protein-ligand interactions in an undergraduate teaching lab. Even though dC 5 and dC 2 AC 2 binding to RNase A are clearly illustrated in Figure 1, the presence of the P i adducts complicated the mass spectra and broadened the signals for free RNase A and the ligand-bound forms of RNase A. In-source collision-induced dissociation was explored to reduce P i adduct formation, but it appeared to disrupt the RNase A+dC 5 and RNase A+dC 2 AC 2 complexes, and so this approach was abandoned (data not shown). Although it was not attempted, centrifugal desalting of the RNase A stock solution might have eliminated P i adducts and improved the quality of the mass spectra in Figure 1. As an added benefit, in the context of an undergraduate lab, desalting would also introduce students to a common sample preparation technique. It is unclear why the decrease in the total ion abundance for the ligand-bound forms of RNase A was observed at higher deoxyoligonucleotide concentrations ( Figure 2). Previously, the ion intensity ratio of free RNase A to the RNase A+cytidine 2′-monophosphate (2′-CMP) complex was observed to vary with charge state as follows: +8 (0.65), +7 (0.73), +6 (1.1) 12 . This led Zhang et al. to suggest that either the binding of ligand, or the presence of ligand in the analyzed RNase A samples, created a change of the charge state distribution for the protein-ligand complex 12 . In the present work, the binding of deoxyoligonucleotide, or the presence of deoxyoligonucleotide in samples, could have shifted some of the total ion abundance of free and/or ligand-bound RNase A from the +8 charge state to lower charge states, which were beyond the mass range of our ion-trap mass analyzer. This highlights an inherent limitation of this work, which was the inability to gather data for all free and ligand-bound RNase A charge states. Kitova et al. stated the importance of including all ligand-bound and free protein ions in the calculation of R, and emphasized that the "sometimes-used practice" of employing a particular charge state to determine K a should be avoided 13 . Thus, the lack of data for the +7 and +6 charge states of RNase A hindered accurate collection of total ion abundance data, which may have affected calculations of R and led to the negative K d values at low ligand concentrations (Table 6). Other factors to consider, that could have affected measurements, include non-ideal ionization conditions and non-specific ligand binding. Benkestock et al. showed that instrument-derived parameters (e.g. capillary-to-cone distances) could affect the proteinligand complex to free protein ratio 19 . They also demonstrated that compared to pneumatically assisted ESI, which was used in this work, nESI better reflects the equilibrium between free protein and protein-ligand complexes in solution. Furthermore, Kitova et al. noted that changes in the magnitude of K a , with changes in ligand concentration, might indicate nonspecific ligand binding 13 . As seen in Table 6, K d values varied with the deoxyoligonucleotide concentration. Therefore, it is reasonable to suspect that non-specific binding may have contributed to the decreased total ion abundance of the ligand-bound forms of RNase A at higher ligand concentrations ( Figure 2). Notwithstanding these possibilities, the positive K d values in Table 6

Competing interests
No competing interests were disclosed.

Grant information
This work was supported by the College of Natural Sciences and the Department of Chemistry and Biochemistry at California State University-Chico.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
on ribonuclease A using nano-electrospray ionization mass spectrometry.
students can often learn more when things do not work exactly as expected. The experiment presented here offers an opportunity for students to understand equipment limitations and may help students obtain a stronger understanding of how the diversity of charged states impacts the ability to collect an accurate mass spectra.

Is the work clearly and accurately presented and does it cite the current literature? Yes
Is the study design appropriate and is the work technically sound? Yes

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed.
useful for undergraduate students that are early in their scientific career.
For future investigations and undergraduate studies, it would be Suggestion ( ): no change requested beneficial to use centrifugal desalting columns in an attempt to remove the phosphate adducts. This would result in improved ion response, less convoluted spectra, and would introduce undergraduate students to common sample preparation used in native-like protein MS experiments.
For future investigations and undergraduate studies, a native MS Suggestion ( ): no change requested technique that has been used to address the upper m/z limitation of ion trap is to "supercharge" proteins . This can be done by adding as low as 1% v/v sulfolane or m-nitrobenzyl alcohol to the sample.
There are two issues that the Additional comments to conclusion ( ): minor revisions requested authors addresses regarding the data from this study. (1) The observation of negative dissociation constants and (2) decreasing PL abundance at the highest L concentration. To point (1), although these experiments are being performed under native conditions (i.e. non denaturing solvents), there are still other factors during electrospray ionization that need to be considered during native experiments. For example, Benkestock, et al show data that suggests that the "capillary-to-cone" distance and the electrospray probe internal diameter can affect the PL to L ratios. To point (2), the decrease in PL abundance at the highest L concentration may be due to non-specific binding as described in the already cited Kitova et al. (2012) (Section 2.4 and Figure 3). The author should add a couple sentences to the conclusion addressing how "non-ideal" ionization conditions and non-specific binding could have affected the measurements.

Are the conclusions drawn adequately supported by the results? Partly
No competing interests were disclosed.

Competing Interests:
Referee Expertise: Native Mass Spectrometry I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias You can publish traditional articles, null/negative results, case reports, data notes and more The peer review process is transparent and collaborative Your article is indexed in PubMed after passing peer review Dedicated customer support at every stage For pre-submission enquiries, contact research@f1000.com