Complementary Square Wave Voltammetry and LC-MS/MS Analysis to Elucidate Induced Damaged and Mutated Mitochondrial and Nuclear DNA from in Vivo Knockdown of the BRCA1 Gene in Mouse Skeletal Muscle

Abstract

Breast cancer 1 gene (BRCA1) DNA mutations impact skeletal muscle function. Inducible skeletal muscle specific Brca1 homozygote knockout (Brca1KO^smi, KO) mice accumulate mitochondrial DNA (mtDNA) mutations resulting in loss of muscle quality.¹ Dual electrochemical-mass spectrometry analysis was utilized to rapidly assess mtDNA or nuclear DNA (nDNA) extracted directly from mouse skeletal muscle. Oxidative peak currents (I_p) from DNA immobilized layer-by-layer (LbL) were monitored using square wave voltammetry (SWV) via Ru(bpy)₃²⁺ electrocatalysis. I_p significantly decreased (p < 0.05) for KO mtDNA compared to heterozygous KO (Het) or wild type (WT), indicative of decreases in guanine content. nDNA I_p significantly increased in KO compared to WT (p < 0.05), suggesting an accumulation of damaged nDNA. Guanine or oxidatively damaged guanine content was monitored via appropriate m/z mass transitions using liquid chromatography-tandem mass spectroscopy (LC-MS/MS). Guanine in both KO mtDNA and nDNA was significantly lower, while oxidatively damaged guanine in KO nDNA was significantly elevated versus WT. These data demonstrate a loss of guanine content consistent with mtDNA mutation accumulation. Oxidative damage in KO nDNA suggests that repair processes associated with Brca1 are impacted. Overall, electrochemical and LC-MS/MS analysis can provide chemical level answers to biological model phenotypic responses as a rapid and cost-effective analysis alternative to established assays.

INTRODUCTION

Breast cancer 1 early onset gene (BRCA1) encodes for the DNA repair enzyme, breast cancer type 1 susceptibility protein (BRCA1, human). BRCA1 is highly susceptible to the development of mutations within the gene, many of which impact functionality and predispose humans to an increased risk of cancer.^2–6 More recent studies have linked BRCA1 to other biological systems acting in a DNA repair capacity,⁷ extending beyond cancer development, including cardiovascular function and brain development. Jackson et al. utilized an inducible skeletal muscle specific mouse model (Brca1KO^smi) to show that Brca1 (murine) is expressed in skeletal muscle both within the nucleus and mitochondria, resulting in phenotypic changes to muscle quality.^1,8

Phenotypic studies showed that loss of Brca1 function in skeletal muscle leads to reduced mitochondrial respiration, enlarged mitochondria, and susceptibility to osmotically-induced mitochondrial swelling all of which negatively impacted skeletal muscle quality.¹ Loss of skeletal muscle quality included the development of kyphosis and the general loss of isometric specific force production.¹ The altered function of the mitochondria under conditions of reduced Brca1 was characterized by the accumulation of mitochondrial DNA mutations.¹ qPCR was used to qualitatively determine a higher mtDNA mutation frequency in Brca1KO mice when compared to WT mice.¹

Despite clear physiological differences that developed with loss of Brca1 function, the data are only suggestive of DNA damage in the mitochondria and lack precise quantitation. Additionally, the assays do not provide any chemical basis for the potential changes in the DNA that contribute to the phenotypic differences between the KO and WT mice. Therefore, to provide a quantitative and more mechanistic insight into the type of mtDNA and nDNA variations that occur with lost Brca1 function, a dual electrochemical-tandem mass spectrometry analysis was utilized to rapidly assess structure and potential sequence changes from DNA extracted directly from the mouse model.

Electrochemical analysis techniques have been used as a rapid and cost effective means to analyze DNA and other important biochemical processes.^9–19 DNA sensors have been constructed to study a range of small oligomeric segments up to whole genomic DNA.^20–21 The immobilization of the DNA to the electrode surface dictates the analysis strategy.²² To simply assay for a general DNA changes or damage, DNA immobilized via layer by layer (LbL) on an electrode surface has been shown to be very effective.^23–24 Here, polymers of alternating charge are individually applied to an electrode surface, with the addition of each subsequent polymer effectively switching the charge on the surface.²⁵ Negatively charged DNA is immobilized onto the electrode by sandwiching between alternating layers of an inert cationic polymer.²⁵ Detection can take place by the addition of a redox active compound, usually some variant of ruthenium trisbipyridine (Ru(bpy)₃²⁺), to the analysis solution or polymer in the film architecture.^26–28 The transition metal compound is oxidized at the electrode at approximately +1 V vs. SCE, and the oxidized compound primarily electrocatalytically oxidizes guanine bases in the DNA.²⁹ Detection takes place by monitoring peak current or electrochemiluminescence (ECL).²⁹ If the DNA on the electrode is altered or damaged, the output signal will change due to the kinetic rate of the ruthenium-guanine oxidation.^29–30 In cases where DNA damage was assessed in vitro, damaged bases have been shown to be more accessible to the ruthenium compound, enhancing the kinetic rate of the oxidation of the DNA bases and resulting in higher peak currents or ECL output.^31–33 We have previously used this technique to assay DNA extracted directly from an organism. We demonstrated the detection of DNA damage onset upon exposure to increasing concentrations of Ni²⁺ in nematodes.³¹ Oxidative peak currents directly increased upon exposure to higher concentrations of Ni²⁺, which suggested an accumulation of DNA damage in the nematodes due to the metal exposure.^31–32

Liquid-chromatography-tandem mass spectroscopy (LC-MS/MS) methods have been extensively employed to validate the selectivity of the electrochemical sensor, specifically the electrochemical findings that suggested changes in DNA base structure resultant from altered DNA. Through the utilization of MS/MS techniques, where DNA is often treated in a similar manner to the electrochemical assay—i.e. an in situ DNA damage assay—before collecting structurally altered bases, these alterations can be both identified and quantified through unique product ion m/z transitions.^34–37 In addition to DNA damage assays, MS techniques have been commonly used to monitor nuclear and mitochondrial DNA for base composition analyses, most notably as an alternative to gel electrophoretic separation methods.^37–43

Here, we build upon these previous reports to couple LC-MS/MS validation with biological source DNA electrochemical assay, the bridging of which is innovative. In this manner, LC-MS/MS analysis can be used to specifically identify and quantify DNA structural changes that are detected electrochemically, thus providing a rapid and cost-effective analytical approach to provide answers to difficult genetic issues.

EXPERIMENTAL

Materials.

Tris buffer was from Fluka, poly(diallyldimethylammonium chloride) (PDDA), poly(sodium 4-styrenesulfonate) (PSS), and tris(2,2′-bipyridyl)dichlororuthenium (Ru(bpy)₃²⁺), formic acid, iron (II) chloride, hydrogen peroxide, acetonitrile and methanol (HPLC Grade) were from Sigma. Calf thymus DNA (sodium salt) was from Calbiochem. Purified 18 MΩ DI water was generated using a Siemens high-purity water system. All other chemicals for the electrochemical experiments were from Sigma and were reagent grade.

DNA Isolation and Preparation.

DNA was extracted from 30-week-old HSA-mER-Cre-mER (+) Brca1(fl+/fl+) (homozygous) or HSA-mER-Cre-mER(+) Brca1(fl+/fl-) (heterozygous) male and female mice following previously established protocols.¹ Briefly, to eliminate Brca1, the 10 week old mice were injected for 5 consecutive days with tamoxifen (2mg/day) or vehicle solution (sunflower seed oil/ethanol). Injection of the vehicle only does not induce recombination of the flox sites within Brca1, thus these mice act as the control or wild type mice. Mice were allowed to age for 20 more weeks before euthanasia and tissue collection. Mitochondria and nuclei were isolated as previously described and then DNA was extracted using Qiagen mini-spin kit (Qiagen 27104).

The following terminology will be used for simplicity: homozygous genotype mice administered tamoxifen = Brca1KO^smi or KO, heterozygous genotype mice administered tamoxifen = Het, and mice receiving vehicle only = WT.

Electrode Preparation.

Pyrolytic graphite (PG, McMaster-Carr (Elmhurst, IL) source) electrodes (2 mm dia.) were manufactured in-house and abraded on 800 grit SiC (Buehler) paper followed by rinsing and sonication in DI H₂O. After drying under an argon stream, the electrodes were then exposed to the following solutions: PDDA (30 μL droplet, 3 mg mL⁻¹ in DI H₂O with 50 mM NaCl) for 15 min followed by PSS (3 mg mL⁻¹ in DI H2O with 50 mM NaCl). These steps were repeated twice. DNA (0.065 mg mL⁻¹ in 10 mM Tris, 10 mM NaCl, pH 7.4) for 30 min was substituted for PSS in the following rounds and was applied for 30 min. The electrode was rinsed with DI water between each layer. The final film formation for the electrodes was (PDDA/PSS)₂(PDDA/DNA)₂.

Electrochemical Measurements.

All electrochemical measurements were performed using a CH Instruments (Austin, TX) Model 660A potentiostat. Modified electrodes were placed in a standard three-electrode electrochemical cell (Pt counter, saturated calomel (saturated KCl) reference in 10 mL buffer (10 mM tris, 10 mM NaCl, pH 7.4) with 50 μM Ru(bpy)₃²⁺. Square-wave voltammograms (SWV) were obtained using the following parameters: scan from 0 to +1.3 V, 15 Hz frequency, 4 mV step height, 25 mV amplitude.

LC-MS DNA Preparation.

DNA samples (0.65 mg mL⁻¹) were diluted in equal volume of 44% formic acid followed by heating for 1 hr at 120°C in 1 mL vacuum hydrolysis tubes. The solutions were then transferred to Amicon Ultra-0.5 mL 3K filter tubes and centrifuged for 40 min at 30,000 x g. The filtrate was then transferred to LC vials containing 10 μL inserts. For standard addition analysis, following hydrolysis, 1.0 nmol guanine standard was added before filtration. Following the aforementioned filtration steps, the solution was analyzed using LC-MS as described below.

In-situ LC-MS Fenton DNA Reaction Preparation.

Calf thymus DNA (3 mg mL⁻¹) was prepared in 50 mM ammonium acetate buffer (pH 6) and exposed to 0.25 M solution of H₂O₂ and 1.5 mM of FeCl₂. Incubations were carried out for 2 hrs at 37°C. The solutions were then prepared for LC-MS analysis as described above.

LC-MS/MS.

A SciEx 3200 triple quadrupole LC-MS/MS equipped with a Gemini 3μm NX-C18 110 Å LC column (50 × 2 mm) was used. Two MS detection methods were developed. First, guanine content in both mtDNA and nDNA samples was assayed, and second, mass transitions consistent with 8-oxoguanine formation were monitored. For guanine content, analysis was performed in positive ion multiple reaction monitoring (MRM) mode for two mass transitions characterizing fragments of guanine: m/z 152 → 135, conditions: 150 msec scan time, 36 V declustering potential, 8 V entrance potential, 12 V collision entrance potential, 27 V collision energy, 4 V collision exit cell potential and m/z 152 → 110, conditions: 150 msec scan time, 36 V declustering potential, 8 V entrance potential, 12 V collision entrance potential, 29 V collision energy, 4 V collision exit cell potential. LC solvents were Solvent A: 0.1% Formic Acid, 95:5 DI H₂O and Acetonitrile and Solvent B: Methanol. Gradient elution took place via 100% A 0–3 mins, ramp to 75% A mins 3–5. The total analysis time was 5 mins. The flow rate was 0.200 mL min⁻¹ and injection volume was 1.00 μL.

For 8-oxo-guanine content, analysis was performed in positive ion MRM mode for two characterizing fragments of 8-oxo-guanine: m/z 158 → 85, conditions: 150 msec scan time, 16 V declustering potential, 6.5 V entrance potential, 12 V collision entrance potential, 17 V collision energy, 4 V collision exit cell potential and m/z 158 → 126, conditions: 150 msec scan time, 16 V declustering potential, 6.5 V entrance potential, 12 V collision entrance potential, 11 V collision energy, 4 V collision exit cell potential. LC solvents were A: 0.1% Formic Acid, 95:5 DI H₂O and Acetonitrile; B: Methanol; C: Acetonitrile. Gradient elution took place via 45:50:5 A:B:C for 0–1 mins, decreasing A and increasing C by 5%, respectively, for 1–3 mins, holding 35:50:15 A:B:C for 3–4 mins, ramping back to 45:50:5 at 4–6 mins by increasing A and decreasing C by 5%, respectively. Total analysis time was 6 mins. The flow rate was 0.100 mL min⁻¹ and injection volume was 1.00 μL.

Data Analysis.

Statistical analysis was performed using Excel for MAC 2019, version 16.24 and SCIEX MultiQuant Software, version 3.0. Raw MS and SWV were analyzed using OriginPro 9 software. For SWV plots, background-subtracting was performed using the same SWV electrochemical response for a (PDDA/PSS)₂ electrode containing no DNA. Figures were modified using Adobe Illustrator CS6.

RESULTS

Electrochemical Analysis.

Our goal was to monitor DNA that had been extracted from skeletal muscle of wildtype, heterozygous (Het) or Brca1KO^smi mice by applying similar approaches based on our previous report showing LbL methods ^{27, 44–46} can be effective in assessing biological DNA samples.⁹ It was previously shown phenotypically, as well as genotypically using qPCR, that the mitochondria of the KO mice were altered. Based on this evidence, similar DNA-modified electrodes were constructed using mtDNA or nDNA extracted from the peripheral skeletal muscle of WT or Brca1KO^smi mice. Briefly, DNA can be concentrated on an electrode surface via electrostatic forces by switching the effective charge of the electrode surface by immobilization of a polymer of opposite charge. Here, PDDA was used in between layers to capture and preconcentrate DNA at the electrode surface. Guanine is the most easily oxidizable and thus the most commonly studied base in these films.^11,13 The output current generated by guanine oxidation is enhanced by redox-active ruthenium trisbipyridine (Ru(bpy)₃²⁺).^27,31 Ru(bpy)₃ ²⁺ is oxidized at the electrode followed by regeneration through the oxidation of accessible guanines in the DNA according to the following:¹¹

$$\text{Ru}{(\text{bpy})}_3{}^{2+}\to \text{Ru}{(\text{bpy})}_3{}^{3+}(\text{at}\text{electrode})$$ (1)

$$\text{Ru}{(\text{bpy})}_3{}^{3+}+\text{G}(\text{DNA})\to {\text{G}}^{\cdot +}(\text{DNA})+\text{Ru}{(\text{bpy})}_3{}^{2+}$$ (2)

The rate at which the electrocatalytic reaction occurs is important because it can reveal critical differences in DNA sequences. Typically, the electrochemical response monitored in situations where the possibility of DNA damage is suspected results in an increase in signal between the test case vs. the control. More facile guanine access via mispairing or DNA damage leads to closer ruthenium approach, faster ruthenium 2+ regeneration via guanine oxidation, and higher electrochemical currents.^47–49 The main, but not trivial, difference here compared to previous reports is the use of mtDNA in the LbL films.

Fig. 1a shows representative background-subtracted mtDNA SWV for KO, Het, and age-matched WT samples. The figure shows that the oxidative peak current decreased stepwise as the Brca1 genetic knock-out was made more severe from WT to one allele and then two allele deletion (i.e. from WT to Het to KO). Decreases in peak current at approximately +1.05 V versus SCE for KO indicate a decrease in the kinetic rate of Ru(bpy)₃²⁺ oxidation at the electrode, suggesting that the KO samples do not exhibit elevated amounts of mtDNA damage compared to the WT samples. The data suggest a significant change in the mtDNA from WT to KO samples, consistent with a guanine content difference. Fig. 1c shows the average peak currents for each DNA type. A statistically significant (p < 0.05) average current decrease of 19.8% was seen when comparing peak currents from KO mtDNA versus WT samples, whereas Het samples exhibited a slightly less severe current decrease versus WT. Note that the mouse samples used in these analyses were both male and female; however, no significant trends in either mtDNA or nDNA were determined in relation to gender differences. Larger peak current variations in WT and Het samples are due to the biological nature of the DNA sample – i.e. DNA from individual mice - not due to the LbL electrode modification or electrochemical application. They are also mirrored in the LC-MS data (vide infra). We were able to obtain reproducible intra-sample peak currents using different electrodes exhibiting 6% RSD (n = 3) on separate, successive analyses. A representative sample of raw SWV summarizing these data is presented in Figure S1.

Figure 1.

Background subtracted SWV plots showing typical voltammetric response between WT (black), Het (blue) and KO (red) for a) mtDNA and b) nDNA. c) Average peak current responses for each denoted DNA type. Error bars denote S.D. for mtDNA WT (n = 17), Het (n = 8), KO (n = 9) and nDNA WT (n = 5) and KO (n = 5). Asterisks denote statistically significant differences (p < 0.05).

Fig. 1b shows a background-subtracted nDNA SWV comparison with the Brca1KO^smi sample shown in red and the age matched WT in black. In contrast to the decreasing currents when comparing KO vs. WT mtDNA, the nDNA comparison shows that there is an increase in peak current at approximately +1.05 V for KO vs. WT, indicating an increase in the kinetic rate of Ru(bpy)₃²⁺ oxidation at the electrode when KO is monitored, consistent with the presence of damaged guanine bases. A statistically significant (p < 0.02) average peak increase of 14.7% in KO vs. WT nDNA samples was seen (Fig. 1c). There were not enough collected Het nDNA samples for those analyses to be included here.

In order to validate the electrochemical findings, mass spectrometry analyses were performed (vide infra). Those studies included an acid hydrolysis preparation step. To determine the impacts of acid hydrolysis conditions on the amount of guanine detected, we performed an additional SWV analysis to show that the electrochemical assay and the MS data mirror each other. Fig. 2 shows these data where ctDNA was used as a model to study the hydrolysis.

Figure 2:

SWV comparison showing ctDNA electrochemical response before (green) and after exposure to Fenton (red) or hydrolysis (black) conditions. Blue shows response of collected filtrate after DNA hydrolysis.

This figure highlights a number of important aspects. First, consider the SWV comparison where equal concentrations of ctDNA were either immobilized onto an electrode and oxidatively analyzed (green trace), as previously outlined, or exposed to hydrolysis conditions prior to collecting the DNA, reconstituting to the original concentration (as confirmed by UV-Vis) and immobilized onto the electrode (black trace) for analysis. The figure shows that the electrochemical current response is effectively muted following hydrolysis, consistent with the complete removal of electrochemically accessible guanine in the films. When exposing the filtrate to the electrode, the electrochemical response was recovered (blue trace), showing that the guanine was captured via hydrolysis and filtration. This suggests that our hydrolysis conditions are removing many of the electrochemically accessible guanine bases to capture and detect using LC-MS/MS.

Additionally, the figure also shows the response of Fenton reaction-oxidized DNA that was captured and then put on the electrode (red trace). Both this analysis and the isolated base analysis exhibit increased oxidative peak currents, consistent with more easily accessible guanine bases. For the oxidatively damaged sample, this is consistent with the accumulation of DNA damage via 8-oxo-guanine (8oxoG) formation, whereas the free bases are easily accessible by their nature of being removed from the polymer through hydrolysis. We return to these data in the context of their overall importance with MS/MS analysis below.

Guanine Content Mass Spectrometry Analysis:

In order to validate the electrochemical results and provide more details on the chemical basis of these changes, LC-MS/MS analyses were performed on Brca1KO, Het, and WT DNA samples.

Here, we utilized methodology coupling liquid chromatography separation to electrospray ionization-triple quadrupole tandem MS that is well established for analysis of DNA bases.^37–39 First, because of its role in the electrochemical assay, guanine content was determined for both mtDNA and nDNA. Guanine standard was used to calibrate the instrument response, first by identifying the correct parent ion via Q1 MS scanning in positive mode, followed by product ion scanning while optimizing fragmentation parameters to obtain mass transitions of m/z 152→135 for quantitation and 152→110 for qualitative confirmation. This spectrum is shown in Supplemental Information (Fig. S2). The transitions at m/z 135 and 110 are consistent with previous literature describing MS analysis of guanine.^37–39 The analytical response showing the 152→135 transition peak area vs. guanine solution is shown in Fig. S3.

DNA from mouse samples was hydrolyzed in formic acid to release purinic bases before filtering for LC-MS/MS detection.⁵⁰ This sample preparation technique was effective in the removal of accessible guanine bases from the larger DNA polymer as shown and described previously in Fig. 2. Figure 3a shows a representative chromatogram normalized for input mtDNA concentration comparing the detected m/z 152→135 guanine product ion transition for KO (red) vs. Het (blue) and WT (black). The figure shows that the guanine response for KO samples was substantially less than that for WT and Het. Fig. 3b shows the average guanine detected from each sample, highlighting a 33.3% average decrease in detected guanine content for KO mtDNA samples vs. WT. The variation in the WT and Het samples mirrored that seen in the electrochemical data, reflecting the biological sample nature. The difference in guanine content was statistically significant between the KO and WT samples (p < 0.0015), but the difference between Het and WT samples was not quite statistically significant. Similar analysis was performed using nDNA, and the example chromatogram comparing KO (red) vs. WT (black) as well as the average responses are shown in Figs. S4 and 4b, respectively. The KO nDNA also showed a statistically significant decrease (p < 0.02) in guanine content compared to WT samples. These data are summarized for simple comparison in Table 1.

Figure 3.

a) LC-MS/MS chromatograms of WT, Het and Brca1KO mtDNA samples monitoring m/z 152→135 transition. Normalized for input mtDNA concentration. b) Average normalized guanine content between WT (black), Het (blue) and KO (red) for mtDNA and nDNA. Error bars represent S.D. for n = 16 (WT), 13 (Het), and 12 (KO) mtDNA and n = 14 (WT) and 17 (KO) nDNA. Asterisks denote statistically significant differences (p < 0.002 mDNA and p < 0.02 nDNA) between the denoted groups.

Figure 4.

Normalized LC-MS/MS chromatograms showing response monitoring m/z 158→85 from a) mtDNA and b) nDNA WT (black) or Brca1KO (red) samples. Intensity was normalized to starting DNA concentration. c) Average normalized 8oxoG peak areas between WT (black), Het (blue), and KO (red) from different denoted DNA types. Error bars represent S.D. from n = 6 (WT), 4 (Het), and 3 (KO) mtDNA and n = 14 (WT) and 19 (KO) nDNA. Asterisks denote statistically significant differences between the groups (p < 0.02 for mtDNA and p < 0.005 for nDNA).

Table 1.

Summary of mass spectrometry data for mtDNA and nDNA.

	Guanine					8oxoG
		Avg. Gua.a	SDb	n	% Changec	Avg. Peak Aread	SD	n	% Changea,e
mtDNA	WT	2.48	0.781	16	--	1610	541	7	--
	Het	2.33	0.985	13	−6.2	1190	422	4	−26.2
	KO	1.66	0.0191	12	−33.5	950	188	3	−40.9
nDNA	WT	4.17	1.25	16	--	1770	137	8	--
	KO	3.15	1.13	17	−24.5	1900	129	10	6.8

^{a -}x 10⁻³ ng guanine (ng DNA)⁻¹

^{b -}x 10⁻³

^{c -}percent difference comparing treatment to WT

^{d -}normalized to starting input DNA concentration

^{e -}corrected for signal acquired due to hydrolysis treatment

To assess for any potential ion suppression or recovery issues, standard addition procedures were followed as an alternative to using isotopically labeled standards.^51–53 Similar procedures have been utilized for organic biomolecule analysis.⁵¹ Following hydrolysis, we added a known amount of standard guanine to the solution, filtered, and analyzed via our LC-MS parameters. Based on our previous guanine standard analysis (Figure S3) and hydrolysis of mtDNA or nDNA and the amount of guanine we expect to detect (vide infra), it was determined that our guanine detection recovery was 96.7±0.4% (n = 3). A summary of these calculations is outlined in SI. Based on this recovery value, ion suppression does not significantly inhibit the analyses. This is consistent with what has been reported for MS analysis of biomolecules originating from less complex matrices as compared to those originating from matrix-heavy samples such as blood or plasma.⁵²

8-oxo-Guanine Content Mass Spectrometry Analysis.

Beyond strictly assaying for hydrolyzed guanine content, we also monitored the DNA samples for m/z that were consistent with DNA damage to determine if its presence in the samples could impact the electrochemical outputs. The overall decrease in guanine content shown in Fig. 3 supports the electrochemical assay findings for mtDNA, where Brca1KO^smi samples exhibited a peak current decrease compared to WT samples upon oxidation.

However, to explain the electrochemical current increase for nDNA Brca1KO samples (Fig. 1b) coupled with the decrease in guanine content (Fig. 3), we analyzed the filtered DNA for m/z consistent with oxidative damage, predominately 8oxoG. To optimize MS conditions for 8oxoG analysis, we initially exposed known concentrations of guanine and calf thymus DNA (ctDNA) to Fenton conditions using Fe³⁺ and hydrogen peroxide.^54–56

Standard guanine and test DNA were allowed to react for set periods of time before undergoing similar hydrolysis and filtration steps as outlined above for DNA when monitoring guanine content. It has been previously shown that m/z consistent with the oxygen atom addition to guanine oscillate over reaction time;^57,58 however, the 8oxoG m/z 158 rearrangement product provides a more stable ion to monitor.⁵⁹ This guanidinohydantion rearrangement, shown in Fig. S5, has been shown to fragment into reliable quantifying and qualifying m/z in previous MRM analyses.⁵⁹

Under our reaction conditions, m/z 158 was detected in increasing amounts as a function of reaction time or starting concentrations of either guanine or ctDNA (Figs S6–7). The product ion spectrum for m/z 158 is shown in Fig. S8. Upon product ion scanning, m/z 158→85 and 158→126 transitions were selected for quantitation and qualitative confirmation, respectively. Overall, the detection of m/z 158 and the representative product ions are consistent with 8oxoG content in the samples and allows for its more facile detection and monitoring in biological samples.⁵⁹

Figure 4 shows representative LC-MS/MS chromatograms monitoring the m/z 158→85 product ion response comparison between KO (red) and WT (black) mtDNA (Figure 4a) and nDNA (Figure 4b) samples, respectively. The figure shows that the 8oxoG response for KO nDNA samples is elevated as compared to WT in nDNA while in mtDNA, KO samples showed a depleted amount of 8oxoG content compared to WT. The plot in Fig. 4c shows the average normalized responses for each of these sample populations. Table 1 summarizes the 8oxoG data for each mouse sample population. The percent changes for the KO samples as compared to the WT samples are also summarized in the table.

The data demonstrate that the amount of 8oxoG product detected in the mtDNA KO samples was significantly less (p < 0.05) by approximately 40%. This result is somewhat surprising in that the deletion of Brca1 was thought to result in the accumulation of damage to mtDNA based on the impacted DNA repair processes, but this is likely the result of mutations leading to less detectable guanine present.⁷ The data demonstrate that higher amounts of DNA damage occur in the KO nDNA samples, as the amount of 8oxoG product detected in the nDNA KO samples was significantly higher by approximately 7% (p < 0.05). These data are consistent with the electrochemical nDNA analyses in that higher amounts of oxidative damage result in higher electrochemical peak currents due to enhanced ruthenium redox kinetics in the presence of the damaged DNA. All of these data are consistent with the detection of guanine-involved variations in both mtDNA and nDNA; however, the data do reveal that the mtDNA process is quite intricate. Overall, while not initially obvious, these data are consistent with the electrochemical and guanine quantification LC-MS/MS data previously presented, which is discussed further below.

Mutation Analysis.

Of particular interest is the relationship between the percent guanine content change determined and the absolute amount of guanine originating from each sample. MS analysis showed that the concentration of guanine detected was much lower than what would be expected if every guanine base was captured from the DNA sample. We utilized what is known about mtDNA and ctDNA to determine the percentage of guanine from an original sample that is detected. This allowed us to form a baseline to gauge the total concentration of nucleobases initially present in the DNA based on established quantification formulae and the known G-C content percentage in mtDNA or ctDNA.⁶⁰ This mathematical overview is summarized in SI. Analyzing WT mtDNA samples provided an average detected guanine percentage of 4.86(±1.53)%. Similar values were also calculated after hydrolyzing, filtering, and analyzing ctDNA using similar concentrations as the mtDNA samples (data not shown).

Although ~5% represents a small yield of guanine collected and detected, it is congruent with the known sluggish and inefficient reaction yields that have been reported when performing base analysis using non-specific nucleotide acid hydrolysis.⁵⁰ Additionally, this analysis shows that the oxidative peak currents seen in the electrochemical assay originate from a significant minority of the guanines in the original sample (Fig. 2). This is consistent with large DNA polymers adopting a protective, yet similar morphology on the electrode that allows the analysis of roughly the same guanine population depending on technique. Future studies are needed to fully discern this possibility.

More specific to the Brca1 application, however, is that this value gives us the ability to use the detected amounts of guanine between Brca1KO^smi and WT samples to reach a mutation frequency. We would expect to capture a statistically consistent population of guanines from either sample if both were exposed to identical conditions. Based on the detected guanine amounts, we calculated a mutation frequency per base pair of 0.065 in Brca1KO^smi mtDNA. This is an elevated value, and it suggests a mutation burden of ~1000 within the entire Brca1KO^smi mtDNA, but it is consistent with a recent model showing mtDNA mutation frequencies of 6×10⁻³ – 0.1 over the course of 70 years.⁶¹ Phenotypic studies showed the Brca1 knockout results in muscular tissue that is consistent with this type of aging.¹ Additionally, this elevated frequency is consistent with elevated mutation rates reported in DNA polymerase γ (POLG) mouse models carrying specific mutations in the proofreading region, that were shown to exhibit increased levels of mtDNA mutations determined using qPCR⁶² and next-generation sequencing (NGS).⁶³ A similar calculation in Brca1KO^smi nDNA leads to a mutation frequency estimate of 7×10⁻⁴ – 1.0×10⁻³.

DISCUSSION

Both mtDNA and nDNA were extracted from tamoxifen or vehicle injected HSA-mER-Cre-mER (+)/Brca1(fl⁺/fl⁻) or HSA-mER-Cre-mER (+)/Brca1(fl⁺/fl⁺) mice.¹ This mouse model was allowed to age for a total of 20 weeks prior to DNA extraction, to investigate the impacts of Brca1 deletion in skeletal muscle of adult mice. Initially, phenotypic studies, including an electron microcopy analysis showing the presence of swollen mitochondria in homozygous KO mice, alluded to an increase in mutations in Brca1KO mtDNA samples.¹ However, quantitative mtDNA mutation analysis measures are both time and cost prohibitive mainly due to complexities associated with mtDNA. For instance, sequencing mtDNA is difficult due to its polymorphic nature, making the use of consistent primers troublesome.⁶⁴ Additionally, mtDNA sequencing may introduce mutations, and while each mtDNA molecule can be amplified individually, these assays typically lack sensitivity to provide a consistent mutation frequency analysis.⁶⁵ A solution is to employ the random mutation capture (RMC) PCR assay that exploits mutations in Taq1 digestion sites.⁶⁵ Mutations in any of the four bp sequence sites results in an intact segment of DNA that can be subsequently amplified and detected.⁶⁴ The number of amplicons per total bp monitored provides the means to calculate mutation frequency vs. WT.⁶⁵ Through employing this assay, DNA mutations were shown to be more relatively prevalent in the mtDNA of the Brca1KO^smi mice.¹

Some drawbacks of the RMC assay include the possibility of WT DNA amplicons providing a false positive owing to digestion-precluded mtDNA structural motifs, and despite improvements compared to other sequencing assays, it is still cost and time intensive compared to other analytical methods, mainly owing to the costs associated with the RT-PCR instrumentation and significant costs associated with the high-fidelity DNA polymerase and fluorescent agents. The assay itself requires an overnight digestion and multi-hour instrument runs after that initial preparatory step. Finally, while the assay is responsive to mutations in any of the four bases, it does not provide information regarding which base is mutated, nor does it offer reasons to why the mutation occurs. Improvements to these drawbacks are presented here by employing the electrochemical and LC-MS/MS approaches to study the mtDNA from Brca1KO^smi mice.

The electrochemical assay results show how this approach can be used as a rapid screen for the possibility of DNA variations in a biological model. LbL electrode preparation and electrochemical detection are both rapid and inexpensive compared to previously discussed genotypic assays. The electrochemical results show a stepwise decrease in oxidative currents upon analyzing WT, Het, and KO mtDNA. Since the electrochemical assay is extremely sensitive to the amount of guanine in the DNA films, these decreases in signal are indicative of lower amounts of guanine across this mouse model spectrum. This signal response was validated by hydrolyzing the DNA samples and analyzing guanine content with LC-MS/MS, which provides the chemical structure information to complement the initial qPCR mutation analysis.¹ MS data showed that the KO mtDNA samples exhibit lower amounts of guanine compared to WT or Het samples. The guanine decrease is consistent with an increase in mtDNA mutations as the Brca1 gene was sequentially silenced in the mice.

LC-MS/MS analysis also showed a decrease in guanine content in KO vs. WT nDNA, but electrochemical nDNA analysis showed an increase in oxidative current for KO vs. WT mice, opposite of the mtDNA responses. Increases in current signify the presence of damaged DNA vs. control.^47–49 Using LC-MS/MS, we detected an increased concentration of m/z consistent with 8oxoG formation in KO nDNA as compared to WT. The elevated amounts of chemically altered guanine in KO nDNA likely contributes to the lower amount of native guanine detected in KO samples as well. Therefore, taken as a whole, the LC-MS/MS validated electrochemical results are consistent with KO mice accumulating DNA or guanine damage in the nucleus, and increased mutations at guanine sites in mtDNA due to Brca1 knockout.

Overall, these findings are congruent with the estimation that 8oxoG develops in mtDNA ten times more frequently in comparison to nDNA, with approximately 50% of damages resulting in a G to T transversion mutation.⁶⁶ The absence of functional Brca1 also incorporates additional sources of DNA variances, albeit differing responses in nDNA and mtDNA. In both nDNA and mtDNA, Brca1 regulates the base excision repair (BER) pathway, which acts as a repair mechanism for oxidized DNA by increasing activity of BER enzymes, including 8oxoG DNA glycosylase (OGG1).⁶⁷ However, mtDNA has a specific mechanism to maintain integrity of its circular structure through the introduction of strand breaks that result in linear mtDNA fragments.^68,69 Although these fragments have been shown to be a precursor to expulsion from the organelle, they are much more susceptible to mutagenic lesions in comparison to intact circular mtDNA.⁶⁸ Thus, through the combination of increased levels of ROS exposure in comparison to nDNA and loss of Brca1, mtDNA may undergo additional damage more rapidly that can lead to an increased presence of transversion mutations, which results in the subsequent detection of lower amounts of guanine content. Indeed, we detected a mutation frequency that was approximately 70 times higher in KO mtDNA compared to the nDNA, which is consistent with what is known about the mutation rates between the two.^70,71 Specifically, the point mutation, G→A, hypothesized through these analyses has been shown to be 70 times more prevalent in mtDNA than in nDNA in previous pyrosequencing studies in Drosophilia.⁷¹ The mutation frequency we calculated based on the LC-MS/MS results and compared to the known mtDNA base content is on the order of that detected in the POLG mutator mouse, a model known to accumulate mtDNA mutations due to the expression of deficient polymerase γ protein.⁶³

The impact of Brca1 on the skeletal system is still a burgeoning and intense area of research. Brca1 has been shown to localize in both the nucleus and mitochondria, and its loss drastically impacts the musculature of an organism. Here, we have shown how rapid and inexpensive electrochemical and more detailed LC-MS/MS analytical tools can provide a complementary approach to give detailed answers regarding mtDNA mutations and shed light on the role Brca1 plays in DNA damage and repair in both organelle locations. While we focused primarily on guanine, mainly due to the higher frequency of its damage compared to other bases and its role in the electrochemical DNA assay, future studies will focus on the analysis and detection of mutations that may occur at additional base sites. Based on our previous research showing DNA damage detection in nematodes and the work presented here analyzing murine mtDNA and nDNA, it is conceivable that this electrochemical/LC-MS complementary approach can be useful in providing genetic insight from a wide range of biological models as an alternative to DNA sequencing.