DNA-modified ITO surfaces for affinity MALDI-MS

Abstract

Indium Tin Oxide (ITO) coated surfaces were compared with fused silica and aluminium probes for conventional MALDI-MS of proteins, followed by experiments to determine if the advantages of ITO over the poorly conducting fused silica surface would carryover to affinity MALDI-MS, in which the probe surface is modified with DNA for protein capture and detection. Improved precision of m/z values and increased detectability by 1 to 2 orders of magnitude were observed for capture and detection of insulin and insulin-like growth factor 2 at surfaces modified with sequences from the insulin-linked polymorphic region (ILPR) of the human insulin gene that have previously been shown to bind with high affinity to these proteins. Results also indicate capture and detection of the insulin β-chain from DTT-treated human serum at the DNA-modified ITO surfaces. The improved performance is accompanied by the same ease of covalent DNA attachment and indefinite reusability of the DNA-modified surfaces previously demonstrated for fused silica probes. The use of ITO-coated surfaces is an important advance toward the routine use of DNA-modified surfaces in affinity MALDI-MS for rapid screening of DNA–protein interactions for applications such as biomarker discovery and detection of low abundance proteins in complex biological samples.

---

Introduction

The insulin-linked polymorphic region (ILPR) of the human insulin gene contains tandem repeats of similar G-rich sequences, some of which form intramolecular G-quadruplex structures in vitro.^1–5 Previous work has established affinity binding of insulin and insulin-like growth factor 2 (IGF-2) to intramolecular G-quadruplexes formed by ILPR variants a and h but not to variant i, which does not form an intramolecular G-quadruplex.^4,6 The results of prior work on the insulin–ILPR system^4–8 provide a basis for design of DNA binding ligands for insulin and IGF-2 and support a new approach to discovery of DNA affinity binding ligands based on genome-inspired sequences rather than the traditional combinatorial selection route to aptamer discovery. A second outcome of the work was the demonstration of DNA binding ligands in affinity MALDI-MS for capture and detection of low abundance proteins and peptides from human biological samples.^6,9,10

In previous work, DNA oligonucleotides were immobilized at fused silica surfaces rather than metal surfaces that are typically used in MALDI-MS. Fused silica has poor conductivity, which may cause charge stacking during laser bombardment that reduces the efficiency of the MALDI process and therefore decreases detectability of analytes. Furthermore, the silica surface is hydrophilic due to the presence of silanol groups, which can cause spreading of the sample droplet on the surface. These factors can affect ionization of analytes and result in poor spectral resolution and inaccurate m/z values.^11,12 The surface conductivity of metal probe surfaces, on the other hand, is sufficiently high to permit dissipation of surface charges after laser bombardment and the surface hydrophobicity promotes homogenous mixtures of matrix and analytes for better sample preparation.^13–15

Here we describe the investigation of Indium Tin Oxide (ITO)-coated glass surfaces as an alternative to fused silica surfaces as the substrate for immobilization of ILPR oligonucleotides for affinity MALDI-MS capture and detection of insulin and IGF-2. ITO surfaces are well suited to this purpose because of their high conductivity, hydrophobicity and ease of covalent DNA attachment. Because of these properties, ITO films have become important materials for use as conductive substrates for DNA sensors^16–18 and as probe surfaces in atmospheric pressure and imaging MALDI-MS.^19,20

In this work, we compare ITO-coated glass slides with aluminium MALDI probes and fused silica plates for conventional MALDI-MS of insulin and IGF-2. We then demonstrate affinity MALDI-MS capture and detection of human insulin and IGF-2 on ITO plates modified with the ILPR variants a and h and compare the results with previous work that ILPR-modified fused silica plates.⁶ ILPR variant i and bare ITO surfaces served as controls since they do not exhibit affinity toward the proteins. Finally, the ILPR variant a and variant h modified ITO surfaces were employed in affinity MALDI-MS of a commercial human serum sample.

Materials and methods

Materials

Oligonucleotides that were either thiol-modified or amine-modified at the 5′-end were synthesized by Eurogentec (San Diego, CA). The sequences of the oligonucleotides included 2-repeats of ILRP variant a (ILPRa) (5′-(ACAGGGGTGTGGGG)₂-3′), ILPR variant h (ILPRh) (5′-(ACAGGGGTGTGGGC)₂-3′) and ILPR variant i (ILPRi) (5′-(ACAGGGTCCTGGGG)₂-3′). Recombinant human proteins including insulin and insulin-like growth factor 2 (IGF-2) and pooled human serum were from Sigma (St Louis, MO). Other reagents, including phosphate buffered saline (PBS) buffer (10 mM potassium phosphate buffer, 150 mM potassium chloride, pH 7.32), sinapinic acid (SA), 1,2-dichloroethane, trichloroethylene, acetone, isopropyl alcohol, l,3-aminopropyltrimethoxysilane (APTES), dimethylformamide (DMF), pyridine, 1,4-phenylene diisothiocyanate (PDITC), tris(2-carboxyethyl)phosphine (TCEP), sulfo-succinimidyl-4-(N-maleimidomethyl) cyclohexane-1-carboxylate (sulfo-SMCC), α-cyano-4-hydroxycinnamic acid (CHCA), trifluoroacetic acid (TFA), dithiolthreitol (DTT), and reagents for preparation of the phosphate buffer (10 mM potassium phosphate, 150 mM NaCl, pH 7.4) were also from Sigma. Distilled deionized water was used in all procedures and solution preparations.

Preparation of oligonucleotide-coated MALDI probe surfaces

Prior to surface attachment the oligonucleotides were heated to melt any intermolecular and intramolecular structures and then cooled to room temperature. Circular dichroism spectra were then taken of the solutions to confirm the presence of intramolecular G-quadruplexes by ILPRa and ILPRh as evidence by peaks at 295 nm and the absence of such intramolecular structures in ILPRi prior to immobilization. The ILPRi serves as a control since it does not exhibit affinity binding toward insulin or IGF-2.^6,22

Fused silica chips (20 mm × 20 mm × 0.75 mm) were made by Valley Design (Westford, MA). 5′-Thiol modified DNA oligonucleotides (ILPR variants or scrambled control) were covalently attached to the fused silica chips as previously described.^6,9,10 In brief, the chip surface was first cleaned and activated by rinsing with methanol, water, and sodium hydroxide. The activated chip was then immersed in a 10% solution of 3-APTES at 100 °C for 4 h, followed by addition of the heterobifunctional linker sulfo-SMCC to the 3-APTES-coated fused silica surface to create spots that were approximately 1 mm in diameter.

The 5′-thiol modified oligonucleotides were treated with TCEP to give a free sulfhydryl group that was reacted with the linker at the chip surface. Finally, the DNA-modified plates were rinsed with buffer to remove excess reagents including free oligonucleotide, dried with ultra-high purity nitrogen, and stored at room temperature. As an additional precaution to ensure the removal of any non-covalently bound oligonucleotides that could otherwise form intermolecular G-quadruplex structures with the bound oligonucleotides, surfaces were treated with the acidic MALDI matrix to unfold any secondary oligonucleotide structures and rinsed with 50% acetonitrile in water to remove unbound material prior to first use.

ITO-coated glass slides (8–12 Ω per square surface resistivity, Aldrich) were cut into plates with dimensions of 20 mm × 20 mm × 1.1 mm. Prior to surface chemical modification, the chips were treated in an oxygen plasma (Tegal Plasma Asher) for 30 min at 50 W in order to create more reactive hydroxyl groups on the surface for later functional group attachment. The 5′-amine-modified DNA oligonucleotides were covalently attached to the ITO coated glass slides through a covalent linker using a method from the literature.²¹ First, the ITO coated glass chip was pre-treated by solvent washing in trichloroethylene and sonicated for 10 min in order to clean the surface and prepare the samples for silanisation. The chip was then placed in acetone at 50 °C for 10 min and then sonicated in isopropyl alcohol for 10 min. After drying with nitrogen, the chip surface was silanised in a MeOH : H₂O (19 : 1) solution containing 30% APTES for 30 min at room temperature. After rinsing with methanol and then water, the chip was cured at 120 °C for 15 min in a fan-operated oven. The surface was then activated for DNA attachment by immersion in a DMF solution containing 10% pyridine and 10 mM PDITC for 2 h. The chip was washed sequentially with DMF and 1,2-dichloroethane and dried with nitrogen. DNA immobilization was achieved by spotting 1 µL of the amine-modified oligonucleotide (500 µM) onto pre-defined circles with dimension of 1 mm in diameter on the ITO substrate surface followed by overnight incubation at 37 °C in a humid chamber. The chip was finally washed with methanol and water and dried in nitrogen. A 2 µL aliquot of PBS buffer was then spotted onto each DNA spot to maintain the G-quadruplex ILPR structures and stored in the refrigerator for later use. A diagram of the DNA-modified ITO surface is shown in Fig. 1.

Fig. 1 Schematic of DNA-modified ITO probe surface.

Affinity MALDI-MS of insulin and IGF-2

For affinity MALDI-MS protein capture experiments, a 1 µL aliquot of protein sample (insulin or IGF-2) in phosphate buffer was incubated on each oligonucleotide-coated spot at room temperature for 30 min. The surface was then rinsed with deionized water for 30 s to remove any weakly bound or unbound proteins as well as buffer and salts, and dried under nitrogen gas. This incubation–rinse cycle was repeated two more times, each with a new aliquot of sample, to concentrate captured protein at the surface. A 1–1.5 µL aliquot of MALDI matrix (a 1 : 1 mixture of water and saturated SA solution containing 0.3% TFA) was then applied to the spot and allowed to crystallize. The chip was then mounted on a custom Al MALDI target designed to accommodate the chip and each spot was analyzed by MALDI-time-of-flight-mass spectrometry (MALDI-TOF-MS) using a Bruker AutoFlex II instrument (Billerica, MA). Each mass spectrum is the sum of 1000 laser shots taken at different points across the sample spot. After use the spots were rinsed with 50% acetonitrile in water to remove the MALDI matrix, peptides and any concomitants. PBS buffer was applied to the surface to reconstitute it prior to reuse. Surfaces were periodically tested for protein carryover, contamination and degradation of the oligonucleotide-modified surface by MALDI-MS in the absence of applied protein sample. In no case was any carryover, contamination or degradation indicated.

Affinity MALDI-MS of human serum and in situ digestion of captured proteins

Insulin and IGF-2 (1 µM in PBS buffer) and human serum that was diluted 10⁴-fold in PBS buffer were treated in DTT (50 mM) and boiled at 95 °C for 5 min. For each sample, 2 µL was incubated on an oligonucleotide-modified spot at room temperature for 30 min. Chips were then rinsed with deionized water for 30 s to remove any weakly bound or unbound proteins and dried under nitrogen gas. This incubation–rinse cycle was repeated two more times, each with a new 2 µL aliquot of sample, to concentrate captured proteins at the surface. For in situ digestion, chymase solution (1 µL of 0.37 µM in PBS buffer) was placed on the spots following the three incubation/rinse cycles and the plate was incubated at 37 °C for 15 min.²² After deposition of MALDI matrix (saturated CHCA solution containing 50% of ACN and 0.3% trifluoroacetic acid), the ITO chip was placed on the customized MALDI target and loaded into the MALDI-MS system for MS analysis. The CHCA matrix was found to be superior to SA matrix for the serum studies. After use, the spots were rinsed with 50% acetonitrile in water to remove the MALDI matrix, peptides and any concomitants. PBS buffer was applied to the surface to reconstitute it prior to reuse.

In silico digestion of the insulin β-chain was performed using online software “Protein Prospector” from the website of the University of California, San Francisco (http://prospector.ucsf.edu/), with the following search parameters: Database: user protein; Digest: chymotrypsin (chymase digestion is not available in this software but chymotrypsin cleaves at the same sites and is a valid substitution); Variable Modification: DTT treated; Min peptide length: 3; Max missed cleavages: 5; Instrument: MALDI-TOF.

Results

MALDI-MS analysis of protein samples on three surfaces

The m/z values (raw, uncalibrated), peak intensities, signal-to-noise (S/N) and peak areas for recombinant human insulin (172 nM, M_w 5808) and IGF-2 (100 nM, M_w 7505) were compared for conventional MALDI-MS on fused silica, ITO-coated and Al probe surfaces. Results are shown in Table 1 for insulin. The m/z values show similar precision, while the peak intensity, S/N and peak area are similar for the ITO-coated and Al plates but greatly decreased for the fused silica plate. Results for IGF-2 (Table 2) are of a much lower quality compared to insulin due to the relatively weak signals that resulted in large run-to-run variability, but the overall trends in peak intensity, S/N and peak area are similar to those for insulin. One exception is the large imprecision of the m/z values on fused silica compared to ITO and Al, likely due to the relatively poor detectability of IGF-2 that increases the impact of surface conductivity on the experimental m/z. These results confirm that the overall performance for MALDI-MS of insulin and IGF-2 is greatly improved by using ITO-coated glass slides instead of fused silica, which we had used in affinity MALDI-MS studies up to this point.

Table 1 MALDI-MS analysis of 1 µL of 172 nM (172 fmol) insulin on different surfaces^a

	m/z ± SD	Intensity ± SD	S/N ± SD	Peak area ± SD
a Results are mean ± one standard deviation for 6 replicate analyses.
Fused silica	5914 ± 3	(7 ± 4) × 10^3	(8 ± 4) × 10^2	(5 ± 2) × 10^5
ITO	5817 ± 3	(3 ± 1) × 10^4	(5 ± 1) × 10^3	(3 ± 1) × 10^6
Al	5852 ± 2	(3.8 ± 0.8) × 10^4	(3.9 ± 0.8) × 10^3	(4 ± 1) × 10^6

---

Table 2 MALDI-MS analysis of 1 µL of 100 nM (100 fmol) IGF-2 on different surfaces^a

	m/z ± SD	Intensity ± SD	S/N ± SD	Peak area ± SD
a Results are mean ± one standard deviation for 4 replicate analyses (fused silica) or 8 replicate analyses (ITO and Al).
Fused silica	7648 ± 145	(2 ± 1) × 10^2	26 ± 13	(2 ± 2) × 10^3
ITO	7542 ± 14	(1.0 ± 0.8) × 10^3	84 ± 40	(1 ± 1) × 10^5
Al	7469 ± 6	(1 ± 1) × 10^3	91 ± 41	(2 ± 3) × 10^5

---

Affinity MALDI-MS of insulin and IGF-2 on ILPRa- and ILPRh-modified ITO plates

Having established that ITO-coated surfaces are better than fused silica for detection of insulin and IGF-2 in conventional MALDI-MS, we next proceeded to test if this improvement would carryover to the detection of the captured proteins on the DNA-modified surfaces in our affinity MALDI-MS experiments. Fig. 2 shows results for insulin capture on ILPRa-modified ITO surfaces. Concentrations as low as 17 pM (51 amol total for 3 incubation/rinse cycles with 1 µL per cycle) were readily detected. This is two orders of magnitude improvement over previously reported affinity MALDI-MS of insulin on ILPRa-modified fused silica plates, which was able to detect insulin only down to the nM range.⁶ No peaks were observed for the capture experiments performed for insulin on ILPRi-modified (control) spots (also shown in Fig. 2) or on an unmodified ITO surface (not shown).

Fig. 2 Capture of insulin on ILPRa-modified and ILPRi-modified (control) ITO probe surfaces. Total insulin applied after 3 incubations of 1 µL of 1.72 nM (5 fmol), 17.2 nM (50 fmol) and 172 nM (500 fmol).

Based on the signal intensities of captured insulin at incubation concentrations of 1.72 nM, 172 pM and 17.2 pM, insulin exhibits higher binding affinity when it is present at sub-nM concentrations. This does not happen for insulin capture on ILPRh-modified surfaces, shown in Fig. 3, for which 1.7 nM solution gave a small peak and 17 pM was barely detectable. This interesting result supports the conclusions from previous surface plasmon resonance binding studies,⁶ in which insulin showed anomalously high affinity binding with ILPRa at very low (pM) insulin concentrations (K_D of 4 × 10⁻¹³ M for sub-nM insulin compared to K_D of 7 × 10⁻⁷ M for insulin at concentrations above nM), while a consistent K_D of 3 × 10⁻⁸ M was observed for insulin with ILPRh over the entire concentration range.

Fig. 3 Capture of insulin on ILPRh-modified ITO probe surface. Inset for 17.2 pM shows peak on expanded scale.

Affinity MALDI-MS results for IGF-2 capture on ILPRa- and ILPRh-modified ITO-coated surfaces are shown in Fig. 4. A concentration of 1 nM (3 fmol total) was readily detected on ILPRa and detectable on ILPRh as well. This is more than an order of magnitude improvement over the previously reported affinity MALDI-MS results for IGF-2 using ILPR-modified fused silica plates.⁶ As was the case for insulin, no peaks were observed for the capture experiments performed for IGF-2 on ILPRi-modified (control) spots or on a bare (unmodified) ITO surface (not shown).

Fig. 4 Capture of IGF-2 on ILPRa- and ILPRh-modified ITO probe surfaces.

Affinity MALDI-MS of human serum

Affinity MALDI-MS of human serum was performed on ILPRa- and ILPRh-modified ITO probe surfaces. Analysis of the serum without dilution and with moderate (10-fold and 100-fold) dilution yielded several peaks that were attributed to non-specific association of high abundance proteins (not shown). The serum sample was then diluted 10³-fold prior to analysis based on previous studies that showed that detection of low abundance proteins improved dramatically with increasing dilution of serum.¹⁰ This did not improve results and so the serum was treated with DTT prior to application in order to separate the disulfide-linked α and β chains of insulin. A peak was then detected at both the ILPRa- and ILPRh-modified ITO surfaces at m/z 3416 with a second peak at slightly longer m/z (Fig. 5). This is consistent with capture of the insulin β-chain (m/z 3430 for the unmodified peptide based on amino acid composition), which has been shown to contain the site of affinity interaction with ILPRa.^4,22 No peaks for the DTT-treated serum were detected on the ILPRi-modified or bare ITO control surfaces that were subjected to the identical capture/rinse protocols described for the ILPRa and ILPRh surfaces (not shown). It is important to note that proteins captured from serum may have undergone post-translational modifications and so it is difficult to predict the exact m/z. Assignment of the peak to the insulin β-chain is based not only on m/z but also on our previous studies that established the high affinity and selectivity of the interactions of insulin and IGF-2 with ILPRa and ILPRh.^6,22

Fig. 5 Affinity MALDI-MS of DTT-treated human serum (10⁴-fold dilution in PBS buffer) on ILPRa- and ILPRh-modified ITO surfaces.

In order to further support the identification of the captured protein as the insulin β-chain, in situ digestion of the captured proteins by chymase was performed directly on the ILPRa-modified ITO surface prior to application of MALDI matrix and analysis. Fig. 6 compares the results for serum with those for insulin and for a blank containing only chymase. The black rectangle in Fig. 6 contains peaks that are apparent for serum and insulin but not the chymase-only blank. In silico digestion of the insulin β-chain yielded seven peaks in the m/z 1450–1750 range included in the rectangle in Fig. 6, although the experimental peaks are not adequately resolved for specific assignments. Nevertheless, these results are consistent with the assignment of the peak at m/z 3416 in Fig. 5 to the insulin β-chain, which has seven possible digestion fragments at m/z 1456, 1505, 1520, 1603, 1618 1667 and 1683 in the m/z 1450–1750 range (not accounting for any post-translational modifications).

Fig. 6 In situ chymase digestion and MALDI-MS analysis of captured components from DTT-treated insulin and DTT-treated human serum on ILPRa-modified ITO surfaces. Top to bottom: chymase control (blank), DTT-treated insulin, DTT-treated human serum. Rectangular box indicates components tentatively attributed to insulin.

Conclusions

The results show that the anticipated improvements in m/z reproducibility and protein detectability obtained for conventional MALDI-MS analysis of insulin and IGF-2 through the use of ITO-coated surfaces instead of fused silica carryover to the ILPRa- and ILPRh-modified surfaces in affinity MALDI-MS. The improvements in performance are accompanied by the same ease of covalent DNA attachment and indefinite reusability of the DNA-modified surfaces demonstrated for fused silica in our previous work.^6,9,10 The use of ITO-coated surfaces is an important advance toward the routine use of DNA-modified surfaces in affinity MALDI-MS for rapid screening of DNA–protein interactions for applications such as biomarker discovery and detection of low abundance proteins in complex biological samples.

Acknowledgements

This material is based upon work supported by the National Science Foundation under CHE-0911108.

References

1. G. I. Bell, M. Selby and W. J. Rutter, The highly polymorphic region near the human insulin gene is composed of simple tandemly repeating sequences, Nature, 1982, 295, 31–35.
2. M. C. Hammond-Kosack, B. Dobrinski, R. Lurz, K. Docherty and M. W. Kilpatrick, The human insulin gene linked polymorphic region exhibits an altered DNA structure, Nucleic Acids Res., 1992, 20, 231–236.
3. A. Lew, W. J. Rutter and G. C. Kennedy, Unusual DNA structure of the diabetes susceptibility locus IDDM2 and its effect on transcription by the insulin promoter factor Pur-1/MAZ, Proc. Natl. Acad. Sci. U. S. A., 2000, 97, 12508–12512.
4. A. C. Connor, K. A. Frederick, E. J. Morgan and L. B. McGown, Insulin capture by an insulin-linked polymorphic region G-quadruplex DNA oligonucleotide, J. Am. Chem. Soc., 2006, 128, 4986–4991.
5. Z. B. Yu, J. D. Schonhoft, S. Dhakal, R. Bajracharya, R. Hegde, S. Basu and H. B. Mao, ILPR G-quadruplexes formed in seconds demonstrate high mechanical stabilities, J. Am. Chem. Soc., 2009, 131, 1876–1882.
6. J. F. Xiao, J. A. Carter, K. A. Frederick and L. B. McGown, A genome-inspired DNA ligand for the affinity capture of insulin and insulin-like growth factor-2, J. Sep. Sci., 2009, 32, 1654–1664.
7. W. Yoshida, E. Mochizuki, M. Takase, H. Hasegawa, Y. Morita, H. Yamazaki, K. Sode and K. Ikebukuro, Selection of DNA aptamers against insulin and construction of an aptameric enzyme subunit for insulin sensing, Biosens. Bioelectron., 2009, 24, 1116–1120.
8. J. D. Schonhoft, A. Das, F. Achamyeleh, S. Samdani, A. Sewell, H. Mao and S. Basu, ILPR repeats adopt diverse G-quadruplex conformations that determine insulin binding, Biopolymers, 2010, 93, 21–31.
9. L. W. Dick, Jr and L. B. McGown, Aptamer-enhanced laser desorption/ionization for affinity mass spectrometry, Anal. Chem., 2004, 76, 3037–3041.
10. J. R. Cole, L. W. Dick Jr, E. J. Morgan and L. B. McGown, Affinity capture and detection of Immunoglobulin E in human serum using an aptamer-modified surface in matrix assisted laser desorption/ionization mass spectrometry, Anal. Chem., 2007, 79, 273–279.
11. M. L. Vestal, et al., MALDI plate construction with grid, US pat., 7 030 373, April 18, 2006, H01J 49/04.
12. B. C. Senn, P. J. Pigram and J. Liesegang, Surface electrical resistivity and wettability study of fused silica, Surf. Interface Anal., 1999, 27, 835–839.
13. R. Knochenmuss, Photoionization pathways and free electrons in UV-MALDI, Anal. Chem., 2004, 76, 3179–3184.
14. V. Frankevich, J. Zhang, S. D. Friess, M. Dashtiev and R. Zenobi, Role of electrons in laser desorption/ionization mass spectrometry, Anal. Chem., 2003, 75, 6063–6067.
15. V. Frankevich, R. Knochenmuss and R. Zenobi, The origin of electrons in MALDI and their use for sympathetic cooling of negative ions in FTICR, Int. J. Mass Spectrom., 2002, 220, 11–19.
16. M. T. Bhatti, A. M. Rana and A. F. Khan, Characterization of rf-sputtered indium tin oxide thin films, Chem. Phys., 2004, 84, 126–130.
17. J. Davenas, S. Besbes and H. B. Quada, NIR spectrophotometry characterization of ITO electronic property changes at the interface with a PPV derivative, Synth. Met., 2003, 138, 197–200.
18. C. Liu, T. Mihara, T. Matsutani, T. Asanuma and M. Kiuchi, Preparation and characterization of indium tin oxide films formed by oxygen ion beam assisted deposition, Solid State Commun., 2003, 126, 509–513.
19. M. C. Galicia, A. Vertes and J. H. Callahan, Atmospheric pressure matrix-assisted laser desorption/ionization in transmission geometry, Anal. Chem., 2002, 74, 1891–1895.
20. P. Chaurand, S. A. Schwartz, D. Billheimer, B. J. Xu, A. Crecelius and R. M. Caprioli, Integrating histology and imaging mass spectrometry, Anal. Chem., 2004, 76, 1145–1155.
21. E. Moore, D. O'Connell and P. Galvin, Surface characterisation of indium-tin oxide thin electrode films for use as a conducting substrate in DNA sensor development, Thin Solid Films, 2006, 515, 2612–2617.
22. J. Xiao and L. B. McGown, Mass spectrometric determination of ILPR G-quadruplex binding sites in insulin and IGF-2, J. Am. Soc. Mass. Spectrom., 2009, 20, 1974–1982.

---