REPRODUCIBLE AND EFFICIENT SEPARATION OF AGGREGATABLE ZEIN

PROTEINS BY CAPILLARY ZONE ELECTROPHORESIS USING A VOLATILE

BACKGROUND ELECTROLYTE

Guillaume L. Erny a, Maria Luisa Marina b and Alejandro Cifuentes a,*

a Institute of Industrial Fermentations (CSIC), Juan de la Cierva 3, 28006 Madrid, Spain

b Department of Analytical Chemistry, Faculty of Chemistry, University of Alcalá, Ctra.

Madrid-Barcelona Km. 33.600, 28871 Alcalá de Henares (Madrid), Spain.

*Corresponding author: Dr. Alejandro Cifuentes, Fax#: 34-91-5644853, e-mail:

acifuentes@ifi.csic.es

Running title: CZE of zein proteins

Abbreviations: DMA, N,N-dimethylacrylamide; EpyM, ethylpyrrolidine methacrylate.

Keywords: ammonium gradient, coating, corn, protein separation,

1

Abstract

Zein proteins are a complex mixture of polypetides that belong to the alcohol soluble storage

proteins group (prolamines) in corn. These proteins constitute about 50-60% of the total

endosperm protein and are classified in different groups on the basis of differences in their

solubility and sequence. Among them, zein proteins are considered the majority group

showing a high tendency to aggregate what makes very difficult their analysis by any

analytical method. Thus, capillary zone electrophoresis (CZE) of these proteins require the

use of very complex background electrolytes (BGEs) non-compatible with on-line ESI-MS

analysis. The aim of this work was to find a new BGE for the CZE separation of zein protein

fully compatible with ESI-MS while providing further light on the complex CZE separation

of aggregatable proteins. Thus, it is demonstrated in this work that efficient and reproducible

CZE separations of zein proteins can be achieved by using a BGE composed of water,

acetonitrile, formic acid and ammonium hydroxide. Besides, it is shown that zeins analysis is

significantly improved by including the effect of an ammonium gradient during their

separation. It is experimentally verified that the ammonium gradient can easily be achieved in

CZE by either working with a sample zone with a low concentration of ammonium and a

background electrolyte (BGE) with a high concentration, or conversely, working with a

sample zone with high ammonium concentration and a BGE with low concentration of

ammonium, giving rise in both cases to a significant improvement in the CZE separation of

these proteins. It is demonstrated that this procedure can give rise to efficiency improvements

of up to 20-folds in the CZE separation of zein proteins. Under optimized conditions, 20

proteins could be separated with average efficiencies higher than 400000 theoretical plates/m.

Some possible explanations of this effect are discussed including stacking, protein-capillary

wall adsorption, protein solubility and protein-salt interactions.

2

1. Introduction

Maize (Zea mays L) is the second most important crop in the world, with an annual

production of 705 million tones in 2004. It is expected that by 2020 the worldwide demand

for maize will exceed that for wheat1. The former will require the adoption of superior

varieties, improved management strategies, and where appropriate, new technologies in order

to optimize maize production. Quality of maize flour can be determined, among other

procedures, by evaluating the quantity and quality of its proteins [1].

The alcohol soluble storage proteins (prolamines) of corn are a mixture of polypetides that

constitute about 50-60% of the total endosperm protein [2]. These proteins are classified in

different groups on the basis of differences in their solubility and sequence [3-5] among

which α-, - and -zeins are considered the majority group. The function of these proteins is

apparently to store nitrogen for the developing seed. The multiple applications usually given

to these proteins show their considerable interest in many fields [6].

The composition of zein proteins is usually characterized by sodium dodecyl sulfate–

polyacrylamide gel electrophoresis (SDS-PAGE) based on their molecular weight (Mw) [7].

However, SDS-PAGE exhibits poor resolution and low accuracy in Mw assignment. More

recently, matrix-assisted laser desorption ionization mass spectrometry (MALDI-MS) [8] has

been used, demonstrating that this procedure provides much better information in terms of

mass accuracy. However, the complexity of these zein fractions makes the direct analysis of

all the proteins that can be found a very difficult task, especially those at low concentration.

1 http://www.cimmyt.org/

3

High performance liquid chromatography (HPLC) and capillary electrophoresis (CE) have

also been used to separate zein proteins [9]. Basically, peptides and proteins can be separated

by their affinity to a stationary phase in HPLC or by their electrophoretic mobility in CE

[10]. Although these approaches provide excellent resolution, the absence of information

concerning the relative molecular mass of the separated proteins is an important limitation. In

this regard, coupling liquid separation techniques (HPLC or CE) with mass spectrometry can

be an interesting strategy to overcome this limitation. Thus, CE-MS (or HPLC-MS) allows

combining the high separation power of modern analytical separation techniques with the

information provided by MS about the relative molecular mass of analyte (or even its

fragmentation pattern). Numerous reviews dealing with the application of LC-MS [11,12] and

CE-MS [13-16] to protein analysis have been published in recent years. However, to our

knowledge, neither LC-MS nor CE-MS methods have been developed to analyse zein

proteins. This might be due to the high tendency of zein proteins to interact with each other

forming aggregates [17-20] that obligates to use background electrolytes (BGEs) non-

compatible with the subsequent ESI-MS step. Indeed, all published separations of zein

proteins were performed using chaotropic species (urea [21,22], glicine [23]) and/or

surfactants (SDS [24]) making the separation step highly incompatible with the subsequent

MS analysis.

The aim of this work was, therefore, to find a new BGE for the CE separation of zein proteins

fully compatible with a subsequent ESI-MS coupling, while providing further light on the

complex CZE separation of aggregatable proteins..

2. Materials and methods.

4

2.1 Chemicals.

Sodium hydroxide, formic acid and 25% ammonium were from Merck

(Darmstad, Germany). Sodium dodecyl sulfate and acetonitrile were from

Fluka (Seelze, Germany) and 2-propanol, HPLC grade, from Sharlau

(Barcelona, Spain). The zein protein mix (Zein F4000, purity 92%) was

purchased from Freeman Industries LLC (Tuckahoe, NY). Water was

deionized with a Milli-Q system (Millipore, Bedford, MA, USA). The

ethylpyrrolidine methacrylate :N,N-dimethylacrylamide (EPyM:DMA) copolymer used as

physically adsorbed cationic coating [25] have been synthesized in-house as previously

described [26].

2.2 Preparation of the background electrolyte.

All BGEs were prepared using a 20 ml volumetric flask. BGEs used to

study the effect of ACN content (see section 3.1 below) were made by

firstly adding the ACN content (12 ml of ACN for a 60% ACN BGE for

example), then 4 ml of formic acid (2% formic BGE), and to complete with

water till 20 ml. BGEs used to study the effect of ammonium content (see

section 3.2 below) were prepared using the precedent BGE and by adding

the required amount of 25% ammonium solution until reaching the

required concentration.

2.3 Capillary electrophoresis.

5

CE experiments were performed using a P/ACE 2050 CE instrument from Beckman

(Fullerton, CA, USA). The capillaries used were bare fused silica with 50 μm id and 363 μm

od (Composite Metals, Worcester, UK.), with 370 mm of total length (300 mm detection

length). Before first use, capillaries were conditioned using a 20 min rinse with a washing

solution composed of 1% SDS (w/v) in a 0.1 M NaOH aqueous solution. Between injections,

the capillary was flushed with the washing solution for 1 min and BGE solution for 1 min. To

obtain the EPyM:DMA coated capillary, after flushing the capillary with the washing

solution, the capillary was washed for 1 min with a solution containing 0.1 mg of the polymer

in 1 ml of water.

Sample injections were performed using a pressure of 3.5 Pa. The wavelength was set at 200

nm. The ramp time of the voltage (time to go from zero to the marked value) was set at 10 s.

The capillary was thermostated at 25C. With the bare silica the separation was performed at

20 kV, and at -20 kV with the EPyM:DMA coated capillary. Electrophoretic peak analysis

was carried out using a commercial peak fitting software (PeakFit version 4, Systat,

Richmond, USA) following a procedure already described [27]. A Gaussian function was

used as fitting function.

The migration time repeatability of zein proteins was measured under optimized conditions

(see below) after six successive injections using the washing routine between runs described

above, obtaining RSDn=6 values up to 0.8%. Interestingly, a classical 0.1M sodium hydroxide

washing routine was also used between injections under identical conditions giving RSDn=6

values up to 2.1%, demonstrating the better capability of the SDS-NaOH dissolutions to

regenerate the capillary wall in CZE when working with proteins.

6

2.4 Proteins extraction.

Proteins from grounded maize were extracted using a similar procedure as the one described

by Bean and Lookhart [23]. Briefly, 50 mg of maize were extracted with 1 ml of a buffer

composed of 60:5:35 (v:v:v) ACN:2-mercaptoethanol:water. The sample was vigorously

shaken for 5 minutes and centrifuged at 7000 rpm for 5 min. After centrifugation, part of the

supernatant was kept for injection as it is. The proteins were then precipitated by adding

acetone at a ratio 1:8. After 30 minutes, the sample was centrifuged, the supernatant was

discarded, and the pellet containing the proteins air dried. The proteins were then re-dissolved

in the adequate solvent (either 60:2:38 (v:v:v) ACN:formic acid:water or 60:2:38 (v:v:v)

ACN:formic acid:water with NH4OH at a final concentration of 134 mM) with a

concentration factor of 2 (i.e. 0.2 ml of extraction buffer were mixed with 1.6 ml acetone, the

pellet was dissolved in 0.1 ml solvent). Before injection the samples were kept frozen at -

20ºC.

3. Results and discussion

The multiple BGEs used in this work were solely composed of water, formic acid,

ammonium hydroxide and acetonitrile (ACN). The choice of these substances was based on

the following points: i) zein proteins are only soluble in organic mixture and good solubility

and separations were achieved in 40 to 60% ACN [23] and ii) all these compounds are MS

compatible. As a strategy, it was decided to use a BGE containing a high concentration of

formic acid (2%, ~530 mM). Such BGE has already been successfully used in CE-MS [28,

29] and will provide a low pH even at high organic content, providing a good ionization of

the zein proteins whose pI values are between 5.5 and 8.5 [30], and minimizing electrostatic

7

interactions between the silica wall and the charged proteins. The BGE was then optimized

by adjusting the organic content to obtain good solubility, and by adding ammonium

hydroxide to adjust pH and ionic strength.

3.1 Effect of the acetonitrile content.

In this first step, the effect of the amount of ACN used in the BGE on the CE analysis of

zeins was studied. All BGEs were composed of ACN:formic acid:water (v:v:v) at percentages

given by x:2:(100-x-2), where x was varied between 20 and 80 (zein protein are insoluble

below 20% ACN in this condition). It has to be noted that at this low pH (below 3) the high

concentration of [H+] increases the buffer capacity of this BGE [31]. Samples were prepared

by diluting the required amount of zein proteins mix in the prepared BGEs (1 mg of proteins

per ml of BGE). Between each run, the capillary was washed with the washing solution to

regenerate the capillary wall [32].

The effect of the percentage of ACN can be observed in Figure 1. Figure 1 shows the

obtained electropherograms at 30 (A), 40 (B), 60 (C) and 70 % (D) ACN. At 20 % ACN, the

electropherograms exhibit (date not shown) only two broad peaks. At 80 % ACN an

important change is observed in the electrophoretic pattern with an abrupt decrease in the

electrophoretic mobility of all peaks (data not shown). As it can be observed, from 30 to 60

% ACN, a group of six peaks can be observed (peaks 1 to 6 in Figures 1A to 1C). At 60 and

70 % an additional peak can be observed (peak 7 in Figures 1C and 1D) whose velocity is

very close to the electroosmotic flow velocity. However, since this peak was not detected at

280 nm, it seems that this peak does not correspond to any protein.

8

Whereas in Figures 1A to 1C similar patterns are observed, Figure 1D shows a peak area

decrease, despite the lower BGE viscosity (see Table 1), which might indicate a decrease in

the zein proteins solubility. This Figure is also characterized by broader peaks. From 30 to 60

% ACN, the main difference between the three electropherograms is the shoulder before peak

1 (indicated by an arrow). This shoulder decreases as the ACN content increases, and

disappears at 60 % ACN (Figure 1C). Whereas the migration of zein proteins seems to occur

in a similar way from 30 to 70% ACN, this is not the case at lower or higher organic content.

This is probably due to changes in their solubility.

Figures of merits reported in Table 1 correspond to peak number 2 in Figure 1. Also, some

parameters related to the BGE and CE conditions (including EOF mobilities, pH values (see

below), electrical current, etc) are given. In Table 1, corresponds to the pH of the BGE

measured using the electrode calibrated in aqueous solution, whereas corresponds to the

pH that would have been obtained calibrating the electrode in acetonitrile-water mixtures. In

case of water-ACN solutions (less than 60% ACN), can be estimated from by

[33],

(1)

where xACN is the molar fraction of acetonitrile in the mixture. In this equation, the liquid

junction potential is assumed to be the same in every binary mixture. Although this is an

approximation, the error when using a 1M KCl junction is below 0.1 pH units [33], which is

acceptable in our case. The current in Table 1 is the one recorded by the CE instrument

9

during the separation and the EOF mobility was measured using acetone as a neutral marker.

These values are discussed below.

As can be seen in Table 1, the efficiency (N) obtained for peak 2 is very low and ranges from

1000 to 6000 theoretical plates, with an optimal value at 40% of ACN (Figure 1B). The

was found constant and equal to 2.4 ± 0.1. However, after correcting this value using

Equation 1, the was found to depend on the ACN content. The important variations

detected in the electrical current (28.8 A at 20% ACN, but 4.4 μA at 60 % ACN) confirmed

this dependence. Indeed with a BGE composed mainly of formic acid, the electrical current is

carried by the formate ions and the ions resulting from the auto-dissociation of the solvents

(water and, in a less extent, ACN), and, therefore, the current is expected to have a strong

dependence on 10-pH as observed here. The EOF mobility was also found to depend on the

ACN content. From 20 to 40% ACN, the EOF mobility was constant and equal to 1.023 ±

0.003 (n = 3) ×10-8 m2 V-1 s-1, and then increased (1.20 at 50%, 1.52 at 60% and 1.59 at 70%)

until reaching 2.73 ± 0.05 (n = 3)×10-8 m2 V-1 s-1 at 80% ACN.

The apparent electrophoretic mobility of the main protein peak (peak 2 in Figures 1A to D)

was also measured and after correction with the EOF mobility, a linear dependence of the

mobility on the ACN content was observed. The mobilities range from 0.29 at 20% ACN to

1.24×10-8 m2 V-1 s-1 at 70% ACN. However, at 80% ACN, peaks migrate with the EOF. For

small ions, it has already been shown that the inverse of the effective mobility, assuming fully

charged species, depends linearly on τ/εη, where η, τ and ε are the solvent viscosity, the

Debye dielectric relaxation time and the permittivity, respectively [34,35]. However, such

general equation does not apply to proteins, since the ACN content will modify not only the

10

pKa of the various acidic and basic sites [36] but will also modify the three dimensional

structure of the zein proteins [37, 38] and thus their electrophoretic mobilities.

Experiments were next performed at 60% and 40% ACN with lower formic concentrations.

Although a lower current was observed and, therefore, a faster EOF obtained, no significant

differences to those shown in Figure 1 were observed in the separation. Since the use of

formic acid allows the use of ammonium hydroxide, controlling better the pH and improving

the buffer capacity of the BGE, all the following experiments were done using 2% formic

acid. The ACN content was fixed at 60% considering the CE separations obtained so far and

assuming that this particular percentage of ACN is the optimal to extract zein proteins from

grounded maize [23].

3.2 Effect of the ammonium concentration.

To study the effect of the concentration of ammonium ions, various volumes of 25%

ammonium hydroxide were added to a BGE composed of 60:2:38 (v:v:v) ACN:formic

acid:water to achieve a final concentration of x mM NH4OH. As before, the current and

of the different BGEs were recorded, and the EOF and apparent mobility of the main protein

peak were measured. Those values are resumed in Table 2. The samples were prepared as

above dissolving 1 mg of zein proteins in 1 ml of the different running BGEs. As expected,

the conductivity increases in a linear manner with the ammonium concentration. Thus, a

current of 4 μA was measured at 0 mM NH4OH, 26 μA at 67 mM NH4OH and 47 μA at 134

mM NH4OH (see Table 2). Some selected electropherograms are displayed in Figure 2. As

can be seen, at 0 (Figure 1C), 13 (Figure 2A), 27, 67 and 134 mM NH4OH (Figure 2D)

11

similar electropherograms were obtained with a small shift of the apparent mobility of the

main protein peaks (peak 2) that can easily be explained through increase of and ionic

strength. However noticeable changes can be observed in peak 7, whose electrophoretic

mobility increased with the pH (-0.03 and 0.14 ×10-8 m2 V-1 s-1 at 13 and 137 mM NH4OH,

respectively). A partial separation in this peak was observed at higher pH (comparing Figure

2A and 2D for example). On the whole, it should be noted that upon addition of NH 4OH the

overall separation degrades: whereas in Figure 1, 6 to 7 peaks were observed at 30, 40 and

70% ACN, in Figure 2 only three broad peaks are observed in the best case (peak 1, 2 and 7).

However, at 40 mM NH4OH (Figure 2B) a different pattern is observed with a main single

peak closes to the EOF, and a very important fronting. An evidence of peaks 1 and 2 can be

observed (peak 2) but at a lower intensity. At 53 mM NH4OH (Figure 2C) and 134 mM

NH4OH (Figure 2D) the electropherogram is similar to the one obtained at lower

concentration (Figure 2A), but significant differences can still be observed, such as a higher

effective mobility (see Table 2) and an additional peak marked as X. From the results given

in Figure 2B it could be though that the separation pH under these conditions is very close to

the isoelectric point of these proteins what could explain this strange behaviour. However, the

high pI values for these proteins (between 5.5 and 8.5 [30]), together with the low pH values

of the BGEs indicated in Table 2 (lower than 4.1), seem to contradict this hypothesis, In any

case, the CE separation of these zein proteins needs to be improved further as can be deduced

from the poor separations shown in Figures 1 and 2.

An interesting phenomenon was observed when the ammonium concentration in the sample

zone was different from that in the BGE (see Figure 3). Separations of Figures 3A to 3D were

12

obtained under the same conditions that Figures 2A to 2D but dissolving the proteins in the

former case in 60:2:38 ACN:formic acid:water without ammonium. Thus, although no

significant differences are observed using low ammonium concentrations in the BGE (see

e.g., Figure 2A and Figure 3A both with 13 mM NH4+ in the BGE), using higher ammonium

concentrations (see e.g. Figure 2D and Figure 3D both with 134 mM NH4+) a clear

improvement in the separation is observed when the proteins are dissolved in 60% ACN and

2% formic acid without ammonium. As in Figure 2, different patterns are observed at 40 mM

NH4OH (Figure 3B) and 53 mM NH4OH (Figure 3C). However, whereas Figure 3B exhibits

a very large peak, Figure 3C shows symmetrical narrow peaks with very low selectivity. A

comparison between Figure 2D and 3D could suggest that a classical stacking phenomenon

(i.e. field amplified sample stacking [39] or pH mediated sample stacking [40]) is responsible

for the efficiency and resolution enhancement. However, the electropherograms obtained in

Figures 2B, 2C, 3B and 3C suggest that another mechanism occurs. Indeed, assuming a

sample stacking from either ionic strength or pH discontinuity between the sample zone and

the BGE as the main explanation for the good efficiency observed in Figure 3D, a regular

increase of efficiency from Figures 3A to 3D should have also been observed due to a regular

increase in the pH and ionic strength in the BGE (see Table 2) what is not the case.

Additional experiments were carried out to further investigate this effect. The last series of

experiments were repeated using a BGE composed of 40:2:58 (v:v:v) ACN:formic

acid:water, adding ammonium hydroxide to a final concentration of x mM. Proteins were

dissolved in a 40:2:58 (v:v:v) ACN:formic acid:water solvent in this case. By using BGEs

containing NH4OH concentrations varying from 0 to 40 mM (Figure 4A) no significant

changes were observed in the electropherograms. However, using a BGE containing 67 mM

NH4OH a single broad peak was obtained, while the electropherogram obtained at 40% ACN

13

with 137 mM NH4OH (Figure 4B) showed a similar pattern to the one obtained at 60% ACN

with 53 mM NH4OH (see Figure 3C), what could indicate some residual electrostatic effect

between the proteins (carrying positive charge) and the silica surface (carrying a small

negative charge)..

To test whether the silica surface is responsible for this effect, experiments were repeated

using a cationic coated capillary. The capillary was dynamically coated by flushing with a

ethylpyrrolidine methacrylate N,N-dimethylacrylamide (EPyM:DMA) copolymer at a

concentration of 0.1 mg/ml [25]. This coating should provide a strong electrostatic repulsion

between the capillary wall and the proteins, since all of them are positively charged at this pH

range [41]. Figure 4C shows the separation obtained in a EPyM:DMA coated capillary when

the BGE is composed of 60:2:38 (v:v:v) ACN:formic acid:water with 134 mM NH4OH, and

the proteins dissolved in 60:2:38 (v:v:v) ACN:formic acid:water. Due to the strong reverse

EOF at this pH [41], the separation was performed in reverse polarity. As can be seen,

comparing Figures 3D and 4C, a surface effect does not seem to be responsible for the

separation improvement. The main difference between both separations is the reverse

migration order due to the reverse polarity. Differences in resolution can also be observed in

accordance to the well-known EOF effect.

The separation shown in Figure 4D was also performed in the EPyM:DMA coated capillary

and it was achieved using a BGE composed of 60:2:38 (v:v:v) ACN:formic acid:water, but

dissolving in this case the proteins in 60:2:38 (v:v:v) ACN:formic acid:water with 134 mM

NH4OH (i.e., the sample has a much higher ionic strength than the BGE). Surprisingly, the

separation was also very good in this case. The main differences between Figures 4C and 4D

are probably due to the difference of pH between the two BGEs. Figures 1 to 4 show that

14

when the zein proteins are injected in BGE, relatively poor separations are achieved.

However, if an ammonium concentration step exists between the sample zone and the BGE, a

significant improvement in the separation can be observed (see Figures 3D, 4C and 4D). Any

classical stacking-like effect does not seem to be the main reason, as, first, if the separation

was enhanced due to a stacking effect the separation would improve continuously with the

amplitude of the step increase until reaching a plateau. This behavior was not observed in

Figure 3. Secondly, stacking effects are due to a higher velocity in the sample zone than in

the BGE. If the conditions are reversed, a de-stacking occurs which will cause a poor

separation. However, in the cases described previously, both when the pH and the ionic

strength are lower in the sample zone (Figure 4C), or when the pH and the ionic strength are

lower in the BGE (Figure 4D), a better separation is achieved than when the pH and ionic

strength are the same in the sample and BGE zones. These results clearly indicate that,

whereas stacking phenomena will occur, they are not the dominant effect.

Protein solubility is a complex function of the physiochemical nature of the proteins, pH,

temperature, salt concentration, etc. It also depends on whether the salt is kosmotropic

(stabilizes water structure) or chaotropic (disrupts water structure). In general, the solubility

of proteins increases slightly at low salt concentrations (salting in). But at high salt

concentrations the solubility of proteins drops sharply (salting out) [42]. These effects are

part of what is called the Hofmeister effects [43]. The solubility of proteins as a function of

salt concentration passes by a maximum before decreasing sharply. An increase of protein

solubility is usually linked to the loss and/or change of its secondary structure (the general 3D

shape of the local segment), tertiary structure (the global 3D shape of the protein) and/or

quaternary structure (interaction between several chains of polypeptides) [44,45]. Indeed,

high solubility in general is achieved when there are preferential protein-solvent interactions

15

rather than protein-protein interactions. Curtis and Lue suggested that a preferential protein-

salt binding, as observed during salting-in, would result in effective protein-protein repulsions

[46]. In our case, this effect can also be responsible for the observed improvement in the

separation of zein proteins. Zein proteins are known to interact with each other and to

naturally form dimers, trimers and even more complex aggregates [17-20]. It is therefore

supposed that at the beginning of the separation the concentration of NH4+ will either decrease

(see Figure 4C) or increase (see Figure 4D) in the sample zone and the protein will pass by

their maximum of solubility in which the protein-protein interactions will become minimum.

This should allow, upon the effect of the applied electrical field, to separate aggregates

leading to the observed efficiency improvement. This might also explain the behavior

observed in Figures 2B, 3B, 3C and 4B. This method should be applicable to other proteins

where protein:protein interactions are important. However, more experiments should be done

to further validate this finding.

3.3 Extraction and separation of zein proteins from maize.

To validate our approach using real samples, zein proteins were extracted from maize. The

extraction buffer was composed of 60% ACN, 5% 2-mercaptoethanol and 35% water at a

ratio of 1:20 (50 mg of grounded maize for 1 ml of extraction buffer). After centrifugation,

part of the supernatant is kept for injection as it is. The proteins were then precipitated by

adding acetone at a ratio 1:8. After 30 minutes, the sample was centrifuged, the supernatant

was discarded, and the pellet containing the proteins dried. The proteins were then

redissolved in the adequate solvent (either 60% ACN, 2% formic or 60%ACN, 2% formic

and 134 mM NH4OH) with a concentration factor of 2 (i.e. 0.2 ml of extraction buffer were

mixed with 1.6 ml acetone, the pellet was dissolved in 0.1 ml buffer). Figure 5A shows the

16

electropherogram of the protein extract from corn separated by CZE in 60:2:38 (v:v:v)

ACN:formic acid:water with 134 mM NH4OH. The electropherogram shows important

similarities with the one obtained previously (see Figure 4C). The small differences might be

due to cultivar variations or, more likely, to the different conditions in which zein proteins

were obtained. The observed resolution and efficiency are satisfactory (8 peaks clearly

separated with an efficiency ranging from 10000 to 40000 plates/m, and an average

efficiency of 25000). However, higher resolutions should be expected if a lower EOF could

be used [47].

In our case, lower EOF was achieved by using in the BGE 20% of 2-propanol plus 40% of

ACN as organic modifiers, as it is well known that 2-propanol strongly reduces the EOF due

to its much lower permititivity/viscosity ratio. By using a BGE composed of 40:20:2:38

(v:v:v:v) ACN:2-propanol:formic acid:water with 134 mM NH4OH, the EOF velocity was

reduced by a factor close to 2. The CZE electropherogram obtained under these conditions

can be seen in Figure 5B. Thus, 11 peaks can now be separated with efficiency values ranging

from 10000 to 170000 theoretical plates, and an average efficiency of 80000. Although, in

theory, the EOF could have been further decreased by increasing the 2-propanol content, the

peak shapes are significantly deteriorated over 20% of this organic solvent. Separations were

also performed with an EPyM:DMA coated capillary. As with silica capillary, addition of 2-

propanol has an important quenching effect on the EOF. Figure 5C shows the separation

obtained in this case. Under the same conditions used in Figure 4C, better efficiency and

resolution values were observed with 19 peaks separated and the efficiency ranging from

30000 to 400000 theoretical plates, and an average efficiency of 170000, what supposes more

than a 20-fold improvement compared to the original data obtained in Figure 1. From these

results it can be concluded that better separations were obtained in general using the coated

17

capillary as shown in Figure 5C. This is mainly due to the difference in EOF that allows a

slower separation, but also probably to the diminishing of some residual protein-capillary

wall interactions.

Besides, the volatile nature of this buffer makes it compatible with MS coupling, making

possible the CE-MS analysis of these maize proteins. In order to confirm the proteinaceous

nature of the detected peaks, a new CE-ESI-MS method was developed. Results using this

CE-ESI-MS method demonstrated the proteinaceous nature of the 15 compounds detected

under these conditions with relative molecular mass values ranging from 17240 to 27264 Da.

A detailed description of the optimization of this CE-ESI-MS procedure will be the topic of a

forthcoming paper.

4. Concluding remarks

In this work, we have demonstrated that the use of an NH4+ concentration gradient during the

CZE separation of zein proteins can bring about an improvement in their separation

efficiency of up to 20-folds. It is postulated that this effect can be due to a reduction of

protein:protein interactions, which allows separation of protein aggregates. Such effect has

been observed with zein proteins. However, similar effects might be achievable with other

proteins. It is also anticipated that the adequate type of salt, pH and organic modifier in the

BGE will have to be determined case-by-case, since effects of salts or organic solvents on

protein arrangement are not fully understood yet. These findings, as underlined in this work,

can have important applications for designing better CZE separations of proteins compatible

with ESI-MS analysis while providing better understanding of the proteins

solubility/folding/refolding/aggregation mechanisms.

18

Acknowledgments

G.L.E. thanks the Spanish MEC for a postdoctoral grant. Authors are grateful to the

AGL2005-05320-C02-01 Project (Ministerio de Educacion y Ciencia) and the S-505/AGR-

0153 Project (Comunidad Autonoma de Madrid) for financial support of this work. Authors

also thank the great help obtained from the web site of Martin Chaplin

(http://www.lsbu.ac.uk/water/protein.html). Authors are thankful to L. Sanchez of CSIC

(Madrid, Spain) for providing grounded maize samples.

19

5. References

[1] Guerrieri, N. in Woodhead, R.Y. (Ed.), Proteins in Food Processing, CRC Press¸

Washington, DC, 2004, Chapter 8.

[2] Lookhart, G. L., in: Lorenz, K. J. and Karel, K. (Eds.), Handbook of Cereal Science

and Technology, New York 1991, pp. 441-468.

[3] Essen, A., Plant Physiol. 1986, 80, 623-627.

[4] Essen, A., J. Cereal Sci. 1987, 5, 117-128.

[5] Monamy, F. A., Sessa, D. J., Lawton, J. W., Selling, G. W., et al. J. Agric. Food Chem.

2006, 54, 543-547.

[6] Lawton, J. W., Cereal Chem. 2002, 79, 1-18.

[7] Parris, N., Dickey, L.C., J. Agric. Food Chem. 2001, 49, 3757-3760.

[8] Wang, J-F., Geil, P.H., Kolling, D.R.J, Padua, G.W., J. Agric. Food Chem. 2003, 51,

5849-5854.

[9] Rodriguez-Nogales, J.M., Garcia, M.C., Marina, M. L., J. Sep. Sci.2006, 29 ,197-210.

[10] Kasicka, V., Electrophoresis 2006, 27, 142-175.

[11] Issaq, H.J., Chan, K.C., Janini, G. M., Conrads, T. P., Veenstra, T. D., J. Chromatogr.

B 2005, 817, 35-47.

[12] Ishihama, Y., J. Chromatogr. A 2005, 1067, 73-83.

[13] Stutz, H., Electrophoresis 2005, 26, 1254-1290.

[14] Huck, C.W., Bakry, R., Huber, L.A., Bonn, G. K., Electrophoresis, 2006, 27, 2063-

2074.

[15] Simó, C., Barbas, C., Cifuentes A., Electrophoresis, 2005, 26 1306-1318.

[16] Hernández-Borges, J., Neusüß, C., Cifuentes, A., Pelzing, M., Electrophoresis, 2004,

25, 2257-2281.

20

[17] Kim, C.S., Woo, Y-M., Clore, A.M., Burnett, R.J., et al. Plant Cell 2002, 14, 655-

672.

[18] Cabra, V., Arreguin, R., Vazquez-Duhalt, R., Farres, A., Biochim. Biophys. Acta

2006, 1764, 1110-1118.

[19] Parris, N., Dickey, L.C., J. Agric. Food Chem. 2001, 49, 3757-3760.

[20] Cabra, V., Arreguin, R., Galvez, A., Quirasco, M., et al. J. Agric. Food Chem. 2005,

53, 725-729.

[21] Righetti, P.G., Olivieri, E., Viotti, A., Electrophoresis, 1998, 19, 1738-1741.

[22] Olivieri, E., Viotti, A., Lauria, M., Simo-Alfonso, E., Righetti, P.G., Electrophoresis,

1999, 20, 1595-1604

[23] Bean, S. R., Lookhart, G. L., Bietz, J.A., J. Agric. Food Chem. 2000, 48, 318-327.

[24] Parris, N., Dickey, L., Craig, J., Cereal Chem. 1997, 74, 766-770.

[25] Simo, C., Elvira, C., Gonzalez, N., San Roman, J., et al. Electrophoresis 2004, 25,

2056-2064.

[26] González, N., Elvira, C., San Román, J., J. Polym. Sci. Part A: Polym. Chem. 2003,

41, 395-407.

[27] Erny, G. L., Bergstrom, E.T., Goodall, D.M. J. Chromatogr. A. 2002, 959, 229-239.

[28] Simo, C., Gonzalez, R., Barbas, C., Cifuentes, A., Anal. Chem.2005, 77, 7709-7716.

[29] Erny, G. L., Cifuentes, A., Electrophoresis 2007, 28, 1335-1344

[30] Consoli, L., Damerval, C., Electrophoresis 2001, 22, 2983-2989.

[31] Beynan, R. J., Easterby, J. S. Buffer solution, BIOS Scientific Publishers, Oxford

(UK), 1996, Chapter 3.

[32] Righetti, P.G., Gelfi, C., Verzola, B., Castelletti, L. Electrophoresis 2001, 22, 603-

611.

[33] Espinosa, S., Bosch, E., Roses, M., Anal. Chem. 2000, 72, 5193-5200.

21

[34] Joyban, A., Kenndler, E., Electrophoresis, 2006, 27, 992-1005.

[35] Roy, K. J., Lucy, C. D., Electrophoresisis, 2002, 23, 383-392.

[36] Porras, S. P., Kenndler, E. Electrophoresis, 2005, 26, 3203-3220

[37] Zhang, L., Zhang, F., Zhang, S.Y., Guo, Y.L., Xu, C.H. Rapid Com. Mass Spec.

2005, 19, 2151-5156.

[38] Kovrigin, E.L., Potekhin, S.A. Biophys. Chem. 2000, 83, 45-59.

[39] Chien, R. L., Burgi, D. S., Anal. Chem.1992, 64, 1046-1050.

[40] Zhao, Y. P., Lunte, C. E., Anal. Chem.1999, 71, 3985-3991.

[41] Erny, G.L., Elvira, C., San Roman, J., Cifuentes, A. Electrophoresis 2006, 27, 1041-

1049.

[42] Melander W., Horvath, C. Arch. Biochem. Biophys. 1977, 183, 200-215.

[43] Kunz, W., Nostro, P.L., Ninham, B.W. Curr. Opin. Colloid. Interface Sci. 2004, 9, 1-

18.

[44] Arakawat, T., Bhat, R., Timasheff, S. N., Biochemistry, 1990, 29, 1914-1923

[45] Mei, G., Di-Venere, A., Buganza, M., Vecchini, P. et al., Biochemistry, 1997, 36,

10917-10922.

[46] Curtis, R.A., Lue, L. Chem. Eng. Sci. 2006, 61, 907-923.

[47] Lucy, C. A., McDonald, T. L. Anal. Chem. 1995, 67, 1074-1078.

[48] Saleh, M. A., Akhtar, S., Ahmed, M. S., Phys. Chem. Liq., 2006, 44, 551-562.

[49] Cunningh, G., Vidulich, G. A., Kay, R. L., J. Chem. Eng .Data 1967, 12, 336-342.

22

Table 1. Electrophoretic figures of merits and solvent physical properties as a function of the acetonitrile content (see text for details).

% ACN xACN

aI

(A)

ηb

(cP)

εc

(10-12 F/m)

μEOF

(10-8m2V-1s-1)

μeff

(10-8m2V-1s-1)

N

(plates/m)

20 0.079 2.35 2.38 28.8 0.87 71.9 1.02 (± 0.01) 0.28 (± 0.05) 500

30 0.130 2.33 2.37 19.4 0.86 68.2 1.02 (± 0.01) 0.49 (± 0.03) 3500

40 0.186 2.39 2.53 12.5 0.81 64.3 1.03 (± 0.01) 0.71 (± 0.05) 6000

50 0.260 2.41 2.63 7.6 0.73 59.6 1.20 (± 0.05) 0.75 (± 0.03) 3000

60 0.339 2.42 2.88 4.4 0.65 55.0 1.52 (±0.12) 0.75 (± 0.07) 2000

70 0.440 2.4 -- 2.3 0.55 49.9 1.59 (±0.14) 0.83 (± 0.09) 1300

80 0.578 2.36 -- 1.2 0.47 44.2 2.73 (±0.22) -- --

a Values calculated using eq. 1b Values extrapolated from [48]

23

c Values extrapolated from [49].

24

Table 2. Electrophoretic figures of merits as a function of the NH4OH concentration (see text for details).

[NH4OH]

(mM)

I

(A)

μEOF

(10-8m2V-1s-1)

μeff

(10-8m2V-1s-1)

N

(plates/m)

0 2.42 2.88 4.4 1.52 (±0.12)a 0.75 (± 0.10)a 2000

13 2.7 3.16 6.4 1.27 (±0.10) 0.79 (± 0.12) 1800

27 2.96 3.42 10.4 1.43 (±0.03) 0.70 (± 0.07) 2150

40 3.1 3.56 15.5 1.37 (±0.05) 0.89 (± 0.03) 1500

53 3.26 3.72 20.5 1.38 (±0.02) 0.94 (± 0.02) 1450

67 3.37 3.83 25.6 1.40 (±0.05) 0.68 (± 0.01) 3500

134 3.63 4.09 46.5 1.38 (±0.01) 0.58 (± 0.02) 2200

a standard deviation values given between parenthesis.

25

FIGURE CAPTIONS

Figure 1. Electropherograms of a standard zein protein mixture separated by CZE using

different BGEs composed of x:2:(100-x-2) (v:v:v) ACN:formic acid:water with x equal to

(A) 30, (B) 40, (C) 60 and (D) 70. The sample (1 mg/ml of a mixture of zein proteins

dissolved in the same BGE) was injected for 6 s at 3.5 Pa. The separation was performed

applying +20 kV using a bare silica capillary of 37 cm (30 to detector) length and 50 μm id,

thermostated at 25C. The detection was performed at 200 nm. Numbered peaks are discussed

in the text.

Figure 2. Electropherograms of a standard zein protein mix separated by CZE using a

60:2:38 (v:v:v) ACN:formic acid: water and addition of (A) 13, (B) 40, (C) 53 and (D) 134

mM of NH4+OH- BGEs. The sample was injected for 6 s at 3.5 Pa and was made of 1 mg/ml

of a standard mix of zein proteins dissolved in BGE. Numbered peaks are discussed in the

text. Other conditions as in Figure 1.

Figure 3. Electropherograms of a standard zein protein mix dissolved in a 60:2:38 (v:v:v)

ACN:formic acid:water and separated in a BGE made of 60:2:38 (v:v:v) ACN:formic

acid:water and (A) 13, (B) 40 , (C) 53 and (D) 134 mM NH4OH. Other conditions as in

Figure 1.

26

Figure 4. Zein standard dissolved in (A and B) 40:2:58 (v:v:v) ACN:formic acid:water; (C)

60:2:38 (v:v:v) ACN:formic acid:water and; (D) 60:2:38 (v:v:v) ACN:formic acid:water with

134 mM of NH4+OH-, and separated in (A) 40:2:58 (v:v:v) ACN:formic acid:water, 40 mM

NH4+OH-, (B) 40:2:58 (v:v:v) ACN:formic acid:water, 134 mM NH4

+OH-, (C) 60:2:58 (v:v:v)

ACN:formic acid:water, 134 mM NH4+OH- and (D) 60:2:58 (v:v:v) ACN:formic acid:water

BGE. In (C) and (D) separations were performed in reverse polarity (-20 kV) in an

EPyM:DMA coated capillary (see section 2 for coating procedure). Other conditions as in

Figure 1.

Figure 5. Zein proteins extracted from ground maize using a 65:5:35 (v:v:v)

ACN:mercaptoethanol:water solvent. Separations performed in (A) 60:2:38 (v:v:v)

ACN:formic acid:water BGE with 134 mM of NH4+OH-, (B) and (C) 40:20:2:38 (v:v:v:v)

ACN:propanol:formic acid:water BGE with 134 mM of NH4+OH-. Separations were

performed either (A) and (B) in direct polarity (20 kV) in a bare silica capillary, or (C) in an

EPyM:DMA coated capillary in reverse polarity (-20 kV). (A) Sample injected after

extraction without any further manipulation, proteins dissolved in (B and C) 60:2:38 (v:v:v)

ACN:formic acid:water after precipitation in acetone. Other conditions as in Figure 4.

27

0

1

3 6 9Time /min

Abs

orba

nce

/10-2

a.u.

12

34

56

A

0

1

3 6 9Time /min

Abso

rban

ce /1

0-2 a.

u.

12

34

5 6

B

0

1

2 5 8Time /min

Abso

rban

ce /1

0-2 a.

u.

1

2

3

4

56

C

7

0

1

3 6 9Time /min

Abso

rban

ce /1

0-2 a.

u.

1

2

34

D

7

Figure 1.

28

0

1

3 5 7

Time /min

Abso

rban

ce /1

0-2 a.

u.

A 2

71

0

2

4

3 5 7

Time /min

Abso

rban

ce /1

0-2 a.

u.

B7

2

0

1

2

3 5 7

Time /min

Abso

rban

ce /1

0-2 a.

u.

C

1

2

7X

0

1

2

3 5 7

Time /min

Abso

rban

ce /1

0-2 a.

u.

D

7

1

2

Figure 2.

29

0

1

4 6 8

Time /min

Abs

orba

nce

/10-2

a.u.

A2 7

1

0

1

2

3 5 7

Time /min

Abs

orba

nce

/10-2

a.u.

B

7

0

1

2

3

3 5 7

Time /min

Abs

orba

nce

/10-2

a.u.

C

7

0

1

2

3

3 5 7

Time /minA

bsor

banc

e /1

0-2 a.

u.

D 7

Figure 3.

30

0

1

4 6 8

Time /min

Abs

orba

nce

/10-2

a.u.

A7

0

2

4

4 6 8

Time /min

Abs

orba

nce

/10-2

a.u.

B

7

0

1

2

5 8 11

Time /min

Abs

orba

nce

/10-2

a.u.

C

7

0

1

2

2 5

Time /minA

bsor

banc

e /1

0-2 a.

u.

D

7

Figure 4.

31

0

1

2

4 5 6Time /min

Abs

orba

nce

/10-2

a.u

. A

0

1

2

3

6 7 8Time /min

Abs

orba

nce

/10-2

a.u

. B

0

1

2

11 13 15 17

Time /min

Abs

orba

nce

/10-2

a.u

.

C

Figure 5.

32