RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 9.97-102 (1995)

Utility of Organic Bases for Improved Electrospray Mass Spectrometry of Oligonucleotides Michael Greig and Richard H. Griffey* Isis Pharmaceuticals, 2292 Faraday Ave., Carlsbad, California 92008, USA

SPONSOR REFEREE: Professor J. A. McCloskey, University of Utah, Salt Lake City, UT 84112, USA

The sensitivity and accuracy of the mass spectrometric analysis of oligonucleotides using electrospray ionization can be compromized when the oligomer is adducted in the gas phase to cations such as sodium or potassium. We have evaluated the addition of mM concentrations of a series of organic bases with solution pK, values ranging from 11.5 to 5.5 and gas-phase proton affinities ranging from 213 to 232 kcal/mol as a method for suppression of signals from alkali-adducted ions. Stronger bases such as triethylamine and piperidine reduce the signals from bound sodium most effectively, but also decrease the total ion current from oligonucleotide. Imidazole, with a solution pH of -8.0, provides modest suppression of sodiumlpotassium adduct ions, but up to a four-fold improvement in sensitivity. Co-addition of imidazole and triethylamine or piperidine produces high ion abundance and good suppression of cation-adducted species for samples of phosphodiester or phosphorothioate oligomers which have not been desalted via preliminary precipitation or by high-performance liquid chroma- tography. Addition of high concentrations of imidazole generates a bimodal distribution of charge states, which may reflect different gas-phase conformations for single-stranded oligomers.

Electrospray mass spectrometry (ES-MS) has proven to be a gentle and sensitive method for the characteriza- tion of nucleic acids. ’.’ Analysis of oligonucleotides can be complicated by the solution affinity of the poly- anionic backbone for ubiquitous cations such as sodium and potassium. The origins of these cations include the controlled-pore glass used in synthetic preparation of DNA or RNA, high-performance liquid chromato- graphic (HPLC) purification procedures, or cations leached from plastics or glassware used in sample handling and analysis. The bound cations reduce sensit- ivity for the analyte and distribute the signal from any charge state across an envelope of cation-containing species, complicating data analysis and impairing the performance of the instrument.

Several methods have been described for desalting natural nucleic acids and oligonucleotides, including reversed-phase HPLC and precipitation from concen- trated ammonium acetate.’-‘ Reversed-phase HPLC adequately removes bound cations from standard DNA and RNA, but fails for modified oligomers such as phosphorothioates. Precipitation from ammonium ace- tate is a simple and efficient method for exchanging cations, but the method is not quantitative, the effec- tiveness is a function of sequence length, and loss of analyte during precipitation is a concern.

A second strategy for reducing the number of phosphate-bound cations is addition of chelators or volatile cations to the sample during analysis. Grotjahn has demonstrated the utility of triethylamine (TEA) for reduction of cation adduction in fast-atom bombard- ment (FAB) mass spectrometric analysis of oligonucleotide^.^ Limbach and coworkers have exa- mined the addition of cyclohexyldiaminetetraacetic acid (CDTA) and TEA to solutions of precipitated

’ Author f o r correspondence

tRNAs and 5s RNA.5 Their results show that magne- sium and sodium adduct ions can be reduced to levels which permit accurate mass determination. Potier et a f . have shown that sodium-adducted species present in precipitated, HPLC-purified synthetic DNA oligomers can be dramatically reduced through addition of high (1%) concentrations of trieth~1amine.l~ In addition, their work suggested that the degree of sodium-ion replacement improved with addition of triethylamine compared to trimethylamine or ammonia.

We have investigated the addition of various concen- trations and mixtures of organic buffers for the suppres- sion of cation-adducted ions and enhancement of signal from oligonucleotides. Buffers have been studied with a range of pK,, volatility, and gas-phase proton affi- nity. Addition of strong bases such as triethylamine and piperidine provides the best suppression of adducts. These strong bases also can suppress ion abundance, and co-addition of a weaker base such as imidazole improves the ES-MS sensitivity for different classes of oligonucleotides, without the need for any sample pre- treatment.

EXPERIMENTAL

Preparation of oligonucleotides. Samples of DNA oli- gonucleotides 1 and 2 were synthesized using the solid- 1 5’-dTTGCTTCCATCTTCCTCGTC (uniform P=S) 2 5’-dTGAGTCAGACGCATCGTCGTCATGG

(uniform P=O)

Sequence and composition of DNA oligomers. 1, DNA 20-mer

phase phosphoramidite method on an ABI 390B syn- thesizer (ABI, Foster City, USA) with commercial phosphoramidites (Millipore, Boston, MA, USA). Phosphorothioate oligomers were prepared using

phosphorothioate; 2, DNA 20-mer phosphodiester.

CCC 095 1-4198/95/010097-06 0 1995 by John Wiley & Sons, Ltd

Received 9 November I994 Accepted 10 November I994

98 IMPROVED ES-MS OF OLIGONUCLEOTIDES

Abundance

50000

40000

30000

20000

10000

0 ‘I 000 1050 1100 1150 1200 1250

I d Z

1 Abundance [M-SH]’- 1252.3

50000

40000

30000

20000

10000

0 ‘I 000 1050 1100 1150 1200 1250

I t&

Figure 1. Expanded electrospray mass spectra of DNA oligorner 1 ethanol-precipitated from 10 M ammonium acetate: (a) control; (b) with dissolution in a 10 rnM irnidazole buffer.

Beaucage reagent to generate the P=S bonds.’ Where indicated, samples were precipitated from 10 M ammo- nium acetate using 9 :1 v : v cold ethanol. Other chemi- cals were purchased as the highest available grade (Aldrich Chemicals, Milwaukee, WI, USA). Mass spectrometry. A Hewlett-Packard (Palo Alto, USA) HP 5989 quadrupole mass spectrometer with an extended mass range (2600 mlz) was employed in the negative ionization mode for all studies. Electrospray was achieved using an HP 59987A source with a pneu- matic nebulizer and a 70psi flow of nitrogen gas. The curtain gas mixture consisted of 12.5 L/min nitrogen and 2L/min oxygen. For all studies samples were dissolved in 100 pL of a 1:l mixture of aqueous buffer + isopropanol to give a final concentration of 3 pmol/pL. Samples were infused at a rate of 6 pL/min with an HP 1090 HPLC pump. The cylinder voltage was main- tained at + 3500 V relative to the needle, with a skim- mer voltage of - 28.5 V and an entry lens voltage of + 98 V. A total of 15 scans was averaged with a step size of 0.1 over the measujed mass range of mlz 600- 1650. Molecular masses were determined using the HP Chemstation analysis package.

RESULTS AND DISCUSSION Adduction of cations to the phosphate backbone is a major problem in the analysis of oligonucleotides. Bulk exchange with ammonium ions via precipitation reduces the level of cations to levels where the molecu- lar ion can be observed in many cases, but residual adducts still complicate the mass spectrometric analy- sis. Cation adduction is severe for phosphorothioate oligomers, as illustrated in Fig. l(a) for DNA phos- phorothioate 20-mer, 1, which has been ethanol- precipitated from 10 M ammonium acetate. The [M - 6HI6- and [M - 5H]’- ions are observed at m/z 1043.4 and 1252.2, respectively, along with the mono-, di-, tri-, and tetrasodium adducts.

Seven bases, listed in Table 1 , have been examined as additives to a 3pmoUyL concentration of the oli- gomer. The bases cover a range of solution pH values and gas-phase proton affinities,’, lo and concentrations of 10 and 50mM have been compared. Above this level, all bases suppress the signals from oligonucleo- tide. The abundance of the [M - 5H]’- ion at mlz 1252 and the associated sodium-adducted ions are listed in Table 1, along with properties of the bases.

IMPROVED ES-MS OF OLIGONUCLEOTIDES 99

Table 1. Correlation of relative ion abundance with physical properties of organic bases for the [M - 5HI5- charge state of oligomer 1 observed at mlz 1252

Relative ion abundanm

Compound pH" PAb MC [M+Na]+ [M+ZNa]+ [M+3Na]+ Control 7.0 - 4200 4600 5000 4800 Triethylamine 11.5 232.3 18300 980 280 Piperidine 10.5 226.4 22500 4200 1500 280

Imidazole 8.0 223.5 49000 27000 14000 14000 3,5-Dimethylpyrazole 6.0 220.6 10800 10900 12000 11400

Pyrazole 5.5 212.7 4000 4200 4400 4200

* Measured pH of aqueous soluton of oligonucleotide with buffer.

measured directly from the 50 mM stock solution of buffer.

dBelow limit of detection.

d

N-Methyl morpholine 9.0 12000 14200 10200 4400

4-Methylpyrazole 5.8 215.7 9000 7200 3900 2200

Value of gas phase proton affinity (kcal mol-') from Refs 9 and 10. Values of pH were

Ion abundances measured at rnlz 1252.2, [M - 5HI5- in presence of 50 mM buffer.

Triethylamine and piperidine are strong, volatile bases and provide up to 100-fold suppression of the signal from sodium-adducted species at a 50 mM concentra- tion. However, the total ion current from oligonucleo- tide decreases with increasing pH of the resulting solu- tion. The relative ion abundance increases with basicity for the pyrazole buffers, but no suppression of bound sodium is observed.

At a 10mM concentration, imidazole provides a four-fold increase in ion abundance for the [M - 5H]'- charge state, and cation suppression of 2 30%. Fig. l(b) contains the spectrum obtained after the oligomer was dissolved in a buffer containing 10 mM imidazole. The ratio of parent ions to sodium adducts has been increased five-fold. The relative abundance of the ion at mlz 1252.2 has been increased four-fold relative to the control spectrum, and increases in abundance are observed for the other charge states. Addition of N-methylimidazole, with higher volatility and proton affinity, does not suppress the signal from sodium- adducted species or improve the sensitivity (data not shown).

Co-addition of 25 mM imidazole and 25 mM piperi- dine or TEA with oligomer 1 provides the cation suppression of the stronger base and a 15-fold increase in ion abundance. As shown in Fig. 2, this buffer combination improves the appearance of the spectrum and allows lower abundance impurities to be resolved and detected. A low-mass shoulder (labeled P=O) is seen in Fig. 2(b) for the [M-5HI5- and [M-4HI4- charge states. This ion is generated from the 2% of the molecules containing a single phosphodiester for phos- phorothioate substitution, and reduces the mass of the oligomer by 16Da. In addition, signals (labeled (N-1)) are observed at lower mlz from 5-6 failure sequences produced by incomplete coupling during the synthesis. These impurities are difficult to discern in the control spectrum.

The distribution of charge states detected for an oligomer is also a function of the added buffer. As shown in Table 2, the relative ion abundance for the higher charge states is increased for TEA and piperi- dine. Lower concentrations of imidazole favor the higher charge states. However, in the presence of 50 mM imidazole, the abundance of the [M - 5H]'- species is enhanced dramatically relative to the higher

charge states. As shown in Fig. 4, a bimodal distribu- tion of charge states is realized for oligomer 1 at a 10 mM concentration of imidazole. This reproducible shift in charge state suggests that imidazole can induce structural alterations in a non-self complementary oli- gomer, which are retained in the gas phase.

Next, we investigated the effects of co-addition of piperidine and imidazole on a phosphodiester oligomer isolated via standard ethanol-precipitation from 0.5 M sodium acetate. The mass spectrum from a non- desalted control sample of oligomer 2 is presented in Fig. 3(a). A family of cation-adducted ions is observed for every charge state grouping. Co-addition of 5 mM piperidine to the sample results in a dramatic improve- ment in the appearance of the data (Fig. 3(b)), with appearance of a single molecular ion from the families of cation-adducted species for each charge state. Finally, co-addition of 2.5 mM piperidine and 2.5 mM imidazole produces a 40-70% increase in ion abun- dance from the [M - 5HI5- through [M - 9HI9- charge states, with good suppression of the signal from adducted ions. This ion exchange during sample injec- tion obviates the need for a prior precipitation or HPLC treatment, and should improve the analysis of oligomer samples such as tissue extracts which contain a range of oligomer lengths.

Ionization in the presence of imidazole and a strong base such as TEA or piperidine provides a dramatic decrease in the amount of associated sodium and an increase in the abundance of ions from oligonucleotide. The pyrazoles, which are weak bases relative to the phosphodiester or phosphorothioate backbone in the gas phase, fail to exchange a proton for sodium. Suppression of cation-adducted ions does not correlate with the gas-phase proton affinities of the bases, which bracket the value expected for a phosphorothioate or phosphodiester. l1 N-Methyl imidazole does not equal the level of sodium suppression of the less volatile imidazole, despite a similar solution pK, and proton affinity. Although imidazole has low volatility, no ions from adducts to the oligonucleotide are observed.

The suppression of cation adducts induced by organic buffer correlates with basicity for both phosphodiester and phosphorothioate DNA. Previous studies have shown that addition of the strong base triethylamine leads to improved suppression of adducted species for

100 IMPROVED ES-MS OF OLIGONUCLEOTIDES

Abundance

90000

80000

700 800 900 1000 1100 1200 1300 1400 1500

Abundance

90000

80000

70000

60000

50000

40000

30000

20000j I 10000~ I

[MJH]'- 1252.3

[M-4Hj4- 1565.6

L r d z

Figure 2. Electrospray mass spectra of DNA oligomer 1 precipitated from 10 M ammonium acetate: (a) control; (b) with dissolution in a buffer containing 25 mM imidazole and 25 mM piperidine.

phosphodiester and phosphorothioate ~ l igomers . '~ The role in increasing ion abundance from oligonucleotides, combination of a strong base and imidazole minimizes since the protonated species can bind to two hydrogen- the effective salt-concentration in solution due to bond acceptors. Loss of imidazole during the desolva- ionized organic species and generates a higher abun- tion process rather than the imidazolium cation would dance of analyte ions. The unique hydrogen-bond- reduce charging of the phosphodiester backbone, as donating/accepting ability of imidazole could play a observed at the higher imidazole concentrations.

Table2. Ion abundance from various charge states of oligomer 1 as a function of buffer composition

Buffer mixture" [M-9HIy [M-XHIX [M-7HI7 [M-hHj' [M-5HI5 [M-4Hj4

Control 2100 4400 4300 9400 8500 1900 1 I Imidazole+TEA 22000 68000 52000 56000 83000 51000 1 1 Imidazole+Piperidine 23000 68000 65000 52000 91000 71000

Piperidine 64000 56000 47000 36000 46000 35000 TEA 12800 27000 26200 20400 18300 20200

"Total buffer concentration fixed at 50 mM

Imidazole 800 6 0 0 1 loo0 31000 71000 4000

IMPROVED ES-MS OF OLIGONUCLEOTIDES 101

Abundance

50000 45000

40000 35000

30000

25000 20000

15000

. . i . .-. . I . . ' . I . . . ' " " 700 800 900 1000 1100 1200 1300 1400 1500 1600

I t &

1282.0

1538.7

700 800 900 1000 1700 1200 1300 1400 1500 1600

Abundance

50000

45000

40000

35000

30000

25000

20000 ;::::/kL,w 5000

0 700 800 900 1000 1100 1200 1300 1400 1500 1600

Figure 3. Electrospray mass spectra from DNA oligomer 2: (a) crude DNA ethanol-precipitated from 0.5 M sodium acetate without pretreatment; (b) with dissolution in a buffer containing 5 mM piperidine; (c) with dissolution in a buffer containing 2.5 mM imidazole and 2.5 mM piperidine.

102 IMPROVED ES-MS OF OLIGONUCLEOTIDES

Abundancei (M6HJ‘ 1043.4

I [M-9HIR 3 6911 40000 -- 36000 I 32000 1 28000 1 I I 24000 20000 16000 12000 8000 4000

1252.3

600 650 700 750 800 850 900 950 1000 1050 1100 1150 1200 1250 m / Z

Figure 4. Effect of a 10 mM imidazole buffer on the electrospray mass spectrum of DNA oligomer 1.

Application of the technique may be limited for mass spectrometric analysis of RNA, where 0.2 M concen- trations of imidazole can catalyze strand cleavage.”

CONCLUSIONS The levels of cation-adducted species observed in the ES-MS analysis of oligonucleotides can be reduced 100- fold through addition of organic cations in the buffer solution. The method works well for phosphoro- thioates, which have a much higher affinity for sodium and are difficult to desalt using conventional tech- niques. A strong base such as TEA or piperidine provides the best suppression of cations adducted to the oligonucleotide. Co-addition of imidazole leads to an additional increase in sensitivity, along with an alter- ation in the ion abundance for low- and high-charge states. The co-addition of imidazole and a strong base to samples of oligonucleotide will allow more accurate mass measurement and provides a convenient alterna- tive to preliminary desalting procedures. The effects of imidazole on gas-phase structure of oligonucleotides and nucleic acids are under further investigation.

Acknowledgements We wish to thank Steve Fisher for helpful discussions and Matt Sorensen for synthesis of DNA samples. We thank Prof. James McCloskey and co-workers for a preprint of Ref. 5.

REFERENCES

1. T. R. Covey, R. F. Bonner, B. L. Shushan, and J . Henion, Rapid Commun. Mass Spectrom. 2, 249 (1988).

2. R. D. Smith, J. A. Loo, C. G. Edmonds, C. J. Barinaga and H. R. Udseth, Anal. Chem. 62, 882 (1990).

3. S. C. Pomerantz and J. A. McCloskey, Methods Enzymol. 193, 796 (1990).

4. J. T. Stults and L. C. Marsters, Rapid Commun. Mass Spectrom. 5 , 359 (1991).

5. P. A. Limbach, P. F. Crain and J. A. McCloskey J. Am. SOC. Mass Spec. (in press).

6. D. P. Little, R. A. Chorush, J. P. Speir, M. W. Senko, N. L. Kelleher, and F. W. McLafferty, J . Am. Chem. SOC. 116, 4893 (1994).

7. L. Grotjahn, H. Blocker and R. Frank, Biomed. Mass Spectrom. 12, 514 (1985).

8. R. P. Iyer, L. R. Phillips, W. Egan, J. B. Regan and S. Beaucage, J. Org. Chem. 55,4693 (1990).

9. J. Catalan, J. L. G. de Paz, M. Yanez, R. M. Claramunt, C. Lopen, J. Elguero, F. Anvia, J. H. Quian, M. Taagepera and R. W. Taft, J. Am. Chem. SOC. 112, 1303 (1990).

10. B. Nogaj, E. Dulewicz, B. Brycki, A. Hrynio, P. Barczynski, Z. Dega-Szafran, P. Koziol and A. R. Katritzky, J. Phys. Chem. 94, 1279 (1990).

11. R. V. Hodges, T. J. McDonnell and J. L. Beauchamp, J. Am. Chern. SOC. 102, 1327 (1980).

12. R. Breslow and R. Xu, Proc. Natl. Acad. Sci. USA 90, 1201 (1993).

13. K. Bleicher and E. Bayer, Biol. Mass Spectrom. 23,320 (1994). 14. N. Potier, A. Van Dorsselaer, Y. Cordier, 0. Roch and R.

Bischoff, Nucl. Acids. Res. 22, 3895 (1994).