Hua ZhangWenrui Jin

School of Chemistryand Chemical Engineering,and State Key Laboratoryof Microbial Technology,Shandong University,Jinan, China

Determination of different forms of humaninterferon-ª in single natural killer cells by capillaryelectrophoresis with on-capillary immunoreactionand laser-induced fluorescence detection

A novel method for determining different forms of human interferon-g (IFN-g) in singlenatural killer cells was developed by capillary electrophoresis (CE) with on-capillaryimmunoreaction and laser-induced fluorescence (LIF) detection. Cells were perforatedwith digitonin and one single cell was electrokinatically introduced into the front endof a separation capillary. The monoclonal antibody labeled with fluorescein isothiocya-nate of IFN-g was hydrodynamically injected into the front end of the capillary aroundthe cell introduced. After the cell was lysed by ultrasonication, the front end of thecapillary was used as a microreactor to allow different forms of IFN-g to process theimmunoreaction with their labeled antibody. Finally, the complexes of different formsof IFN-g with their labeled antibody were separated and detected by CE with LIFdetection with a limit of detection of zeptomoles (10221 mol).

Keywords: Capillary electrophoresis / Interferon-g / Laser-induced fluorescence / Single-cellanalysis DOI 10.1002/elps.200305803

1 Introduction

Interferon-g (IFN-g) with a molecular weight of 20 000–25 000 is a multifunctional protein with three forms, whichplays an important role in anti-virus infections, anti-prolif-eration of cells, and immunity adjustment [1]. It can beformed in human natural killer (NK) cells, which are im-portant effectors in immunity to tumors, intracellular bac-teria, parasites, and virus infections [2]. However, to datethere is no report on the determination of IFN-g in individ-ual NK cells. Capillary electrophoresis (CE) has manyinherent features of its operation suitable for analysis ofchemical contents inside single cells such as extremelysmall sample size, high separation speed and efficiency,and biocompatible environments. In CE, laser-inducedfluorescence (LIF) detection is the most sensitive. Yeung[3] has summarized the work in the area of single-cellanalysis by CE with LIF detection. The major investiga-tions in CE-LIF detection focused on direct determinationof natively fluorescent substances in individual cells.

However, IFN-g as an antigen is natively nonfluorescent.For determination of antigens and antibodies the mostfrequently used method is immunoassay. Only one reportconsidered the use of immunoassay on basis of on-cap-illary immunoreaction and particle-counting to measureglucose-6-phosphate dehydrogenase (G-6-PDH) in indi-vidual human erythrocytes [4]. In this method, G-6-PDHwithin the cell after lysis reacts with the anti-G-6-PDH-coated latex particles on the capillary for 30 min. Aftercompletion of the immunoreaction, the particles withG-6-PDH migrate electrophoretically in the capillary andpass a detection window where a laser beam irradiatescontinuously. The light scattering events generated bythe agglutinated particles are counted. It can be calcu-lated from the electropherogram in [4] that the number oftheoretical plates of the electrophoretic peak is ,4000,which is too low for separation of structure-similar anti-gens (or antibodies) such as IFN-g. Capillary electropho-retic immunoassay (CEIA) [5–7] offers several advan-tages, such as high selectivity, high separation efficiency,low reagent consumption, and short incubation time. Sofar the technique is not applied to single-cell analysis.

In the present work, we developed a convenient and sen-sitive method at the zeptomole (zmol, 10221 mol) levelwith high separation efficiency for monitoring differentforms of IFN-g in single NK cells by CE with on-capillaryimmunoreaction and LIF detection using a commercialinstrument. One single NK cell previously perforated with

Correspondence: Professor Wenrui Jin, School of Chemistryand Chemical Engineering, Shandong University, Jinan 250100,ChinaE-mail: jwr@sdu.edu.cnFax: 186-531-8565167

Abbreviations: Ab*, anti-IFN-g monoclonal antibody labeledwith fluorescein isothiocyanate; IFN-ª, interferon-g; NK cell,natural killer cell

1090 Electrophoresis 2004, 25, 1090–1095

2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Electrophoresis 2004, 25, 1090–1095 Human interferon-g in natural killer cells 1091

digitonin is electrokinetically injected into the front end ofthe separation capillary and there adsorbed. To separateand determine the different forms of IFN-g in single NKcells, an excess amount of anti-IFN-g monoclonal anti-body labeled with fluorescein isothiocyanate (Ab*) ishydrodynamically introduced into the front end of a sepa-ration capillary around the NK cell. After lysing the cell byultrasonication, the different forms of IFN-g in the cellreact with Ab* on the capillary, forming their antigen-antibody complexes. After completion of the on-capillaryimmunoreaction, the produced complexes as well as thefree Ab* are separated and detected by CE with LIFdetection.

2 Materials and methods

2.1 Apparatus

A P/ACE MDQ CE system (Beckman Instruments, Fuller-ton, CA, USA) equipped with a LIF detector (488 nm LaserModule; Beckman Coulter) was used to perform all sepa-rations and detections. Fluorescence was excited by anargon ion laser at 488 nm (3 mW). The fluorescence lightwas detected after passing through a 488 nm cut-offand a 520 nm interference filter. The fused-silica capillarywith 50 mm ID and 375 mm OD was supplied from Yong-nian Optical Conductive Fiber Plant (Yongnian, China). Allseparations were performed on an untreated capillary of31.2 cm total length and an effective length of 21 cm. Thelength of the coolant tubing was 14 cm. The sample tem-perature of 107C and the coolant of 207C were maintainedthroughout.

2.2 Reagents and solutions

A 50 mg/mL recombinant human IFN-g standard solutionwas provided by the Shandong Academy of MedicalSciences (Jinan, China). Before use, the solution wasdiluted to 5.0 ng/mL with phosphate-buffered saline(PBS), which consisted of 0.135 mol/L NaCl and0.02 mol/L NaH2PO4 – NaOH (pH 7.4). A 0.5 mg/mL Ab*solution was purchased from Pharmingen (San Diego,CA, USA). Before use, the solution was diluted to 5 mg/mLwith PBS. Spermine was purchased from Sigma (St.Louis, MO, USA). A 4.0061022 mol/L stock solution ofspermine was prepared by dissolving an appropriateamount of spermine in water. The CE running bufferconsisted of 1.061022 mol/L Na2B4O7–9.361022 mol/LH3BO3 containing 8.061025 mol/L spermine (pH 8.0).A 4.0061022 mol/L stock solution of digitonin was pre-pared by dissolving an appropriate amount of digitoninin PBS. Other reagents were of analytical grade and pur-

chased from standard reagent suppliers. The water usedwas purified with a commercial system (Pall, East Hills,NY, USA). All solutions were stored at 47C when not in use.

2.3 Preparation of NK cells and their extract

The NK cell suspension was provided by the ShandongAcademy of Medical Sciences (Jinan, China). A 1.0 mLaliquot of cell suspension was centrifuged at 5006g for2 min. The supernatant liquid was removed. In order towash the cells, 1 mL PBS was added. After slight vibrat-ing, the mixture was centrifuged and then the supernatantliquid was removed. This step was repeated three times.Finally, the cell suspension was diluted to 150 mL withPBS. The cell number in the cell suspension was countedusing a hemocytometer (Shanghai Medical Optical Instru-ment Plant, Shanghai, China). A part of the cell suspen-sion was used for single-cell analysis. In order to provethe reliability of the results obtained from single-cell anal-ysis, the different forms of IFN-g in the NK cell extractwere also analyzed. The NK cell extract was prepared asfollows. The cell suspension of 100 mL was mixed with9.9 mL PBS. The cell in the suspension was lysed by ultra-sonication. Then, the solution was centrifuged at 30006gfor 30 min. Its supernatant liquid was centrifuged at15 0006g for 1.5 h again. The supernatant liquid wasthe NK cell extract.

2.4 CE-LIF detection of different forms of IFN-ªafter in-solution immunoreaction

In order to optimize the experimental conditions and iden-tify the different forms of IFN-g in the NK cell extract, a20 mL aliquot of standard IFN-g solution in PBS or the NKcell extract was mixed with a 2.0 mL aliquot of 5 mg/mLAb* in PBS and incubated for 15 min in dark. The solutionwas then diluted to 100 mL with PBS to guarantee the con-centration of IFN-g in the detected sample solution to bein the linear range of the calibration curve. Finally, thedifferent forms of IFN-g were separated and detectedby CE-LIF detection. Before each run, the capillary wasflushed with 0.2 M NaOH for 30 min, water for 5 min, andthe corresponding CE running buffer for 10 min, respec-tively, using 20 psi. Then the current of the electrophoreticcircuit was detected by the P/ACE MDQ CE system,when a voltage of 20 kV was applied across the separa-tion capillary. After the current reached a constant value,the pressure injection of the standard solution or the NKcell extract after the in-solution immunoreaction was car-ried out with a pressure of 0.5 psi. Then, the capillary wasmoved from the standard solution or the extract into theCE running buffer. Finally, the separation voltage of 20 kVwas applied and the electropherogram was recorded.

2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

CE

and

CE

C

1092 H. Zhang and W. Jin Electrophoresis 2004, 25, 1090–1095

2.5 CE-LIF detection of different forms of IFN-ªin the cell extract and single NK cells afteron-capillary immunoreaction

We used on-capillary immunoreaction to determine differ-ent forms of IFN-g in single NK cells, and to measure themin the NK cell extract to prove the reliability of the resultsdetected from single-cell analysis. For CE-LIF detectionof different forms of IFN-g in the cell extract, a 20 mL ali-quot of standard IFN-g solution in PBS or the NK cellextract was mixed with a 2.0 mL aliquot of 5 mg/mL Ab* inPBS and diluted to 100 mL with PBS to allow the concen-tration of IFN-g in the detected sample solution to be inthe linear range of the calibration curve. The solution washydrodynamically injected immediately into the capillaryfilled with the CE running buffer for 3 s using 0.5 psi.Then the solution was incubated for 15 min on the capil-lary. For CE-LIF detection of IFN-g in single NK cells, NKcells were first perforated by mixing the cell suspension of20 mL and 1.061024 mol/L digitonin of 80 mL for 5 min at47C. The single-cell injection was performed using a lab-oratory-built setup described in our previous work [8]. Theinjection setup consisted of a capillary cartridge of the P/ACE MDQ CE system, a buffer reservoir, a cell reservoir,and a high-voltage power supply. A single cell could beinjected into the capillary using an injection voltage of3 kV. Since the cell reservoir was placed on the platformof an inverted biological microscope with a magnificationof 4006, the single-cell injection process could be seen.The cell injected into the capillary was adsorbed on thewall of the capillary 30–90 mm away from its tip. Then1.061024 g/L Ab* in PBS was hydrodynamically injectedwith a pressure of 0.5 psi by using the P/ACE MDQ CEsystem. The capillary cartridge was unloaded from theCE system. The front end of the capillary with the NK celland Ab* around it was inserted into a glass vial with PBS,which had been put in an ice bath placed in an ultra-sonicator. The cell adsorbed on the wall was lysed byultrasonication for 3 s. The capillary cartridge was setinto the CE system again. After incubation for 15 min, thecomplexes of the different forms of IFN-g were separatedand detected by CE-LIF detection with a separation volt-age of 20 kV.

3 Results and discussion

3.1 Optimum conditions of separation anddetection of different forms of IFN-ª afterin-solution immunoreaction

IFN-g has three forms (form I, II, and III) corresponding tothe IFN-g glycosylated at a single site (Asn25), both sites(Asn25 and Asn97), and not glycosylated [1]. Figure 1

Figure 1. Electropherograms of 1.061024 g/L Ab*(1) without and (2) with 1.061026 g/L IFN-g after in-solution immunoreaction. Incubation time, 15 min; CErunning buffer, 5.061023 mol/L Na2B4O7–0.18 mol/LH3BO3 (pH 7.4); injection, 0.5 psi for 3 s; separationvoltage, 20 kV. RFU, relative fluorescence unit.

depicts the electropherograms of Ab* and a solution con-taining 1.061026 g/L IFN-g and 1.061024 g/L Ab* afterthe in-solution immunoreaction for 15 min. The electro-phoretic peak of Ab* appears at ,3.6 min. Only a widepeak is corresponding to the three forms of IFN-g, whichcannot be separated under the conditions. The adsorp-tion of IFN-g on the wall of the capillary should be respon-sible for it. Since spermine can overcome the influence ofprotein adsorption on electropherograms [9], sperminewas added and the influence of spermine concentra-tion on the separation of the three forms of IFN-g wasinvestigated. When the concentration of spermine was8.061025 mol/L, the three peaks corresponding to thedifferent forms of IFN-g could be obtained. However,their separation is not complete. Therefore, pH, concen-tration of CE running buffer, and separation voltageshould be optimized.

When 1.061022 mol/L Na2B4O7–9.361022 mol/L H3BO3–8.061025 mol/L spermine (pH 8.0) as the CE running buf-fer and the separation voltage of 20 kV were used, thethree peaks corresponding to the three forms of IFN-g

2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Electrophoresis 2004, 25, 1090–1095 Human interferon-g in natural killer cells 1093

could be well separated with a resolution of 1.9–2.2.Under these conditions, the electropherogram of stand-ard IFN-g solution containing its three forms is shown inFig. 2, curve 1. Here, the three forms of IFN-g are definedas IFN-g I, IFN-g II, and IFN-g III according to their elutionsequence on the electropherogram. Four peaks, elutingat 2.47, 2.53, 2.59, and 5.59 min, are corresponding toIFN-g I-Ab*, IFN-g II-Ab*, IFN-g III-Ab*, and the free Ab*,respectively. The widths at the peak half-height, W1/2, are2.1, 1.9, 2.0, and 2.4 s, respectively. The numbers of thetheoretical plates, N, reach 104–105. Since the standardIFN-g is a mixture of the three forms of IFN-g, we are notable to identify them positively. Figure 2, curve 2, depictsthe electropherogram of the NK cell extract. Only three

Figure 2. Electropherograms of (1) 1.061026 g/L IFN-g,(2) NK cell extract and (3) (2) 1 1.061026 g/L IFN-gafter in-solution immunoreaction. In-solution incubationfor 15 min with 1.061024 g/L Ab*; CE running buffer,1.061022 mol/L Na2B4O7–9.361022 mol/L H3BO3 con-taining 8.061025 mol/L spermine (pH 8.0); injection,0.5 psi for 3 s; separation voltage, 20 kV. IFN-g I-Ab*,IFN-g II-Ab*, and IFN-g III-Ab* are the complexes ofIFN-g I, IFN-g II, and IFN-g III with Ab*.

peaks, eluting at 2.47, 2.53, and 5.59 min, appeared.No peak at 2.59 min was detected. When the standardIFN-g solution was spiked into the extract, the two peaksat 2.47 and 2.53 min increased, the peak at 5.59 mindecreased and a other peak at 2.59 min correspondingto IFN-g III-Ab* appeared (see Fig. 2, curve 3). Obviouslythe first two peaks obtained in the NK cell extract couldbe the peaks of IFN-g I-Ab* and IFN-g II-Ab*, and the peakat 5.59 min could be responsible for Ab*. Based on theexperimental results, we can conclude that the peak, elut-ing at 2.59 min in the electropherogram of the standardIFN-g solution (Fig. 2, curves 1 and 3) could be corre-sponding to the complex of IFN-g III that is not glyco-sylated with Ab*, because it is absent in NK cells [1].From the area ratios of the three peaks before 3 min, theamount ratios of the three forms of IFN-g with a minordifference of molecular weight can be calculated to be52%, 19%, and 29%, respectively.

3.2 Detection of different forms of IFN-ª in thecell extract after on-capillary immuno-reaction

The results obtained from the cell extract can examine thereliability of the results of single-cell analysis. Therefore,two forms of IFN-g in the cell extract were analyzed usingthe on-capillary immunoreaction. In this case, the sameconditions as in single-cell analysis were used. In thismode, the standard IFN-g solution or the cell extract aftermixing with Ab* was injected into the capillary immedi-ately and was incubated for 15 min. Then the differentforms of IFN-g were separated and detected. The electro-pherograms of the standard IFN-g and the cell extract aredepicted in Fig. 3. The position and peak number on theelectropherograms are the same as those after the in-solution immunoreaction shown in Fig. 2. It can be notedthat the three peaks (with the W1/2 values of 2.6, 2.1, and2.4 s) recorded using the on-capillary immunoreaction(Fig. 3) are wider than those detected (W1/2 = 1.9–2.1 s)using the in-solution immunoreaction shown in Fig. 2.The diffusion of the IFN-g zones to the longitudinal axisof the capillary during the on-capillary imunnoreactioncould be responsible for the broadening. Although thesepeaks broaden, the three peaks corresponding to differ-ent forms of IFN-g are still separated with resolutions of1.5 and 1.6. The peak area can be used for quantification.The concentration LODs of the method are 5.0, 4.3, and3.8610213 mol/L for IFN-g I, IFN-g II, and IFN-g III, whenthe molecular weight of 20 000 is used and the signal-to-noise ratio is 3. Using the calculated injection volumeof 9.3 nL according to the Hagen-Poiseuille equation,the mass LODs of the three forms are 4.8, 4.0, and3.6 zmol, respectively. The concentration linear ranges

2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1094 H. Zhang and W. Jin Electrophoresis 2004, 25, 1090–1095

Figure 3. Electropherograms of (1) 1.061026 g/L IFN-gand (2) NK cell extract after on-capillay immunoreaction.On-capillary incubation for 15 min with 1.061024 g/L Ab*;other conditions as in Fig. 2. IFN-g I-Ab*, IFN-g II-Ab*, andIFN-g III-Ab* are the complexes of IFN-g I, IFN-g II, andIFN-g III with Ab*.

are from their LODs to 1.1, 3.8 and 6.0610211 mol/L,respectively (Table 1). Using least-squares treatment ofthe relationships between the peak areas and the con-centrations, the resulting slopes, y-intercepts, and thecorrelation coefficients are also listed in Table 1. The rela-

tive standard deviations of the method for a series of eightinjections of IFN-g of 5.00610211 mol/L (i.e., the concen-trations of the three forms of IFN-g are 2.60, 0.950, and1.45610211 mol/L, respectively) are between 1.2–1.5%for migration time and 1.4–1.9% for peak area, respec-tively.

The average concentrations of IFN-g I and IFN-g II in thecell extract quantified by the standard addition methodare 5.5610211 mol/L and 5.0610211 mol /L, respectively.To prove the reliability of the method, a certain amountof standard IFN-g was added to the cell extract andthen the cell extract was measured. From the detectedconcentrations in the cell extract with and without thestandard IFN-g, the recoveries can be calculated. Therecoveries of the method for the two forms of IFN-g arebetween 93 and 105%. The total IFN-g concentrationof 1.1610210 mol/L agrees with the average value(9.9610211 mol/L) detected by ELISA. From the cell con-centration of 1.406103 cells/mL in the cell extract, thecalculated mean amounts of the two forms of IFN-g in asingle NK cell are 39 and 36 zmol, respectively, accordingto a molecular weight of 20 000.

3.3 Analysis of single NK cells

Chemical reagents such as NaOH, SDS, and organicsolvents can lyse cells, but can also denature IFN-g.We tried to lyse a single NK cell inside the capillary directlyby ultrasonication. Lysing a cell took . 50 s, making itimpossible to quantify IFN-g because of large disturb-ances, which resulted in a wider electrophoretic peak.Moreover, longer ultrasonication lysis time denaturedIFN-g because of heating. When NK cells were perforatedwith 8.061025 mol/L digitonin in PBS for 5 min at 47Cbefore injection, the lysis time could be shortened to3 s by ultrasonication. Figure 4 depicts the electrophero-grams of three individual NK cells obtained using theon-capillary immunoreaction. Three peaks appear on theelectropherograms. Comparing these curves with theelectropherograms of the standard IFN-g and the NK cell

Table 1. Limits of detection, linear ranges, and their statistical treatment of IFN-g

Form ConcentrationLOD(10213 mol/L)

Mass LOD(zmol)

Linear range(mol/L)

Slope(108 RFUa)

min?L?mol21)

y-intercept(1023 RFU min)

Correlationcoefficient

IFN-g I 5.0 4.8 5.0610213–1.1610210 4.34 0.054 0.9997IFN-g II 4.3 4.0 4.3610213–3.8610211 4.24 20.030 0.9993IFN-g III 3.8 3.6 3.8610213–6.0610211 4.28 20.021 0.9995

a) RFU, relative fluorescence unit. Conditions as in Fig. 3

2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Electrophoresis 2004, 25, 1090–1095 Human interferon-g in natural killer cells 1095

extract (Fig. 3) under the same conditions, two peaks,eluting at 2.47 and 2.53 min, can be identified as thepeaks of IFN-g I-Ab* and IFN-g II-Ab*, based on themigration time. The reproducible peak areas, together

Figure 4. Electropherograms of three single NK cellsafter lysis and on-capillary immunoreaction. Conditionsas in Fig. 3.

Table 2. Contents of IFN-gI and IFN-gII determined in 11single NK cells (zmol) after on-capillary immu-noreaction with Ab*

Cell Amount of IFN-gI Amount of IFN-gII

1 50 522 29 313 7.5 7.64 58 595 9.2 8.96 32 327 10 9.38 53 549 29 30

10 51 5111 29 29

Conditions as in Fig. 3

with the large linear dynamic range for standard IFN-g,made it suitable to use external standardization for thequantification of both in individual NK cells. Table 2 liststhe amounts of IFN-g I and IFN-g II determined consecu-tively in 11 individual NK cells. The amounts of IFN-g Iand IFN-g II were between 7.5–53 zmol and 7.6–54 zmol,respectively, when the molecular weight of 20 000 wasused. The corresponding values in the NK cell extract(39 zmol for IFN-g I and 36 zmol for IFN-g II) are withinthe range of the values found in the single-cell analysis.It can be noted that the amounts of IFN-g I and IFN-g IIin every cell were very close.

4 Concluding remarks

We present a novel technique with high sensitivity andseparation efficiency for separation and determination ofdifferent forms of IFN-g in single NK cells. Detection limitsin the zmol range could be achieved. Two forms of IFN-gpresent in NK cells can be separated and detected. Thismethod is simple and potentially applicable to determineantigens or antibodies in single cells by using commercialCE instruments. Since antigen- or antibody-labeled fluo-rescent reagents are commercial available, the method ishighly practicable. CE with on-capillary immunoreactionand CE-LIF detection proved to be a useful tool for sin-gle-cell analysis.

This project was supported by the National NaturalScience Foundation of China (No. 20235010). The authorsthank Professor Haiming Wei in Shandong Academy ofMedical Sciences for assistance in preparing the IFN-gstandard solution.

Received July 20, 2003

5 References

[1] Sun, W., Wang, H., Cytokine Research Methodology, PeopleHealth Press, Beijing 1999.

[2] Hussell, T., Openshaw, P. J. M., J. Gen. Virol. 1998, 79, 2593–2601.

[3] Yeung, E. S., J. Chromatogr. A 1999, 830, 243–262.

[4] Rosenzweig, Z., Yeung, E. S., Anal. Chem. 1994, 66, 1771–1776.

[5] Bao, J. J., J. Chromatogr. B 1997, 699, 463–480.

[6] Schmalzing, D., Nashabeh, W., Electrophoresis 1997, 18,2184–2193.

[7] Schmalzing, D., Buonocore, S., Piggel, C., Electrophoresis2000, 21, 3919–3930.

[8] Zhang, H., Jin, W., Electrophoresis 2004, 25, 480–486.

[9] Legaz, H. H., McManigill, D., J. Chromatogr. A 1996, 719,159–170.

2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim