Separation by capillary electrophoresis followed by dynamic elution

ANALYTICAL

BIOCHEMISTRY

196,178-182

(19%)

Separation by Capillary Electrophoresis by Dynamic Elution Patrick Camilleri,

George

N. Okafo,

~m~thKline BeechamPh~r~eu~i~a~,

Received

December

and Christopher

4,199O

Over the past few years, it has been demonstrated (l-5) that capillary electrophoresis (CE)’ can be used for the high-resolution separation of a variety of substrates, including biomolecules (6-8). Because of its uniquely variable selectivity this analytical tool is now widely considered to be complementary to high-performance liquid chromatography. Currently HPLC is the most established separation technique for the analysis of biomolecules because it can be used preparatively for isolation and structure determination of peptides and proteins. Although CE can achieve a higher separation efficiency (9) than HPLC, the inability to use the technique preparatively has to date been a limitation. The direct coupling of CE with mass spectromet~ (MS) has demonstrated the usefulness of the combination for molecular weight determination (10-12). However, much less success has been achieved in the isolation of pep1 Abbreviations ylthiohydantoin.

178

Southan

Researchand De~elo~~nt, The Frythe, Welwyn, Nerts AL69AR, united Kingdom

In this study we have explored the behavior of peptides after capillary electrophoresis (CE) followed by elution under pressure. The use of D,O- rather than I&O-based buffer solutions appears to restrict the diffusion of peptides after CE, resulting in little loss of resolution when peptides are eluted by dynamic flow. In this paper we present results showing that a simple two-step process, involving CE at a low voltage, switching off the power supply, and connecting the fused capillary at the anode end to a syringe pump for dynamic flow, can retain separation characteristics and can be used for the isolation of picomole quantities of peptides for sequence Q 1991 Aeademic Press, Inc. determination.

roacetic acid, AUFS,

Followed

used: CE, capillary absorbance units

electrophoresis; at full spectrum;

TFA, trifluoPTH, phen-

tides or proteins by CE. This is due mainly to technical difficulties not easily surmountable and, in some cases (13,14), has involved the development of sophisticated instrumentation. MATERIALS

AND

METHODS

Sodium dihydrogen phosphate and disodium hydrogen phosphate were of AR grade (BDH). Buffer solutions were prepared with double-distilled deionized water (Mill;&) or with 99.9 at.% D deuterium oxide (D,O) purchased from Aldrich. The buffer solutions were adjusted to the required pH or pD with 0.5 M NaOH or 0.5 M NaOD (made by diluting a 40 wt% solution of NaOD with D,O) as appropriate. pH and pD were measured using a Radiometer PHM82 pH meter calibrated with standard pH buffers. pD measurements were made by the addition of 0.4 unit to the reading of the meter (15). Glucagon, trypsin, bradykinin, neurotensin, angiotensin II, Leu-enkephalin, and Phe-LeuGlu-Glu-Ile were used without further purification as purchased from Sigma Chemical Co. The tryptic digestion of glucagon was performed by adding trypsin (4 ~1 of a 5 mg/ml solution in water) to glucagon (400 ~1 of a 5 mg/ml solution in water) in Tris-HCI buffer (100 ~1 of a 1.0 M solution) at pH 8.00 and allowed to incubate at 39°C for 16 h. Digestion products were stored at -20°C.

The apparatus for CE has already been described elsewhere (16) and consisted essentially of a Glassman high-voltage dc supply (Whitehouse Station) and a capillary electrophoresis absorbance detector (CV4, ISCO). CE measurements were carried out within a protective 0003-2697/91 $3.00 Copyright 0 1991 by Academic Press, Inc. Ail rights of reproduction in any form reserved.

SEPARATION

BY

CAPILLARY

ELECTROPHORESIS

WITH

DYNAMIC

Perspex enclosure with interlock switches. Fused silica capillaries (Applied Biosystems) of 50 pm internal diameter and 72 cm total length (L) were used. The separation distance (L) from the anode to the detector was 50 cm. Capillaries were cleaned with 0.5 M NaOH or 0.5 M NaOD for 10 to 20 min and filled with the appropriate buffer solution using a high-pressure syringe pump (Harvard Model 440). Platinum electrodes were used for the connection of the voltage supply to the buffer reservoirs at each end of the capillary. All samples were dissolved in either water or the Tris buffer and were loaded electrokinetically by applying 5 kV for 2 s. For the CE separation of peptides in water or D,O, phosphate (20 mM) buffer at a pH or pD of 7.83 was used as the electrolyte. The voltage used for analytical CE separations was 10 kV; when CE separation was followed by dynamic elution a 6-kV potential was applied. Detection of the peptides was by uv absorption at 200 nm.

179

ELUTION

Cl.8 pl mine1

/

23.2

I

I

26.8

I

,

t

,

,

30.4

1.2 pl min.’

iii -!--

Dynamic Elution The CE separation of peptides followed by dynamic elution was carried out by first applying the separation

(a)

, 15.10

/

/ 17.54

,

I

Buffer

: 20mM NaH2P041 Na2HP04 in (a) Hz0 or (b) II20

Capillary

:L =72cm I =55cm

3 2

4 5

i.d = !XOpm

Ial 6.0

l;li: I 12.0

9.0

I 15.0

I

I 21.0

18.0

I x.0

7

I

23.2

t

I

26.8

30.4

1.5 pl min-l

pH : 7.81

’

r

19.98

1

27.0

14.2

Separation voltage

: 1OkV

Injection voltage

: 5 kVl2s

16.8

t

19.4

minutes

Wavelength

: 200 nm

Temperature

: Ambient

23.2

1

f

26.8

,

1

30.4

I

minutes

FIG. 2. CE separation of 6 kV followed by dynamic elution in (a) H,O and (b) D,O buffer solutions of the peptides in Fig. 1. The order of elution is the same as that shown in Fig. 1.

Time (minutes)

voltage (6 kV) for approximately 15 and 23 min for H,O and D,O buffer solutions, respectiveiy. After this time, the applied voltage was turned off and the anodic end of the capillary was removed from the reservoir and carefully attached to a l-ml buffer-filled syringe fixed to the Harvard high-pressure syringe pump. The contents of the capillary were then eluted at flow rates of 0.8 to 1.5 ~1 min-‘. Bradykinin was used as a model substrate for CE separation and collection. The peptide was first separated by CE (15 kV) and then eluted from the capillary (pump flow rate of 2 ~1 min-‘) into a small volume (5 ~1) of the buffer. This procedure was repeated several times until enough peptide was collected for further analysis.

pD : 7.83

1. Bradykinin 2. Neurotensin 3. Angiotensin

II

4. Leu-Enkephalin 5. Phe-Leu-Glu-Glu-lie

I

9.0

I 12.0

I

15.0

I

18.0

I

21.0

I

24.0

1

27.0

I

30.0

HPLC

Time (minutes)

FIG. H,O-

1. GE separation and (b) D,O-based

of five standard buffer solutions.

peptides

(1 mg/ml)

in (a)

Conditions

All solvents and reagents were HPLC grade. The configuration of the Applied Biosystems (Warrington, UK)

180

CAMILLERI,

OKAFO,

140A syringe pumps was modified to eliminate the large dead-volume of the standard dynamic mixer by connecting both pump outlets directly to an Upchurch 3-~1 static mixing tee (Anachem, Luton, UK). A Rheodyne 8125, fitted with a ~-PI loop, was used for sample injection. The uv detector was an Applied Biosystems 725, fitted with a 0.5~1 flow cell. Bubble formation during peptide collection was prevented by helium sparging of the solvents and the connection of a section of capillary tubing, 0.0025 in. i.d., to the flow cell outlet. Solvent A was 0.07% trifluoroacetic acid (TFA) in water; solvent B was 0.08% TFA in 70% acetonitrile/30% water. The column was a 0.95 X lOO-mm glass capillary tube containing 7-pm C8 wide-pore packing. This was emptied from a commercial cartridge (Applied Biosystems) and used to dry-pack the capillary as described in Ref. (17). Sequencing The collected RP-HPLC peaks from both the control and the CE-eluted samples were sequenced on an Applied Biosystems 477a pulsed-liquid phase protein sequencer with a 120a on-line PTH analyzer. After the addition of polybrene the sample disk was subjected to three normal cycles to reduce the PTH background before sample addition. RESULTS

AND

DISCUSSION

We have recently reported (16,lB) on the replacement of H,O- by D,O-based electrolytes for the analysis of a

AND

SOUTHAN

variety of substrates by CE. This methodology has been found to suppress electroosmotic flow and can result in enhanced resolution of substrates without the use of additives (19). Moreover, this technique results in an improved baseline compared to that obtained with H,Obased buffer solutions. D,O has the same electronic structure as H,O. However, the higher mass of deuterium restricts the movement of D,O compared to H,O molecules (20). This effect results in D,O solutions containing more structure and having a higher viscosity than Hz0 solutions. Consequently, substrates are expected to diffuse more slowly in D,O solution after CE. We tested this by comparing the behavior of five standard peptides (bradykinin, neurotension, angiotensin II, Leu-enkephalin, and Phe-Leu-Glu-Glu-Ile) after CE followed by dynamic elution resulting from the application of pressure. Figure 1 shows the CE separation of the five peptides carried out in the normal way in H,O- and D,O-based buffer solutions. Both electropherograms were obtained using a buffer of the same acidity (pH = pD) and a separation voltage of 10 kV. As expected, the migration times in D,O solution are longer than those in H,Obased electrolytes and the baseline is less noisy (fewer “spikes”) in the D,O solution. The next stage after these experiments was to carry out the two-step process: CE at a lower voltage (6 kV), turning off the high-voltage power supply, and carefully connecting the fused silica capillary at the anode end to a syringe pump for dy-

Buffer

:

Capillary

:

20mMSodium Phosphate in 0 20 (pD = 7.83) L =72cm I = 55cm i.d = 5Ot.u-n

w

Separation voltage Injection voltage Wavelength Temperature Dynamic Elution in (b)

I

I

I

I

,

11.8 13.0 14.2 15.4 16.6

I

17.0

I

1

19.0

20.2

I

Map

of the tryptic

digest

5kVi2a

:

200nm

:

Ambient

:

lclfmin-1

I

1

I

I

1

1

1

1

25.6 27.2 28.8 30.4 32.3 33.6 35.2 36.0 Time (minutes)

21.4

Time (minutes) FIG. 3.

:

of glucagon

by (a} CE and (b) CE followed

by dynamic

elution.

.

SEPARATION

BY

CAPILLARY

ELECTROPHORES~S

A

II I 20 Retention

I 10 time (min)

I 0

FIG. 4. Mierop~~arative RP-HPLC of a CE-eluted peptide. Samples were eluted by a gradient of 5 to 40% B in 18 min at 50 pfimin. The AUFS at 215 nm was 0.01. ~~~orn~~ograrn (a) was a control injeetion of 0.5 &I of a 5 rnM (w/v) solution of bradykinin. Chromatogram (b) was 4 pl from five pooled CE bradykinin peaks dynamically eluted. The injection loop was rinsed and a blank gradient run immediately preceding sample (b) to eliminate any carryover from the control peptide. The peaks in both (a) and (b) were collected for sequencing. The small peaks in both (a) and (b) were observed in blank gradients.

WITH

DYNAMIC

ELUTION

181

namic flow. These experiments were carried out in both H,O- and I&O-based electrolytes. Results are shown in Figs. 2a and Zb, respectively. The power supply was switched off at the time (minutes) marked by an arrow (4). As shown, the five peaks are baseline resolved in most cases, especially in the I&O-based buffer solutions. Compared with H,O, the use of DzO allowed the application of higher flow rates without the loss of resolution. Peptide 5 is highly negatively charged at the pH or pD under study so that its migration time is much longer than that of the other four peptides under normal CE (Figs. la and lb). Under dynamic elution the elution time of this substrate is greatly shortened, allowing analysis in a shorter period. Encouraged by the above findings, we applied the technique of CE followed by dynamic elution to the analysis of the tryptic digest of glucagon (1’7). This was a difhcult test case, as three of the expected four fragments migrate within about 0.15 min of one another (Fig. 3a). As shown in Fig. 3b, although some of the resolution of these three peaks has been lost, they are clearly identified. No attempt was made to optimize this separation, To establish whether the two-step technique could be applied preparatively, we carried out experiments on bradykinin. This peptide was injected five times under the conditions detailed under Materials and Methods. The peptide peaks collected by dynamic elution were pooled and injected into a microbore RP-HPLC system. The HPLC results for the CE-eluted bradykinin are shown in Fig. 4. The retention time of bradykinin after eleetrophoresis is identical to that obtained with an aliquot of a control sample. The minor peaks in both Figs. 4a and 4b were observed to a variable extent in injection blanks and additional control runs. The nature of these spurious peaks was not investigated further. The quan-

N-terminal protein sequence of the CE-eluted peptide. The first six cycles of PTH analysis are shown with the residue identified at FIG. 5. each cycle. Other peaks are normal by-products of the Edman chemistry except for a small amount of PTH-Gly contamination eluting at 9.25 min.

182

CAMILLERI,

OKAFO,

tity obtained from CE-dynamic elution was sufficient to determine the complete amino acid sequence of bradykinin after HPLC desalting. The first six sequencing cycles for the CE-eluted peptide are shown in Fig. 5, where the bradykinin N-terminal sequence Arg-ProPro-Gly-Phe-Ser can be clearly assigned. The result for the control bradykinin sample was similar in initial yield (in proportion to HPLC peak area), repetitive yield, and background levels. From t of the CE-eluted peak injected onto RP-HPLC the initial sequencing yield was 21 pmol. This is an underestimate of the total recovery from the CE capillary, as absorption losses will have occurred at each subsequent step and the initial yield by Edman degradation will represent less than the total amount of peptide applied to the sample disk. The method of dynamic elution successfully used in this work will allow the unique selectivity of CE for proteins and peptides to be exploited as a micropreparative technique. This not only will facilitate structural analysis by sequencing but also should provide sufficient material for high-sensitivity mass spectrometry without the voltage or current restrictions associated with current CE-MS interfaces (21). REFERENCES 1. Jorgenson, J. N., and Lukacs, K. D. (1983) Science 222,266-272. 2. Terabe, S., Otsuka, K., and Ando, T. (1985) Anal. Chem. 57,834841.

AND

SOUTHAN

3. Kuhr, W., and Yeung, E. S. (1988) Anal. Chem. f30,2642-2646. 4. Ewing, A. G., Wallingford, R. A., and Olefirowicz, T. M. (1989) Anal. Chem. 61,292A-303A. 5. Olefirowicz, T. M., and Ewing, A. G. (1990) Anal. Chem. 62, 1872-1876.

6. Kok, W. Th., and Bruin, G. J. M. (1988) Eur. Chromatogr. News 2(5),22-26. 7. Bruin, G. J. M., Chang, J. P., Kuhlman, R. H., Zegers, K., Kraak, J. C., and Poppe, H. (1989) J. Chromatogr. 471,429-436. 8. Grossman, P. D., Colburn, J. C., and Lauer, H. H. (1989) Anal. Biochem. 179,28-33. 9. Jorgenson,

J. W., and Lukacs,

10. Lee, E. D., Covey, Chromatogr. 458, 11. Lee, E. D., Muck,

K. D. (1984)

Science

T. R., Muck, W., and Henion, 313-321. W., and Henion, J. D. (1988)

222,266-272. J. D. (1988)

J.

J. Chromatogr.

458,313-321. 12. Udseth,

H. R., Loo,

J. A., and

Smith,

R. D. (1989)

Anal.

Chem.

61,228-232. 13. Tekiyiku, R., Keough, Rapid Commun. Mass

T., Lacey, Spectrosc.

M. P., and Scheider, 4(l), 24-29.

R. E. (1990)

14. Huang, X., and Zare, R. N. (1990) Anal. Chem. 62,443-446. 15. Glasoe, P. K., andLong, F. A. (1960) J. Phys. Chem. 64,188-194. 16. Camilleri, P., and Okafo, G. N. (1990) J. Chromatogr., in press.

17. Southan, T., Ed.),

C. (1989) in Techniques in Protein Chemistry pp. 393-409, Academic Press, San Diego.

(Hugli,

18. Camilleri, P., and Okafo, G. N. (1990) Submitted for publication. 19. Terabe, S., Yashima, T., Teraka, N., and Araki, M. (1988) Anal. Chem. 60,1673-1677. 20. Teraka, N., and Thornton, E. R. (1977) J. Am. Chem. Sot. 99,

7300-7307. 21. The

API

Book

(1989)

Central

Reproductions,

Ontario.