CLIN. CHEM.

37/1, 14-19 (1991)

Capillary Electrophoresis-a New Clinical Tool

Fu-TalA. Chen,Cheng-MIngLIu,You-ZungHsleh,and James C. Sternberg

The potential clinical diagnostic utility of capillary electro- chlorideby conductimetric detection in a Teflon capil-
phoresis in an open-tubular column is established. Sep- lary of 200-gm inner diameter. Jorgenson and Lukacs
aration patterns for serum proteins by conventional aga- (2) did extensive studies on capillary electrophoresis in
rosegelelectrophoresis can be faithfully reproduced by a glass and fused-silica columns for several other applica-
capillary electrophoresis procedure that provides the tions with optically based detection. Some of their ear-
complete run data in 8 mm. Hemoglobin variants can be lier capillary electrophoresis studies with a mixture of
separated within 10 mm. The capillary electrophoresis dansyl amino acids showed a separation efficiency of
separations are performed reliably and reproducibly in an -250 000 theoretical plates in a voltage gradient of 300
untreated 75 Mm (i.d.) x 25 cm fused-silica column. V/cm.
Diluted serum or hemoglobin samples can be loaded on Hjerten et al. (10) and Cohen and Karger (11) intro-
an automated instrument with on-line injection, detection, duced gel-filled capillary electrophoresis for protein and
and quantification, providing a truly “walkaway” electro- nucleic acid separation with great resolution. Terabe et
phoresis system. al. (12) expanded open-tube capillary electrophoretic
separation for noncharged species, using an ionic deter-
AddItIonalKeyphrases:proteins hemoglobin urine gent as the charged carrier; the separation principle
cerebrospinal fluid agarose gel electrophoresis compared they used is based on the partition of solutes between an
aqueous phase and a micellar phase (acting as a quasi-
stationary phase), similar to the chromatographic sys-
Electrophoresis is an extremely powerful tool for
tem named electrokinetic chromatography (EKC), do-
separating ionic species, especially biopolymers such as
veloped by Terabe (13).
proteins and DNA. During the last 40 years, zone
The application of capillary electrophoresis (CE) to
electrophoresis on paper, starch gel, cellulose acetate,
and agarose or polyacrylamide gels has been used to clinical samples, such as urine, has been demonstrated
by Jorgenson and Lukacs (3) with an open-tube capil-
separate peptides, proteins, and polynucleotides. This is
lary. Plasma sample separations have also been at-
indeed one of the most important and fundamental tools
tempted by Jorgenson and Lukacs (4), using a coated-
in molecular biology and diagnostic medicine. Offsetting
tube capillary.
its versatility and utility, however, is the reality that
The objective of this report is to establish the poten-
this method is labor-intensive, skill-dependent, and rel-
tial utility of capillary electrophoresis for clinical diag-
atively slow.
nostic applications, with particular emphasis on using
Recent developments with capillary columns repre-
open-tube, untreated fused-silica capillary columns.
sent an important technical advancement in analytical
Analyses of serum proteins and hemoglobin variants by
zone electrophoresis (1-9). The technique with capillary
a capillary electrophoresis method are compared with
columns is well-suited for automation with real-time
those obtained in conventional agarose gel-based elec-
data analysis. Sample quantities are in the nanoliter
trophoresis systems. Several other diagnostic applica-
range, and microliters of buffer reagent are consumed.
tions of capillary electrophoresis, e.g., to urine and
In a sense, the buffer in a capillary is as stable against
spinal fluid samples, are also described. The following
convection and diffusion as a viscous gel, so that sepa-
report summarizes our approach and the results ob-
rations obtained in an open capillary are quite similar
tained with this method, in comparison with several
to those obtainable in an agarose gel. Analytes in the
other established methods.
sample are similarly electrophoretically separated into
discrete zones without significant diffusion. Materials and Methods
Capillary zone electrophoresis was first developed by
Reagents. Serum protein analysis and hemoglobin
Mikkers et al. (1) for separation of small anions such as
assay by an agarose gel electrophoresis method were
performed on the Paragon#{174}
electrophoresis system, with
Applied Research Department, Diagnostic Systems Group, “SPE gel” and “Acid-Jib gel” from Beckman Instru-
Beckman Instruments Inc., 200 S. Kraemer Blvd., Brea, CA ments, Inc., Brea, CA. All procedures were performed
92621. according to the manufacturer’s instructions. Control
Presented at the 22nd annual Oak Ridge Conference on Ad- samples of serum protein and a mixture of hemoglobin
vanced Analytical Concepts for the Clinical Laboratory, Tampa,
FL, April 1990. A, 5, and C variants were obtained from Beckman
Received June 14, 1990; accepted October 17, 1990. Instruments, Inc. A mixture of hemoglobin control A, F,

14 CLINICAL CHEMISTRY, Vol. 37, No. 1, 1991

S, and C variants was purchased from Isolab Inc.,
Akron, OH.
o.o
Patients’ samples. Serum samples and controls were
diluted 20-fold in buffer containing 75 mmol of sodium
chloride and 20 mmol of potassium phosphate per liter
at pH 7.0. Hemoglobin controls were diluted 10-fold in
the buffer. Diluted samples were injected by the elec- 0.01

trokinetic method.
Capillary electrophoresis (CE) system. A CE system
(Beckman Instruments, Inc., Palo Alto, CA) was used
with Beckman system software, controlled by an IBM
PS/2 model Z-50. Data analysis was performed with
System Gold TM software (Beckman Instruments, Inc.,
San Ramon, CA).
The CE system contains built-in 214-, 254-, 280-, and
415-nm narrow-band filters for on-line detection. Elec-
trophoreses were performed in a fused-silica tube, 75
pin (i.d.) x 25 cm, supplied by Polymicro Technologies,
Inc., Phoenix, AZ. The detection window is located -6.5
cm from the column outlet. The capillary column was
assembled in the cartridge format of Beckman Instru-
ments, Inc.
Capillary electrophoresis procedures. Samples were
placed on the inlet tray of the CE instrument. The
system was programmed for various assays, with auto-
mated injection by the electrokinetic method for 3 to 10
s at 1 kV. Serum proteins and of hemoglobin variants
were separated in <10 mm with use of a column voltage
gradient of 200 V/cm. The capillary column was washed
and reconditioned between each run. The system can be
progranuned to run samples in a random-access manner
for as many as 10 different assay methods.
Buffer for the serum protein assays is a Beckman
proprietary buffer at pH of 10.0. Hemoglobin variants A,
F, S, and C were separated with use of another Beckman
proprietary buffer at pH of 8.6. (These buffers will be
made available to the research community.) All buffers
were ifitered through a 0.45-pm (pore size) filter.
Results
Figure 1A shows the separation of the hemoglobin
variants control in an untreated fused-silica capillary
column. The effective separation of hemoglobin A, S,
and C is evident within 10 mm and the resolution is
similar to, if not better than, that obtained with a
traditional agarose gel (Figure 1B).
At pH 8.6 for the electrophoresis buffer, all hemoglobin
species contain a net negative charge, yet migrate
towards the cathode. This is due to the electroendosmotic
flow resulting from the negative charge-bearing fused-
silica capillary, which attracts a layer of positive ions
from the solution. As these ions flow towards the cathode
under the influence of the electric potential, bulk solution
must flow in this direction to maintain electroneutrality.
This electroendosmotic flow provides a fixed velocity
component that sweeps both neutral species and ionic
species, regardless of charge, towards the cathode. Each
of the hemoglobin species is also driven towards the
anode, against the electroendosmotic flow, at its own
velocity, determined by its electrophoretic mobility. Each
0
Tim. Imnuic)
B
(+) (-)
Fig. 1. Electropherogramof hemoglobincontrolA, 5, and C
(A) Capillary electropherogram of hemoglobin A, S, and C. Conditions:
Fused-silicacapIllary, 75 m (id.) x 25 cm; detection at 415 nm; applied
potential, 5 kV. Total hemoglobIn,
2.5 g/L; hemoglobinA, 65.3%; S, 23.9%;
( Electropherogramof the same sample on Acid-Hb gel with
and C, 10.8#{176}!..
the Paragon agarose gel electrophoresissystem
hemoglobin species thus has its own net downstream
velocity (the difference between the general electroendos-
motic velocity and its own electrophoretic velocity), lead-
ing to the separation of the different hemoglobin species.
Similarly, hemoglobin F can be separated from A, 5, and
C species in the same buffer by using slightly modified
electrophoresis conditions (Figure 2).
We sought to perform capillary electrophoresis of
serum proteins in an untreated capillary column in a
manner that would closely reproduce the serum protein
patterns obtained from conventional agarose gel-based
electrophoresis. Figure 3 exhibits the patterns of normal
control serum protein in the capillary electrophoresis
and agarose gel electrophoresis methods.Each of the
serum protein fractions can clearly be distinguished in
the capillary method: gamma globulin appears first,
followed by beta, alpha2, alpha1, and finally albumin.
The capillary electrophoresis pattern (Figure 3A) was
plotted reversely for easy comparison with the pattern
from agarose gel (Figure 3B).
An abnormal control serum sample shows almost
identical separation patterns by capillary and agarose
gel electrophoresis methods (Figure 4). Several serum
samples from myeloma patients were analyzed in both

CLINICALCHEMISTRY,Vol.37, No. 1, 1991 15

 0.02 - A

I. 001

6 ‘1 I,,
Imc Incinul.)

Fig. 2. Capillary electropherogram of hemoglobin control A, F, S, and

C
Conditions:Fused-silicacapillary,75 im (id.) x 25cm; detection at 415 nm;
applied potentIal, 7.5 kV. Total hemoglobIn,2.5 g/L; hemoglobinA, 32.4%; F,
26.9%; S, 24.2%; and C, 16.5%

006

(+)
(-)
006

Fig. 4. Electropherogramof abnormal control serum protein

Conditionsas in Fig. 3

but the difference is less than 10 a. This indicates a

well-behaved reproducibility of the system.
Peptide bonds absorb strongly around 210 nm. Serum
proteins in capillary electrophoresis are monitored by

A
0
B
their absorbance at this wavelength. Figure 7 shows the
absorbance for equal concentrations
min, and immunoglobulins.
of peptides, albu-
The absorbance curve for
each compound is quite similar to each of the others on
a mass basis. In agarose gel, however, the dye binding
(+) ) constant for each protein is quite different, so that the
mass concentration of individual fractions cannot be
obtained directly from the intensity reading of a densi-

jt
!iii; .
tometer. Therefore, the absorbance measured in capil-
lary electrophoresis appears to be a reliable quantita-
fii! .

I(f!!Ti
tive index for proteins in general.
Cerebrospinal fluid (CSF), with a protein concentra-
Fig. 3. Electropherogram of normal control serum protein tion of 300 mg/L in most healthy persons, can be
(A) Capillary electropherogramof normal serum protein.Conditions:Fused- analyzed in capillary electrophoresis by direct sample
silica capillary,75 an (l.d.) x 25cm; detectionat 214 nm; applied potential,5
kV. ( Electropherogram of the same sample on SPE gel with the Paragon injection, without pre-concentration of the specimen.
agarose gel electrophoresls system Figure 8A shows a normal pattern for proteins in CSF,
similar to that for proteins in normal serum, but includ-
ing a distinct peak for prealbumin, which clearly mi-
capillary and agarose gel electrophoresis systems for grates after albumin. Several peaks outside of the nor-
comparison; all exhibit almost identical patterns by the mal protein bands are presumably small molecules,
two methods (Figure 5). Figure 6 shows the electro- which are detectable in capillary electrophoresis be-
pherograms for sequential runs of a normal serum cause they are not lost in a fixing and staining process.
sample in an untreated fused-silica capillary. After 20 Figure 8B exhibits the electropherogram of the same
repeated runs, the elution time of each peak does vary, CSF sample after dialysis through a membrane having

16 CLINICAL CHEMISTRY, Vol. 37, No. 1, 1991

 0012

oas

J 0064
(+) (-)

I
(+) (-)

(+) (-)
002

Tun. (,.nncncl
Fig. 5. Electropherograms of serum proteins from (bottom) an IgG kappa myeloma patient, (middle) an 1gMlambda myeloma patient, and (top)
an lgG3 myelomapatient
Conditionsas in FIg.3

a molecular mass cutoff of 14 kDa. DIscussIon

Figure9 demonstrates the capillary electrophoresis of a
normal urine specimen, showing the fairly complex pat-
terns of absorbance at 214 and 254 nm. The pattern
indicates the presence of many small molecules, including
urea and creatinine, and degradation products of proteins,
probably primarily peptides. The region between 3.5 and 6
minis where ordinary serum proteins would appear; thus,
a dialyzed urine sample displays the more typical protein
pattern in Figure 1(18, whereas an undialyzed urine sam-
pie exhibits the complex pattern in Figure 1(14.
Proteins have been known to adsorb to the surfaces of
both fused silica and glass. Earlier attempts to electro-
phorese proteins in untreated fused-silica capillaries
resulted in irreproducible migration of all sample zones
(4). Jorgenson and Lukacs (4) attempted serum protein
analysis by using glycol-modified fused-silica capillaries
(14), with limited success. The instability and limited
tolerable pH range of most surface-modified capillaries
prevent their routine utilization.
Lauer and McManigill (15) proposed that protein

CLINICALCHEMISTRY,Vol.37, No. 1, 1991 17

 0.04

Upper trace . 204 006

lowe. 0608 . 254 Ofl

I 51
002

I
I

I4:
.

4 0 ,, 8 0
Ti,,,. ii, inirnales I ,,,c (minulel

Fig.6. Reproducibility of capillaryelectrophoresisstudy:(A) 1st run, Fig. 9. Capillary electropherogram of a normal urine sample moni-
(B)20th run tored at 214 and 254 nm
Conditions as in Fig.3A Conditionsas in Fig. 3A

00.1

002

2 4 6 8 10
tim. ,n,flo,0(

-0.01
200 216 224
Wane length
001

Fig. 7. UltravIolet spectra of proteins from serum, dissolved at 50

mg/i. in Isotonicsaline: (A) albumin;(B) alpha1-antitrypsin (alpha1-
globulin); (C) immunoglobulin G (gamma globulin); (L hapto-glob- I
ulin (beta-globulin)
j 0.02

001$ I
0.01

- tm,sm,no,.)

Fig. 10. Capillary electropherogram of urine from a patient with

Bence Jones proteins before (A) and after (B) dialysis through a
membranehavinga molecularmasscutoff of 14 kDa
Conditions as in Fig. 3A

adsorption could be minimized under conditions in

which both the proteins and the surface are negatively
charged. High salt buffers can also reduce protein ad-
-01.0
sorption. Lysozyme migrates as a sharp zone faster than
I,n,, (n,,n,,,eI
the electroendosmotic flowin a pH 9.0high saltbuffer,
Fig. 8. Capillary electropherogram of a CSF sample before (A) and
after (B) dialysis through a membrane having a molecular mass even though the buffer is 2 pH units below the isoelec-
cutoffof 14 kDa tric point of lysozyme, 11. Therefore, Jorgenson (16)
Conditionsas in Fig.3A suggested that the protein adsorption may be due to the

18 CLINICALCHEMISTRY,Vol.37, No. 1, 1991

ion-exchange interaction between catiomc sites in the mance zone electrophoresis. J Chromatogr 1979;169:11-20.
protein and silicate moieties on the fused-silica surface. 2. Jorgenson JW, Lukacs KD. Zone electrophoresis in open-tubu-
lar glass capillaries. Anal Chem 1981;53:1298-302.
Consistent with the above observations and ration- 3. Jorgenson JW, Lukacs KD. Free-zone electrophoresis in glass
ales, we could achieve a well-resolved electropherogram capillaries. Clin Chess 1981;27:1551-3.
of hemoglobin variants in a buffer of pH 8.6 (the isoelec- 4. Jorgenson JW, Lukacs KD. Capillary zone electrophoresis.
ti-ic point of normal human hemoglobin is 6.7). Variants Science 1983;222:266-72.
of hemoglobin such as hemoglobin S, as in sickle cell 5. Gordon MJ, Huang X, Penteney SL, Zare RN. Capillary elec-
trophoresis. Science 1988;247:224-8.
anemia, and hemoglobin C differ from normal hemoglo- 6. Ewing AG, Wallingford RA, Olefirowicz TM. Capillary electro-
bin by a single amino acid at the 6 position of their phoresis. Anal Chem 1989;61:292A-303A.
beta-chain subunit, which is valine, lysine, and glu- 7. Cohen AS, Terabe S, Smith JA, Karger BL. High-performance
tamic acid, respectively. The isoelectric points of hemo- capillary electrophoretic separation of bases, nucleosides, and
globin S and C are above 6.7, but well below 8.0. Buffer oligonucleotides: retention manipulation via micellar solutions
and metal additives. Anal Chem 1987;59:1021-7.
at pH 8.6 turns out to be an ideal electrophoresis buffer 8. Grossman PD, Colburn JC, Lauer HH, et al. Application of free
for the separation of hemoglobin A, 5, C, and also F. solution capillary electrophoresis to the analytical scale separation
Similarly, in serum protein analysis, the buffer pH is of protein and peptides. Anal Chem 1989;61:1186-94.
adjusted to 10.0, well above the isoelectric points of all 9. Gross L, Yeung ES. Indirect fluorometric detection ofcations in
capillary electrophoresis. Anal Chem 1990;62:427-31.
the serum proteins. Between runs, the capillary column
10. Hjerten 5, Elenbring K, Kilar F, et al. Carrier-free zone
is thoroughly washed and reconditioned to obtain repro- electrophoresis, displacement electrophoresis and isoelectric focus-
ducible results. ing in a high-performance electrophoresis apparatus. J Chro-
matogr 1987;403:47-61.
In conclusion, we have demonstrated the potential 11. Cohen AS, Karger BL. High-performance sodium dodecyl
sulfate polyacrylanude gel capillary electrophoresis of peptides
applicability of capillary electrophoresis to several clin- and proteins. J Chromatogr 1987;397:409-17.
ical diagnostic problems. Automation, quick separation, 12. Terabe 5, Otsuka K, Ando T. Electrokinetic chromatography
and on-line data analysis are the most important at- with micellar solutions and open-tubular capillaries. Anal Chem
tributes of this technique. By using untreated fused- 1985;57:834-41.
13. Terabe S. Electrokinetic chromatography: an interface be-
silica capillary columns, routine analyses of serum pro- tween electrophoresis and chromatography. Trends Anal Chess
teins and variants of hemoglobin can be achieved within 1989;8:129-34.
10 mm without intense labor. Urine and CSF samples 14. Chang SH, Gooding KM. Regnier FE. Use of oxiranee in the
can be analyzed without sample pre-concentration. preparation of bonded phase supports. J Chromatogr
1976;120:321-33.
Above all, multiple samples can be run sequentially in 15. Lauer HH, McMarngill D. Capillary zone electrophoresis of
an automated fashion. proteins in untreated fused silicatubing.Anal Chess
1986;58:166-9.
References 16. Jorgenson JW. Capillary electrophoresis. Chapter 13. In: New
1. Mikkers FEP, Everaerta FM, Verheggen ITEM. High-perfor- directions in electrophoretic methods, ACS Symp Ser 1987;335.

CLINICAL CHEMISTRY, Vol. 37, No. 1, 1991 19