- Email: [email protected]
Capillary electrophoresis of peptides and proteins in acidic, isoelectric buffers: recent developments
J. Biochem. Biophys. Methods 40 (1999) 1–15
Capillary electrophoresis of peptides and proteins in acidic, isoelectric buffers: recent developments a, a a Pier Giorgio Righetti *, Alessandra Bossi , Erna Olivieri , Cecilia Gelfi b a
University of Verona, Department of Agricultural and Industrial Biotechnologies, Strada Le Grazie, Ca` Vignal, 37134 Verona, Italy b ITBA, CNR, Via Fratelli Cervi 93, Segrate 20090 Milano, Italy Received 15 August 1998; received in revised form 23 October 1998; accepted 24 October 1998
Abstract The use of isoelectric buffers in capillary zone electrophoresis is here reviewed. Such buffers allow delivery of very high voltage gradients (up to 1000 V/ cm in relatively large bore capillaries, e.g. 75–100 mm I.D.), permitting separations of the order of a few minutes and thus conserving (in fact favouring) very high resolution due to minimal, diffusion-driven, peak spreading. Isoelectric Asp (pI 2.77 at 50 mM concentration and 258C) provides a medium of high resolving power for generating peptide maps. In difficult cases, of coincident titration curves, the pH can be moved up to higher values (e.g. pH 3.0 for 30 mM Asp) thus eliciting separation of unresolved peptides at pH 2.77. This was illustrated by running peptide maps of tryptic digests of human b globin chains. Also imino diacetic acid (pI 2.33 at 50 mM concentration) allows generation of high resolution peptide maps. Isoelectric Asp, in presence of 7 M urea and 0.5% hydroxyethyl cellulose (Mn 5 27 000 Da) is also the preferred medium for fast separation and analysis of storage proteins in cereals, such as gliadins in soft and durum wheat and zeins in maize. A solution of 50 mM iminodiacetic acid (pI 2.23) containing 7 M urea and 0.5% hydroxyethylcellulose (apparent pH 3.2) is effectively used as background electrolyte for fast separation of heme-free, denatured globin (a and b) chains. In the presence of neutral to neutral amino acid substitutions, it is additionally shown that the inclusion of 3% surfactant (Tween 20) in the sample and background electrolyte induces the separation of the wild-type and mutant chains, probably by a mechanism of hydrophobic interaction of the more hydrophobic mutant with the detergent micelle, via a
Abbreviations: IDA, imino diacetic acid; HEC, hydroxyethyl cellulose; Hb, hemoglobin; CZE, capillary zone electrophoresis; TFE, 2,2,2-trifluoroethanol *Corresponding author. Tel. / fax: 1 39-45-809-8901. E-mail address: [email protected] (P.G. Righetti) 0165-022X / 99 / $ – see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S0165-022X( 99 )00010-X
2
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
mechanism similar to ‘micellar electrokinetic chromatography’. All rights reserved.
1999 Elsevier Science B.V.
Keywords: Isoelectric buffers; Capillary electrophoresis; Proteins; Peptides
1. Introduction Protein chemists have relied on electrophoresis for more than half a century as one of the most valued tools for purification, characterization and analysis of proteins. A variety of electrophoretic techniques have been used in protein separations [1], ranging from (i) sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) and (ii) gel electrophoresis at near-physiological pH to retain native three dimensional structure, to (iii) isoelectric focusing [2] and immobilized pH gradients [3] and (iv) two-dimensional (2-D) gel electrophoresis [4]. A major advantage of gel systems is that it is relatively easy to maintain the integrity of a separation while components are being subjected to further manipulations. Dyes may be infused into gels where the separated proteins are denatured and stained, substrates may be driven into gels for zymogramming, or proteins may be electroblotted into adsorptive membranes for immunological assays. All these operations can be achieved with minimal alteration of zone integrity. Once a protein zone has been located it is even possible to bring this material back into solution for further analysis. Another major advantage of conventional gel-slab electrophoresis is the open nature of the format. The gel acts as both a fraction collector and a transport medium, which, after a separation in one dimension, may be used to carry sample components into another separation dimension without disturbing the pattern previously obtained (e.g. as adopted into 2-D maps). An extra advantage of gel-slab systems is that they allow samples to be profiled, i.e. they produce a picture of all components present in it, particularly in the case of 2-D maps, which can be used to examine extremely complex biological extracts [4]. One major drawback of all gel-slab methodologies, however, is that they are labour intensive and require too many manipulations for obtaining final results: gel preparation, electrophoretic step, staining, destaining and densitometry or photographic recording. The advent of capillary zone electrophoresis (CZE) in the nineties has greatly simplified electrophoretic procedures, by permitting automatic protocols for sample loading and data acquisition and storage [5]. The extra bonus of CZE is also the rapid sample analysis, due to the possibility of delivering high voltage gradients in narrow-lumen separation channels, and the minute sample requirements. Peptide and protein analysis has thus received a new impetus in the CZE format, as documented by several reviews [6–12]. However, even in CZE, peptide and protein separations are besieged by severe problems. The major one resides in the very nature of the fused-silica container, which presents a surface studded with silanols, believed to be present at a concentration of 8 mM / m 2 (corresponding to about five silanols per nm 2 ). Due to their weekly acidic character (pK 5.3 as in Ref. [13], or pK 6.3, as in Ref. [14]) they ionize over a broad pH
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
3
range, from pH 3.0 up to ca. pH 10. The problem with extensive silanol ionization is that the fused-silica column acts as a cation exchanger. Severe band spreading and diminished recovery occur when proteins with an isoelectric point higher than the buffer pH (and thus bearing a net positive charge when bathed in such a buffer) interact with the capillary wall [15]. An additional problem is that adsorbed proteins change the z potential at the capillary wall, impacting both the electroosmotic flow and column efficiency. This is a severe problem, because alterations in electroosmotic velocity change the elution time of all substances and compromise reproducibility. Finally, it has been shown that an axially heterogeneous distribution of z potential can diminish the separation efficiency of a column by triggering non-uniform flow along the length of the capillary. This axially heterogeneous distribution of z potential results from the inlet of the capillary being fouled with sample protein. In order to overcome these problems, a number of authors have adopted, especially in the case of peptide maps, acidic buffers (typically phosphate, formate and glycinate), at pH values ranging from 1.95 up to 2.8, in concentrations ranging from 25 up to 100 mM. CZE, if carried out in an acidic milieu, has some distinct advantages: it can be performed in uncoated capillaries, since at such pH values (typically 2.0–2.5) essentially all silanols of fused-silica are protonated and thus unable to adsorb proteins and polypeptides. However, one drawback still remains: due to the very high conductivity of these buffers, only low voltage gradients can be applied, resulting in long separation times. In order to improve the situation, we have reported separations in acidic, isoelectric buffers (notably aspartic and imino diacetic acids) [16,17]: in such amphoteric buffer solutions, much higher voltage gradients (600–1000 V/ cm) can be delivered, eliciting separations of the order of a few minutes. A number of applications of such technique have already appeared: in the analysis of tryptic digests of a and b globin chains [18], in the separation of gliadins [19], of zeins [20] and of intact globin chains [21,22]. The fundamental properties of these isoelectric buffers have been enunciated [23]. Some reviews have already covered the field [24,25]. The present review will highlight some novel aspect of this unique, most powerful technique.
2. CZE separation of peptides in isoelectric Asp Recently, Righetti and Nembri [16] proposed the use of isoelectric aspartic acid as a background electrolyte, operating at pH 5 pI 5 2.77 (at 258C and 50 mM concentration). These authors could produce peptide maps of casein in only 10–12 min (as opposed to . 70 min in standard phosphate buffer, pH 2.0) at voltage gradients as high as 800 V/ cm, with much increased resolution. Adsorption of some larger peptides to the wall was completely eliminated by adding to the background, isoelectric Asp buffer 0.5% hydroxyethyl cellulose (HEC) and 5% trifluoroethanol (TFE). Fig. 1 shows a nice example of such casein peptide maps, as produced by the standard technique (Fig. 1A) and in the presence of isoelectric Asp (Fig. 1B). However, there are some intrinsic problems with the use of isoelectric buffers. A case in point is shown in Fig. 2, which represents the CZE separation of tryptic digests of b globin chains. At the prevailing pH in isoelectric Asp (pH 2.77) two peaks (labelled T2 and T9) co-elute into a single zone.
4
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
Fig. 1. CZE of tryptic digests of b-casein. Conditions (upper panel): 100 mm I.D., 37 cm long capillary, bathed in 80 mM phosphate buffer, pH 2.0. Sample application: by pressure by applying 0.5 p.s.i. for 3 s. Separations were performed at 110 V/ cm (current: 85 mA) and detection was at 214 nm. The three major peaks, are: (1) pI 6.1, fragment b-CN (114–169); (2), pI 6.93, fragment b-CN (49–97) and (3), pI 3.95, fragment b-CN (33–48). Note that the total running time is . 70 min. B: CZE of tryptic digests of b-casein in a 100 mm I.D., 37 cm long capillary, bathed in 50 mM isoelectric aspartic acid (pH approximating the pI value of 2.77) added with 0.5% HEC (Mn: 27 000 Da) and with 5% trifluoroethanol. Running conditions: 600 V/ cm (current: 58 mA) (from Righetti and Nembri, see Ref. [16], by permission).
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
5
Fig. 2. CZE separation of a peptide digest of b globin chains. Run conditions: 37 cm long (30 cm to detector) capillary, run at 22.2 kV (600 V/ cm) at ca. 25 mA. Buffer: 50 mM Asp, pH 2.77, added with 0.5% HEC, 5% TFE and 1% CHAPS. Sample injected for 15 s by hydrostatic pressure. Note that, at the operative pH (2.77), the two fragments T2 and T9 merge into a single peak. The inset on the right shows an enlargement of the boxed area on the left side (transit times from 2.5 to 5.3 min) with proper peak identification (from Capelli et al., see Ref. [18], by permission).
By simulating the theoretical pH / mobility curves of the 13 peptides in the pH 2–4 interval, we could determine that indeed at this pH value there was a cross-over point in the titration curves of these two peptides, which could only be eliminated by working at slightly different pH values, such as pH 3.0 or 3.1. However, if one were to work at this higher pH values with a conventional, non-amphoteric buffer, one would automatically lose the benefit of adopting high voltages and thus considerably shorten the separation time while gaining in resolving power. It should be noted, however, that, as amply illustrated by Bossi and Righetti [17] in the case of IDA, and as anticipated by Righetti and Nembri [16] (see their Fig. 5), the pI of 2.77 is only a theoretical value, reached at high enough concentration of Asp (in this case 50 mM). By progressive dilution of the buffering ion, one can span a small pH interval, up to 0.4 pH units. Thus, it is possible, while still working under isoelectric conditions, to modulate the pH of the background electrolyte simply by varying its concentration. In a buffer composed of 30 mM Asp (still ensuring adequate buffering power), the final pH of the background electrolyte is
6
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
Fig. 3. CZE separation of a peptide digest of b globin chains. Run conditions as in Fig. 2, except that the buffer was: 30 mM Asp, pH 3.0, added with 0.5% HEC, 10% TFE and 50 mM NDSB-195. Sample injected for 15 s by hydrostatic pressure. Note in the inset (an enlargement of the boxed area on the left side) the full splitting of peaks T2 and T9, unresolved in the buffer of Fig. 2 (from Capelli et al., see Ref. [18], by permission).
3.0. When adopting this last buffer (see Fig. 3), it was possible to separate all the 13 b globin peptides. By enlarging the 3–6 min separation window (see Fig. 3, insert) one can indeed appreciate that now the T2 and T9 b-peptides, which co-migrated into a single zone, are base-line resolved. However, as the pH increased to 3.0, there was a higher risk of peptide adsorption to the naked silica wall (only uncoated capillaries are used throughout) and addition of a zwitterion (non-detergent sulphobetaine, NDSB-195) was necessary in order to quench such interaction.
3. CZE separation of proteins in isoelectric buffers The situation with proteins is, of course, more complex: it had first to be demonstrated that these macromolecules would not bind to the silica surface, given the fact that already some large peptides did. In fact (cf. Fig. 1B) already the large casein fragments labelled pI 3.95, pI 6.1 and pI 6.93 gave tailing zones, suggesting adsorption to the few
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
7
ionized silanols at the operative pH of Asp (pI 2.80). An excellent cure to this noxious phenomenon was found with the addition of 0.5% hydroxyethyl cellulose (HEC, average Mn 5 27 000 Da), which probably offers a dynamic coating of the wall. HEC was thus adopted in all instances of separations in acidic, isoelectric buffers. The second problem regards the integrity of proteins at such acidic pH values: it is to be expected that, in general, proteins will be denatured in such a milieu; the result could be exposure of hydrophobic residues, intra-chain interaction, aggregation and precipitation. Thus, our technique cannot quite be recommended for analysis of native proteins, but could be a still valid option for denatured polypeptides. There is at least a very important class of proteins to which our technique could apply: the storage proteins (prolamins) or cereals. Knowledge of cereal proteins is valuable in many ways. For example, because of their heterogeneity and nearly invariant expression, storage proteins provide fingerprints that differentiate cereal genotypes and cultivars. This ability to recognize cultivars is important during breeding, marketing and in research. Through protein analysis, breeders can identify and select optimal parental genotypes and progeny during cultivar development and can ascertain varietal purity. In production agriculture, varietal identification can ensure use of lines with optimal economic return, quality, yield, genetic resistance or adaptability. Millers, bakers, brewers, maltsters and other processors must guarantee that grains and flours are suitable for high quality food and on-food products. Prolamins (called gliadins, in wheat, zeins in maize, hordeins in barley, secalins in oats) are typically synthesized and packaged in the endosperm as storage granules, insoluble in water, salt and buffers solutions. Prolamins are soluble in aqueous alcohol blends and, after extraction, can be efficiently analyzed by CZE in 7–8 M urea solutions. Extensive work in the optimization of CZE profiling of cereal prolamins is already available (for a review, see Ref. [25]), although it is typically performed in high conductivity buffers, such as 100 mM phosphate, pH 2.5. Just as reported by Bean et al. [25], in a 50 mm I.D. capillary it is impossible to obtain a separation in less than 30 min in the phosphate background electrolyte and the run is often plagued by base-line disturbances. On the contrary, when the same separation is performed in our isoelectric Asp buffer, in the presence of 7 M urea and 0.5% HEC, the run is typically over in 10 min (at a voltage gradient of 1000 V/ cm), if one does not consider some minor, late-eluting peaks in the 13–14 min time window and present only in some cultivars (Fig. 4A and B). In Fig. 4A, a series of overlaid, individual runs of four different varieties of durum wheat is given: it is seen that the pattern is highly specific and that the different cultivars can be recognized at a glance. Fig. 4B shows a spectrum of gliadin bands resolved in a total of four different cultivars of bread wheat. Here too the overlaid tracings show that such patterns are indeed cultivar-specific. In both Fig. 4A and B, the fact that the peaks are quite sharp and symmetric, and that the base line, after passage of the whole train of zones, returns to the initial value, suggests that there is no binding of these proteins to the capillary wall. The absence of sharp spikes in the UV tracings also indicates the absence of precipitated material. Fig. 5 shows the reproducibility of our separation system: it represents the overlay of 20 consecutive runs, taken every fifth run. It is seen that there is a tendency for the whole train of peaks to lengthen a bit the transit times in the repeated runs. The overall difference between run No. 1 and No. 20 is 4% (in absolute migration times, not relative
8
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
Fig. 4. Representative CZE run of gliadins, obtained by sequential extraction of flour, from four different cultivars of durum wheat (A) and four different cultivars of bread wheat (B). Runs performed in 40 mM Asp buffer, 7 M urea and 0.5% HEC (apparent pH: 3.9). Conditions: 50 mm I.D., 30 cm long capillary, run at 1000 V/ cm at room temperature. Detection at 214 nm (from Capelli et al., see Ref. [19], by permission).
or normalized migration times), measured on the highest peak eluting around 7.5 min. This is in agreement with the data of Bean et al. [25], who reported a non-linear relationship between transit times and run number. Also the reproducibility in peak areas (again measured on the largest peak at 7.5 min) is quite acceptable: the overall variation is never greater than 62%, averaged over the 20 runs. Fig. 6 shows a representative pattern of a maize line (A69Y 1 ) run in 40 mM Asp, 8 M urea and 0.5% HEC, at 700 V/ cm. Here too one can appreciate a train of sharp peaks, quite symmetric, also suggesting absence of interaction with the wall silanols. At a first glance, it sounds odd that zeins should elute, in a typical electropherogram, in a total of 30 min transit time, although the analysis is performed at 700 V/ cm. When screening wheat gliadins by the same technique, the entire pattern was typically developed in 10 min (at 1000 V/ cm). The reason for that might reside in the unique amino acid composition of zeins: on an average mass of 20 kDa, zeins typically exhibit the following average basic amino composition: one Lys, two His and two Arg residues. On the contrary, they appear to have a vast number of acidic residues: an average of 40 Glu and nine Asp. However, due to the fact that their pI values extend across neutrality, it would appear that ca. 90% of these residues exist, in the native polypeptides, as Gln and
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
9
Fig. 5. Reproducibility of a series of 20 consecutive CZE runs of gliadins from the durum wheat variety Capeiti 8. Run performed in 40 mM Asp buffer, 7 M urea and 0.5% HEC (apparent pH: 3.9). Conditions: 50 mm I.D., 30 cm long capillary, run at 1000 V/ cm at room temperature. Detection at 214 nm. The five overlaid tracings represent the run Nos. 1, 5, 10, 15 and 20, performed in the same day (from Capelli et al., see Ref. [19], by permission).
Asn residues (this was also demonstrated by forced chemical deamidation via alkaline hydrolysis, whereby all zeins exhibited pI values below pH 4). Thus, due to the fact that, in free solution, the mobility of zeins will be largely dictated by the charge / mass ratio,
10
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
Fig. 6. Representative CZE run of zeins, obtained by 70% ethanol extraction of ground endosperm, from the A69Y 1 line of maize. Run performed in 40 mM Asp buffer, 8 M urea and 0.5% HEC (apparent pH: 3.9). Conditions: 50 mm I.D., 30 cm long capillary, run at 700 V/ cm at 308C. Detection at 214 nm (unpublished experiments with Dr. A. Viotti, CNR, Milano).
such a low mobility in a very acidic environment is not surprising, since this ratio would correspond to only one positive charge every 4000 Da. Another field of analysis which we have found very promising is the screening of a, b and g globin chains for detection of point mutations or thalassemic conditions. Although, in the seventies, a most popular technique was isoelectric focusing (IEF) of native, intact Hb tetramers, a valid alternative in the study of hemoglobinopathies was the separation and quantitation of deheminized human globin chains, as first reported in 1966 by Clegg et al. [26], via a chromatographic step on carboxymethyl cellulose. The method had widespread applications for the study of thalassemias and in the analysis of ontogenesis of globin chains for at least a decade until, in 1979, some of us proposed, as a fast and reliable alternative, IEF of heme-free, denatured globin chains dissolved in 8 M urea and a reducing agent (2-mercaptoethanol) [27]. It was also observed that the addition of a neutral surfactant (Nonidet P-40) to the solubilization mixture brought about a unique phenomenon: it greatly improved the separation between b and g (foetal) chains and additionally induced the splitting of the g zone into two peaks, identified as the products of two genes, coding for Ala (Ag) or for Gly (Gg) in position 136 of the
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
11
foetal chains [28]. This surfactant effect, on the resolution of the two genetic variants of foetal chains, Ag and Gg, could also be reproduced by zone electrophoresis on cellulose acetate strips [29]. Fig. 7 shows the fast analysis of a and b globins in an individual heterozygous for a point mutation in the b chains (Hb Debrousse, b96 Leu → Pro). It is seen, in the control run (upper panel), that the mutant b chain is not separated from the wild-type (although this zone is considerably broader than the a chain peak, suggesting a potential analyte heterogeneity). However, when 3% surfactant is added to the sample and background electrolyte, the two phenotypes are well separated (lower panel). An interesting phenomenon can be observed in Fig. 8: in this particular case, a charged mutant was analyzed (Hb Aubenas, b26 Glu → Gly), thus one would have expected resolution of these two chains in the control run (upper panel), in the absence of surfactant. This was not the case, however, and the splitting only occurred in the presence of 3% Tween 20 (lower panel). In a way, this should have been expected (see also Section 4): given the operative pH of our isoelectric buffer, the Glu residue should be extensively protonated, so that the charged amino acid substitution went undetected, but it could still be revealed in presence of surfactant due to the huge difference in hydrophobicity of these two residues.
4. Discussion and conclusions We hope we have demonstrated, in this brief survey, that there is ample space for peptide and protein separations utilizing isoelectric, acidic buffers. It is expected that quite a few more applications will follow, since the technique has much to offer. Nevertheless, the method here reported has some advantages, but also some limitations the readers should be aware off. Perhaps one of the main improvements is the high charge imparted to peptides and proteins at such a low operative pH. For instance, in the case of the globin polypeptides, the respective titration curves show that a chains bear a total of 24 positive charges, b globins 23 and g 21 at pH 3.0. Given the fact that these three chains have very similar mass values, this results in a very favourable charge to mass ratio (on the average, one positive charge every 700 Da), which calls for very fast separations (of course, also favoured by the high voltage gradients applied, typically 600 V/ cm). However, the disadvantage is that, at such low pH values, one looses charge modulation, since essentially all negatively charged amino acids (Glu and Asp residues) are extensively protonated. This results in the paradox of Fig. 8: here a charged mutation was missed, since the Glu residues did not contribute to charge at pH 3 (however, note that this mutation was still spotted by addition of surfactant). Thus, if one desires to resolve directly charged mutants (especially those involving acidic residues) a better pH value for the background electrolyte would be around pH 5, where the three titration curves maximally diverge and where a substantial ionization of Asp and Glu residues occurs. This increment of resolution will come, of course, at the expense of longer running times, due to much lower net positive surface charge on the polypeptides, and with the added risk of adsorption of the globin chains to the ionized silica wall (we use uncoated capillaries at our low operative pH!). The other limitation one should be aware off is that, expecting absolute reproducibility of transit times in protein and peptide
12
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
Fig. 7. CZE separation of Hb Debrousse (b96 Leu → Pro). CZE conditions: 50 mM IDA buffer, added with 7 M urea and 0.5% hydroxyethyl cellulose (apparent pH 3.2) in the absence (control, Ctrl., upper panel) or in presence (lower panel) of 3% surfactant (Tween 20). Capillary: uncoated, 50 mm I.D., 375 mm O.D., 30 cm long (22 cm to detector). Run: in a Water’s Quanta 4000E, at 600 V/ cm (ca. 20 mA current) and 158C. Sample injection at the anodic side for 0.4 s at 5 p.s.i.; detection at 214 nm (unpublished experiments with A. Saccomani, Verona, and H. Wajcman, Paris).
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
13
Fig. 8. CZE separation of Hb Aubenas (b26 Glu → Gly). Upper panel: control run; lower tracing: added with 3% Tween 20. All other conditions as in Fig. 7. Note that, although this mutation involves a charged amino acid, it is not detected in the control run, but only when adding the surfactant (unpublished experiments with A. Saccomani, Verona, and H. Wajcman, Paris).
14
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
analysis, is a chimera: even conditions of suppressed electroendoosmotic flow, as in the present case, do not prevent some shifts in peak position from run to run. There might always be some minute analyte (or impurity) adsorption to the wall, which will change to some extent the charge state of the silica surface. In addition, as shown in the case of globin chain analysis, changing from a control background electrolyte to one containing surfactants brings about adsorption of micelles to the silica wall, again altering the original state of the surface. Migration times are bound to change again and the switching from one type of buffer to another one often requires long and tedious equilibration procedures aimed at restoring the original conditions of the silica surface. Some recent, extensive reviews cover quite thoroughly all these aspects of peptide and protein analysis by CZE (although no mention is given to CZE in acidic, isoelectric buffers!) and we refer the readers to them for a better insight into this field [11,12].
Acknowledgements P.G.R. is supported by grants from Agenzia Spaziale Italiana (No. ARS-98-179) and from MURST (Coordinated Project 40%, Folding / Unfolding of Proteins).
References [1] Andrews AT. Electrophoresis. theory, techniques and clinical applications, Oxford: Clarendon Press, 1986. [2] Righetti PG. Isoelectric focusing: theory, methodology and applications, Elsevier, Amsterdam, 1983, pp. 334–50. [3] Righetti PG. Immobilized pH gradients: theory and methodology, Elsevier, Amsterdam, 1990, pp. 317–33. [4] Celis JE, Bravo R, editors. Two-dimensional gel electrophoresis of proteins, Orlando: Academic Press, 1984. [5] Righetti PG, editor. Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996. [6] Guzman NA. Consecutive protein digestion and peptide derivatization employing an on-line analyte concentrator to map proteins using capillary electrophoresis. In: Righetti PG, editor, Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996, pp. 101–21. [7] Tomer KB, Parker CE, Deterding LJ. Capillary electrophoresis interfaced with mass spectrometry: electrospray ionization and continuous flow atom bombardment. In: Righetti PG, editor, Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996, pp. 123–54. [8] Ganzler K, Warne NW, Hancock WS. Analysis of r-DNA derived proteins and their post-translational modification. In: Righetti PG, editor, Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996, pp. 183–238. [9] Castagnola M, Messana I, Rossetti DV. Capillary zone electrophoresis for the analysis of peptides. In: Righetti PG, editor, Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996, pp. 239–76. [10] Chiesa C, O’Neill RA, Horvath CG, Ofner PJ. Analysis of glycoproteins, oligo- and mono-saccharides. In: Righetti PG, editor, Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996, pp. 277–430. [11] McLaughlin GM, Anderson KW. Peptide analysis by capillary electrophoresis: method development and optimization, sensitivity enhancement strategies and applications. In: Khaledi MG, editor, High performance capillary electrophoresis. Theory, techniques and applications, New York: Wiley-Interscience, 1998, pp. 637–81.
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
15
[12] Regnier FE, Lin S. Capillary electrophoresis of proteins. In: Khaledi MG, editor, High performance capillary electrophoresis. Theory, techniques and applications, New York: Wiley-Interscience, 1998, pp. 683–727. [13] Schwer C, Kenndler E. Electrophoresis in fused-silica capillaries: the influence of organic solvents on the electrosmotic velocity and the z potential. Anal Chem 1991;63:1801–6. [14] Bello MS, Capelli L, Righetti PG. Dependence of electrosmotic mobility on the applied electric field and its reproducibility in capillary electrophoresis. J Chromatogr A 1994;684:311–8. [15] Ermakov S, Zhukov MY, Capelli L, Righetti PG. Wall adsorption in capillary electrophoresis. Experimental study and computer simulation. J Chromatogr A 1995;699:297–313. [16] Righetti PG, Nembri F. Capillary electrophoresis of peptides in isoelectric buffers. J Chromatogr A 1997;772:203–11. [17] Bossi A, Righetti PG. Generation of peptide maps by capillary zone electrophoresis in isoelectric iminodiacetic acid. Electrophoresis 1997;18:2012–8. [18] Capelli L, Stoyanov AV, Wajcman H, Righetti PG. Generation of tryptic maps of a- and b-globin chains by capillary electrophoresis in isoelectric buffers. J Chromatogr A 1997;791:313–22. [19] Capelli L, Forlani F, Perini F, Guerrieri N, Cerletti P, Righetti PG. Wheat cultivar discrimination by capillary electrophoresis of gliadins in isoelectric buffers. Electrophoresis 1998;19:311–8. [20] Righetti PG, Olivieri E, Viotti A. Analysis of maize zeins by capillary electrophoresis in isoelectric buffers. Electrophoresis 1998;19:1738–41. [21] Righetti PG, Saccomani A, Stoyanov AV, Gelfi C. Human globin chain separation by capillary electrophoresis in acidic, isoelectric buffers. Electrophoresis 1998;19:1733–7. [22] Stoyanov AV, Righetti PG. Fundamental properties of isoelectric buffers for capillary zone electrophoresis. J Chromatogr A 1997;790:169–76. [23] Righetti PG, Bossi A, Gelfi C. Capillary isoelectric focusing and isoelectric buffers: an evolving scenario. J Cap Elec 1997;4:47–59. [24] Righetti PG, Gelfi C, Perego M, Stoyanov AV, Bossi A. Capillary zone electrophoresis of oligonucleotides and peptides in isoelectric buffers: theory and methodology. Electrophoresis 1997;18:2145–53. [25] Bean SR, Bietz JA, Lookhart GL. High-performance capillary electrophoresis of cereal proteins. J Chromatogr A 1998;814:25–41. [26] Clegg JB, Naughton MA, Weatherall DJ. Globin chain separation by chromatography on carboxy methyl cellulose. J Mol Biol 1966;19:91–108. [27] Righetti PG, Gianazza E, Gianni AM, Comi P, Giglioni B, Ottolenghi S, Secchi C, Rossi-Bernardi L. Human globin chain separation by isoelectric focusing. J Biochem Biophys Methods 1979;1:47–59. [28] Comi P, Giglioni B, Ottolenghi S, Gianni MA, Ricco G, Mazza U, Saglio G, Camaschella C, Pich PG, Gianazza E, Righetti PG. Gg and Ag globin chain separation and quantitation by isoelectric focusing. Biochem Biophys Res Commun 1979;87:1–8. [29] Guerrasio A, Saglio G, Mazza U, Pich PG, Camaschella C, Ricco G, Gianazza E, Righetti PG. Electrophoretic separation of Ag and Gg human globin chains in Nonidet P-40. Clin Chim Acta 1979;99:7–11.
Capillary electrophoresis of peptides and proteins in acidic, isoelectric buffers: recent developments a, a a Pier Giorgio Righetti *, Alessandra Bossi , Erna Olivieri , Cecilia Gelfi b a
University of Verona, Department of Agricultural and Industrial Biotechnologies, Strada Le Grazie, Ca` Vignal, 37134 Verona, Italy b ITBA, CNR, Via Fratelli Cervi 93, Segrate 20090 Milano, Italy Received 15 August 1998; received in revised form 23 October 1998; accepted 24 October 1998
Abstract The use of isoelectric buffers in capillary zone electrophoresis is here reviewed. Such buffers allow delivery of very high voltage gradients (up to 1000 V/ cm in relatively large bore capillaries, e.g. 75–100 mm I.D.), permitting separations of the order of a few minutes and thus conserving (in fact favouring) very high resolution due to minimal, diffusion-driven, peak spreading. Isoelectric Asp (pI 2.77 at 50 mM concentration and 258C) provides a medium of high resolving power for generating peptide maps. In difficult cases, of coincident titration curves, the pH can be moved up to higher values (e.g. pH 3.0 for 30 mM Asp) thus eliciting separation of unresolved peptides at pH 2.77. This was illustrated by running peptide maps of tryptic digests of human b globin chains. Also imino diacetic acid (pI 2.33 at 50 mM concentration) allows generation of high resolution peptide maps. Isoelectric Asp, in presence of 7 M urea and 0.5% hydroxyethyl cellulose (Mn 5 27 000 Da) is also the preferred medium for fast separation and analysis of storage proteins in cereals, such as gliadins in soft and durum wheat and zeins in maize. A solution of 50 mM iminodiacetic acid (pI 2.23) containing 7 M urea and 0.5% hydroxyethylcellulose (apparent pH 3.2) is effectively used as background electrolyte for fast separation of heme-free, denatured globin (a and b) chains. In the presence of neutral to neutral amino acid substitutions, it is additionally shown that the inclusion of 3% surfactant (Tween 20) in the sample and background electrolyte induces the separation of the wild-type and mutant chains, probably by a mechanism of hydrophobic interaction of the more hydrophobic mutant with the detergent micelle, via a
Abbreviations: IDA, imino diacetic acid; HEC, hydroxyethyl cellulose; Hb, hemoglobin; CZE, capillary zone electrophoresis; TFE, 2,2,2-trifluoroethanol *Corresponding author. Tel. / fax: 1 39-45-809-8901. E-mail address: [email protected] (P.G. Righetti) 0165-022X / 99 / $ – see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S0165-022X( 99 )00010-X
2
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
mechanism similar to ‘micellar electrokinetic chromatography’. All rights reserved.
1999 Elsevier Science B.V.
Keywords: Isoelectric buffers; Capillary electrophoresis; Proteins; Peptides
1. Introduction Protein chemists have relied on electrophoresis for more than half a century as one of the most valued tools for purification, characterization and analysis of proteins. A variety of electrophoretic techniques have been used in protein separations [1], ranging from (i) sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) and (ii) gel electrophoresis at near-physiological pH to retain native three dimensional structure, to (iii) isoelectric focusing [2] and immobilized pH gradients [3] and (iv) two-dimensional (2-D) gel electrophoresis [4]. A major advantage of gel systems is that it is relatively easy to maintain the integrity of a separation while components are being subjected to further manipulations. Dyes may be infused into gels where the separated proteins are denatured and stained, substrates may be driven into gels for zymogramming, or proteins may be electroblotted into adsorptive membranes for immunological assays. All these operations can be achieved with minimal alteration of zone integrity. Once a protein zone has been located it is even possible to bring this material back into solution for further analysis. Another major advantage of conventional gel-slab electrophoresis is the open nature of the format. The gel acts as both a fraction collector and a transport medium, which, after a separation in one dimension, may be used to carry sample components into another separation dimension without disturbing the pattern previously obtained (e.g. as adopted into 2-D maps). An extra advantage of gel-slab systems is that they allow samples to be profiled, i.e. they produce a picture of all components present in it, particularly in the case of 2-D maps, which can be used to examine extremely complex biological extracts [4]. One major drawback of all gel-slab methodologies, however, is that they are labour intensive and require too many manipulations for obtaining final results: gel preparation, electrophoretic step, staining, destaining and densitometry or photographic recording. The advent of capillary zone electrophoresis (CZE) in the nineties has greatly simplified electrophoretic procedures, by permitting automatic protocols for sample loading and data acquisition and storage [5]. The extra bonus of CZE is also the rapid sample analysis, due to the possibility of delivering high voltage gradients in narrow-lumen separation channels, and the minute sample requirements. Peptide and protein analysis has thus received a new impetus in the CZE format, as documented by several reviews [6–12]. However, even in CZE, peptide and protein separations are besieged by severe problems. The major one resides in the very nature of the fused-silica container, which presents a surface studded with silanols, believed to be present at a concentration of 8 mM / m 2 (corresponding to about five silanols per nm 2 ). Due to their weekly acidic character (pK 5.3 as in Ref. [13], or pK 6.3, as in Ref. [14]) they ionize over a broad pH
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
3
range, from pH 3.0 up to ca. pH 10. The problem with extensive silanol ionization is that the fused-silica column acts as a cation exchanger. Severe band spreading and diminished recovery occur when proteins with an isoelectric point higher than the buffer pH (and thus bearing a net positive charge when bathed in such a buffer) interact with the capillary wall [15]. An additional problem is that adsorbed proteins change the z potential at the capillary wall, impacting both the electroosmotic flow and column efficiency. This is a severe problem, because alterations in electroosmotic velocity change the elution time of all substances and compromise reproducibility. Finally, it has been shown that an axially heterogeneous distribution of z potential can diminish the separation efficiency of a column by triggering non-uniform flow along the length of the capillary. This axially heterogeneous distribution of z potential results from the inlet of the capillary being fouled with sample protein. In order to overcome these problems, a number of authors have adopted, especially in the case of peptide maps, acidic buffers (typically phosphate, formate and glycinate), at pH values ranging from 1.95 up to 2.8, in concentrations ranging from 25 up to 100 mM. CZE, if carried out in an acidic milieu, has some distinct advantages: it can be performed in uncoated capillaries, since at such pH values (typically 2.0–2.5) essentially all silanols of fused-silica are protonated and thus unable to adsorb proteins and polypeptides. However, one drawback still remains: due to the very high conductivity of these buffers, only low voltage gradients can be applied, resulting in long separation times. In order to improve the situation, we have reported separations in acidic, isoelectric buffers (notably aspartic and imino diacetic acids) [16,17]: in such amphoteric buffer solutions, much higher voltage gradients (600–1000 V/ cm) can be delivered, eliciting separations of the order of a few minutes. A number of applications of such technique have already appeared: in the analysis of tryptic digests of a and b globin chains [18], in the separation of gliadins [19], of zeins [20] and of intact globin chains [21,22]. The fundamental properties of these isoelectric buffers have been enunciated [23]. Some reviews have already covered the field [24,25]. The present review will highlight some novel aspect of this unique, most powerful technique.
2. CZE separation of peptides in isoelectric Asp Recently, Righetti and Nembri [16] proposed the use of isoelectric aspartic acid as a background electrolyte, operating at pH 5 pI 5 2.77 (at 258C and 50 mM concentration). These authors could produce peptide maps of casein in only 10–12 min (as opposed to . 70 min in standard phosphate buffer, pH 2.0) at voltage gradients as high as 800 V/ cm, with much increased resolution. Adsorption of some larger peptides to the wall was completely eliminated by adding to the background, isoelectric Asp buffer 0.5% hydroxyethyl cellulose (HEC) and 5% trifluoroethanol (TFE). Fig. 1 shows a nice example of such casein peptide maps, as produced by the standard technique (Fig. 1A) and in the presence of isoelectric Asp (Fig. 1B). However, there are some intrinsic problems with the use of isoelectric buffers. A case in point is shown in Fig. 2, which represents the CZE separation of tryptic digests of b globin chains. At the prevailing pH in isoelectric Asp (pH 2.77) two peaks (labelled T2 and T9) co-elute into a single zone.
4
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
Fig. 1. CZE of tryptic digests of b-casein. Conditions (upper panel): 100 mm I.D., 37 cm long capillary, bathed in 80 mM phosphate buffer, pH 2.0. Sample application: by pressure by applying 0.5 p.s.i. for 3 s. Separations were performed at 110 V/ cm (current: 85 mA) and detection was at 214 nm. The three major peaks, are: (1) pI 6.1, fragment b-CN (114–169); (2), pI 6.93, fragment b-CN (49–97) and (3), pI 3.95, fragment b-CN (33–48). Note that the total running time is . 70 min. B: CZE of tryptic digests of b-casein in a 100 mm I.D., 37 cm long capillary, bathed in 50 mM isoelectric aspartic acid (pH approximating the pI value of 2.77) added with 0.5% HEC (Mn: 27 000 Da) and with 5% trifluoroethanol. Running conditions: 600 V/ cm (current: 58 mA) (from Righetti and Nembri, see Ref. [16], by permission).
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
5
Fig. 2. CZE separation of a peptide digest of b globin chains. Run conditions: 37 cm long (30 cm to detector) capillary, run at 22.2 kV (600 V/ cm) at ca. 25 mA. Buffer: 50 mM Asp, pH 2.77, added with 0.5% HEC, 5% TFE and 1% CHAPS. Sample injected for 15 s by hydrostatic pressure. Note that, at the operative pH (2.77), the two fragments T2 and T9 merge into a single peak. The inset on the right shows an enlargement of the boxed area on the left side (transit times from 2.5 to 5.3 min) with proper peak identification (from Capelli et al., see Ref. [18], by permission).
By simulating the theoretical pH / mobility curves of the 13 peptides in the pH 2–4 interval, we could determine that indeed at this pH value there was a cross-over point in the titration curves of these two peptides, which could only be eliminated by working at slightly different pH values, such as pH 3.0 or 3.1. However, if one were to work at this higher pH values with a conventional, non-amphoteric buffer, one would automatically lose the benefit of adopting high voltages and thus considerably shorten the separation time while gaining in resolving power. It should be noted, however, that, as amply illustrated by Bossi and Righetti [17] in the case of IDA, and as anticipated by Righetti and Nembri [16] (see their Fig. 5), the pI of 2.77 is only a theoretical value, reached at high enough concentration of Asp (in this case 50 mM). By progressive dilution of the buffering ion, one can span a small pH interval, up to 0.4 pH units. Thus, it is possible, while still working under isoelectric conditions, to modulate the pH of the background electrolyte simply by varying its concentration. In a buffer composed of 30 mM Asp (still ensuring adequate buffering power), the final pH of the background electrolyte is
6
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
Fig. 3. CZE separation of a peptide digest of b globin chains. Run conditions as in Fig. 2, except that the buffer was: 30 mM Asp, pH 3.0, added with 0.5% HEC, 10% TFE and 50 mM NDSB-195. Sample injected for 15 s by hydrostatic pressure. Note in the inset (an enlargement of the boxed area on the left side) the full splitting of peaks T2 and T9, unresolved in the buffer of Fig. 2 (from Capelli et al., see Ref. [18], by permission).
3.0. When adopting this last buffer (see Fig. 3), it was possible to separate all the 13 b globin peptides. By enlarging the 3–6 min separation window (see Fig. 3, insert) one can indeed appreciate that now the T2 and T9 b-peptides, which co-migrated into a single zone, are base-line resolved. However, as the pH increased to 3.0, there was a higher risk of peptide adsorption to the naked silica wall (only uncoated capillaries are used throughout) and addition of a zwitterion (non-detergent sulphobetaine, NDSB-195) was necessary in order to quench such interaction.
3. CZE separation of proteins in isoelectric buffers The situation with proteins is, of course, more complex: it had first to be demonstrated that these macromolecules would not bind to the silica surface, given the fact that already some large peptides did. In fact (cf. Fig. 1B) already the large casein fragments labelled pI 3.95, pI 6.1 and pI 6.93 gave tailing zones, suggesting adsorption to the few
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
7
ionized silanols at the operative pH of Asp (pI 2.80). An excellent cure to this noxious phenomenon was found with the addition of 0.5% hydroxyethyl cellulose (HEC, average Mn 5 27 000 Da), which probably offers a dynamic coating of the wall. HEC was thus adopted in all instances of separations in acidic, isoelectric buffers. The second problem regards the integrity of proteins at such acidic pH values: it is to be expected that, in general, proteins will be denatured in such a milieu; the result could be exposure of hydrophobic residues, intra-chain interaction, aggregation and precipitation. Thus, our technique cannot quite be recommended for analysis of native proteins, but could be a still valid option for denatured polypeptides. There is at least a very important class of proteins to which our technique could apply: the storage proteins (prolamins) or cereals. Knowledge of cereal proteins is valuable in many ways. For example, because of their heterogeneity and nearly invariant expression, storage proteins provide fingerprints that differentiate cereal genotypes and cultivars. This ability to recognize cultivars is important during breeding, marketing and in research. Through protein analysis, breeders can identify and select optimal parental genotypes and progeny during cultivar development and can ascertain varietal purity. In production agriculture, varietal identification can ensure use of lines with optimal economic return, quality, yield, genetic resistance or adaptability. Millers, bakers, brewers, maltsters and other processors must guarantee that grains and flours are suitable for high quality food and on-food products. Prolamins (called gliadins, in wheat, zeins in maize, hordeins in barley, secalins in oats) are typically synthesized and packaged in the endosperm as storage granules, insoluble in water, salt and buffers solutions. Prolamins are soluble in aqueous alcohol blends and, after extraction, can be efficiently analyzed by CZE in 7–8 M urea solutions. Extensive work in the optimization of CZE profiling of cereal prolamins is already available (for a review, see Ref. [25]), although it is typically performed in high conductivity buffers, such as 100 mM phosphate, pH 2.5. Just as reported by Bean et al. [25], in a 50 mm I.D. capillary it is impossible to obtain a separation in less than 30 min in the phosphate background electrolyte and the run is often plagued by base-line disturbances. On the contrary, when the same separation is performed in our isoelectric Asp buffer, in the presence of 7 M urea and 0.5% HEC, the run is typically over in 10 min (at a voltage gradient of 1000 V/ cm), if one does not consider some minor, late-eluting peaks in the 13–14 min time window and present only in some cultivars (Fig. 4A and B). In Fig. 4A, a series of overlaid, individual runs of four different varieties of durum wheat is given: it is seen that the pattern is highly specific and that the different cultivars can be recognized at a glance. Fig. 4B shows a spectrum of gliadin bands resolved in a total of four different cultivars of bread wheat. Here too the overlaid tracings show that such patterns are indeed cultivar-specific. In both Fig. 4A and B, the fact that the peaks are quite sharp and symmetric, and that the base line, after passage of the whole train of zones, returns to the initial value, suggests that there is no binding of these proteins to the capillary wall. The absence of sharp spikes in the UV tracings also indicates the absence of precipitated material. Fig. 5 shows the reproducibility of our separation system: it represents the overlay of 20 consecutive runs, taken every fifth run. It is seen that there is a tendency for the whole train of peaks to lengthen a bit the transit times in the repeated runs. The overall difference between run No. 1 and No. 20 is 4% (in absolute migration times, not relative
8
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
Fig. 4. Representative CZE run of gliadins, obtained by sequential extraction of flour, from four different cultivars of durum wheat (A) and four different cultivars of bread wheat (B). Runs performed in 40 mM Asp buffer, 7 M urea and 0.5% HEC (apparent pH: 3.9). Conditions: 50 mm I.D., 30 cm long capillary, run at 1000 V/ cm at room temperature. Detection at 214 nm (from Capelli et al., see Ref. [19], by permission).
or normalized migration times), measured on the highest peak eluting around 7.5 min. This is in agreement with the data of Bean et al. [25], who reported a non-linear relationship between transit times and run number. Also the reproducibility in peak areas (again measured on the largest peak at 7.5 min) is quite acceptable: the overall variation is never greater than 62%, averaged over the 20 runs. Fig. 6 shows a representative pattern of a maize line (A69Y 1 ) run in 40 mM Asp, 8 M urea and 0.5% HEC, at 700 V/ cm. Here too one can appreciate a train of sharp peaks, quite symmetric, also suggesting absence of interaction with the wall silanols. At a first glance, it sounds odd that zeins should elute, in a typical electropherogram, in a total of 30 min transit time, although the analysis is performed at 700 V/ cm. When screening wheat gliadins by the same technique, the entire pattern was typically developed in 10 min (at 1000 V/ cm). The reason for that might reside in the unique amino acid composition of zeins: on an average mass of 20 kDa, zeins typically exhibit the following average basic amino composition: one Lys, two His and two Arg residues. On the contrary, they appear to have a vast number of acidic residues: an average of 40 Glu and nine Asp. However, due to the fact that their pI values extend across neutrality, it would appear that ca. 90% of these residues exist, in the native polypeptides, as Gln and
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
9
Fig. 5. Reproducibility of a series of 20 consecutive CZE runs of gliadins from the durum wheat variety Capeiti 8. Run performed in 40 mM Asp buffer, 7 M urea and 0.5% HEC (apparent pH: 3.9). Conditions: 50 mm I.D., 30 cm long capillary, run at 1000 V/ cm at room temperature. Detection at 214 nm. The five overlaid tracings represent the run Nos. 1, 5, 10, 15 and 20, performed in the same day (from Capelli et al., see Ref. [19], by permission).
Asn residues (this was also demonstrated by forced chemical deamidation via alkaline hydrolysis, whereby all zeins exhibited pI values below pH 4). Thus, due to the fact that, in free solution, the mobility of zeins will be largely dictated by the charge / mass ratio,
10
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
Fig. 6. Representative CZE run of zeins, obtained by 70% ethanol extraction of ground endosperm, from the A69Y 1 line of maize. Run performed in 40 mM Asp buffer, 8 M urea and 0.5% HEC (apparent pH: 3.9). Conditions: 50 mm I.D., 30 cm long capillary, run at 700 V/ cm at 308C. Detection at 214 nm (unpublished experiments with Dr. A. Viotti, CNR, Milano).
such a low mobility in a very acidic environment is not surprising, since this ratio would correspond to only one positive charge every 4000 Da. Another field of analysis which we have found very promising is the screening of a, b and g globin chains for detection of point mutations or thalassemic conditions. Although, in the seventies, a most popular technique was isoelectric focusing (IEF) of native, intact Hb tetramers, a valid alternative in the study of hemoglobinopathies was the separation and quantitation of deheminized human globin chains, as first reported in 1966 by Clegg et al. [26], via a chromatographic step on carboxymethyl cellulose. The method had widespread applications for the study of thalassemias and in the analysis of ontogenesis of globin chains for at least a decade until, in 1979, some of us proposed, as a fast and reliable alternative, IEF of heme-free, denatured globin chains dissolved in 8 M urea and a reducing agent (2-mercaptoethanol) [27]. It was also observed that the addition of a neutral surfactant (Nonidet P-40) to the solubilization mixture brought about a unique phenomenon: it greatly improved the separation between b and g (foetal) chains and additionally induced the splitting of the g zone into two peaks, identified as the products of two genes, coding for Ala (Ag) or for Gly (Gg) in position 136 of the
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
11
foetal chains [28]. This surfactant effect, on the resolution of the two genetic variants of foetal chains, Ag and Gg, could also be reproduced by zone electrophoresis on cellulose acetate strips [29]. Fig. 7 shows the fast analysis of a and b globins in an individual heterozygous for a point mutation in the b chains (Hb Debrousse, b96 Leu → Pro). It is seen, in the control run (upper panel), that the mutant b chain is not separated from the wild-type (although this zone is considerably broader than the a chain peak, suggesting a potential analyte heterogeneity). However, when 3% surfactant is added to the sample and background electrolyte, the two phenotypes are well separated (lower panel). An interesting phenomenon can be observed in Fig. 8: in this particular case, a charged mutant was analyzed (Hb Aubenas, b26 Glu → Gly), thus one would have expected resolution of these two chains in the control run (upper panel), in the absence of surfactant. This was not the case, however, and the splitting only occurred in the presence of 3% Tween 20 (lower panel). In a way, this should have been expected (see also Section 4): given the operative pH of our isoelectric buffer, the Glu residue should be extensively protonated, so that the charged amino acid substitution went undetected, but it could still be revealed in presence of surfactant due to the huge difference in hydrophobicity of these two residues.
4. Discussion and conclusions We hope we have demonstrated, in this brief survey, that there is ample space for peptide and protein separations utilizing isoelectric, acidic buffers. It is expected that quite a few more applications will follow, since the technique has much to offer. Nevertheless, the method here reported has some advantages, but also some limitations the readers should be aware off. Perhaps one of the main improvements is the high charge imparted to peptides and proteins at such a low operative pH. For instance, in the case of the globin polypeptides, the respective titration curves show that a chains bear a total of 24 positive charges, b globins 23 and g 21 at pH 3.0. Given the fact that these three chains have very similar mass values, this results in a very favourable charge to mass ratio (on the average, one positive charge every 700 Da), which calls for very fast separations (of course, also favoured by the high voltage gradients applied, typically 600 V/ cm). However, the disadvantage is that, at such low pH values, one looses charge modulation, since essentially all negatively charged amino acids (Glu and Asp residues) are extensively protonated. This results in the paradox of Fig. 8: here a charged mutation was missed, since the Glu residues did not contribute to charge at pH 3 (however, note that this mutation was still spotted by addition of surfactant). Thus, if one desires to resolve directly charged mutants (especially those involving acidic residues) a better pH value for the background electrolyte would be around pH 5, where the three titration curves maximally diverge and where a substantial ionization of Asp and Glu residues occurs. This increment of resolution will come, of course, at the expense of longer running times, due to much lower net positive surface charge on the polypeptides, and with the added risk of adsorption of the globin chains to the ionized silica wall (we use uncoated capillaries at our low operative pH!). The other limitation one should be aware off is that, expecting absolute reproducibility of transit times in protein and peptide
12
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
Fig. 7. CZE separation of Hb Debrousse (b96 Leu → Pro). CZE conditions: 50 mM IDA buffer, added with 7 M urea and 0.5% hydroxyethyl cellulose (apparent pH 3.2) in the absence (control, Ctrl., upper panel) or in presence (lower panel) of 3% surfactant (Tween 20). Capillary: uncoated, 50 mm I.D., 375 mm O.D., 30 cm long (22 cm to detector). Run: in a Water’s Quanta 4000E, at 600 V/ cm (ca. 20 mA current) and 158C. Sample injection at the anodic side for 0.4 s at 5 p.s.i.; detection at 214 nm (unpublished experiments with A. Saccomani, Verona, and H. Wajcman, Paris).
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
13
Fig. 8. CZE separation of Hb Aubenas (b26 Glu → Gly). Upper panel: control run; lower tracing: added with 3% Tween 20. All other conditions as in Fig. 7. Note that, although this mutation involves a charged amino acid, it is not detected in the control run, but only when adding the surfactant (unpublished experiments with A. Saccomani, Verona, and H. Wajcman, Paris).
14
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
analysis, is a chimera: even conditions of suppressed electroendoosmotic flow, as in the present case, do not prevent some shifts in peak position from run to run. There might always be some minute analyte (or impurity) adsorption to the wall, which will change to some extent the charge state of the silica surface. In addition, as shown in the case of globin chain analysis, changing from a control background electrolyte to one containing surfactants brings about adsorption of micelles to the silica wall, again altering the original state of the surface. Migration times are bound to change again and the switching from one type of buffer to another one often requires long and tedious equilibration procedures aimed at restoring the original conditions of the silica surface. Some recent, extensive reviews cover quite thoroughly all these aspects of peptide and protein analysis by CZE (although no mention is given to CZE in acidic, isoelectric buffers!) and we refer the readers to them for a better insight into this field [11,12].
Acknowledgements P.G.R. is supported by grants from Agenzia Spaziale Italiana (No. ARS-98-179) and from MURST (Coordinated Project 40%, Folding / Unfolding of Proteins).
References [1] Andrews AT. Electrophoresis. theory, techniques and clinical applications, Oxford: Clarendon Press, 1986. [2] Righetti PG. Isoelectric focusing: theory, methodology and applications, Elsevier, Amsterdam, 1983, pp. 334–50. [3] Righetti PG. Immobilized pH gradients: theory and methodology, Elsevier, Amsterdam, 1990, pp. 317–33. [4] Celis JE, Bravo R, editors. Two-dimensional gel electrophoresis of proteins, Orlando: Academic Press, 1984. [5] Righetti PG, editor. Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996. [6] Guzman NA. Consecutive protein digestion and peptide derivatization employing an on-line analyte concentrator to map proteins using capillary electrophoresis. In: Righetti PG, editor, Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996, pp. 101–21. [7] Tomer KB, Parker CE, Deterding LJ. Capillary electrophoresis interfaced with mass spectrometry: electrospray ionization and continuous flow atom bombardment. In: Righetti PG, editor, Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996, pp. 123–54. [8] Ganzler K, Warne NW, Hancock WS. Analysis of r-DNA derived proteins and their post-translational modification. In: Righetti PG, editor, Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996, pp. 183–238. [9] Castagnola M, Messana I, Rossetti DV. Capillary zone electrophoresis for the analysis of peptides. In: Righetti PG, editor, Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996, pp. 239–76. [10] Chiesa C, O’Neill RA, Horvath CG, Ofner PJ. Analysis of glycoproteins, oligo- and mono-saccharides. In: Righetti PG, editor, Capillary electrophoresis in analytical biotechnology, Boca Raton: CRC Press, 1996, pp. 277–430. [11] McLaughlin GM, Anderson KW. Peptide analysis by capillary electrophoresis: method development and optimization, sensitivity enhancement strategies and applications. In: Khaledi MG, editor, High performance capillary electrophoresis. Theory, techniques and applications, New York: Wiley-Interscience, 1998, pp. 637–81.
P.G. Righetti et al. / J. Biochem. Biophys. Methods 40 (1999) 1 – 15
15
[12] Regnier FE, Lin S. Capillary electrophoresis of proteins. In: Khaledi MG, editor, High performance capillary electrophoresis. Theory, techniques and applications, New York: Wiley-Interscience, 1998, pp. 683–727. [13] Schwer C, Kenndler E. Electrophoresis in fused-silica capillaries: the influence of organic solvents on the electrosmotic velocity and the z potential. Anal Chem 1991;63:1801–6. [14] Bello MS, Capelli L, Righetti PG. Dependence of electrosmotic mobility on the applied electric field and its reproducibility in capillary electrophoresis. J Chromatogr A 1994;684:311–8. [15] Ermakov S, Zhukov MY, Capelli L, Righetti PG. Wall adsorption in capillary electrophoresis. Experimental study and computer simulation. J Chromatogr A 1995;699:297–313. [16] Righetti PG, Nembri F. Capillary electrophoresis of peptides in isoelectric buffers. J Chromatogr A 1997;772:203–11. [17] Bossi A, Righetti PG. Generation of peptide maps by capillary zone electrophoresis in isoelectric iminodiacetic acid. Electrophoresis 1997;18:2012–8. [18] Capelli L, Stoyanov AV, Wajcman H, Righetti PG. Generation of tryptic maps of a- and b-globin chains by capillary electrophoresis in isoelectric buffers. J Chromatogr A 1997;791:313–22. [19] Capelli L, Forlani F, Perini F, Guerrieri N, Cerletti P, Righetti PG. Wheat cultivar discrimination by capillary electrophoresis of gliadins in isoelectric buffers. Electrophoresis 1998;19:311–8. [20] Righetti PG, Olivieri E, Viotti A. Analysis of maize zeins by capillary electrophoresis in isoelectric buffers. Electrophoresis 1998;19:1738–41. [21] Righetti PG, Saccomani A, Stoyanov AV, Gelfi C. Human globin chain separation by capillary electrophoresis in acidic, isoelectric buffers. Electrophoresis 1998;19:1733–7. [22] Stoyanov AV, Righetti PG. Fundamental properties of isoelectric buffers for capillary zone electrophoresis. J Chromatogr A 1997;790:169–76. [23] Righetti PG, Bossi A, Gelfi C. Capillary isoelectric focusing and isoelectric buffers: an evolving scenario. J Cap Elec 1997;4:47–59. [24] Righetti PG, Gelfi C, Perego M, Stoyanov AV, Bossi A. Capillary zone electrophoresis of oligonucleotides and peptides in isoelectric buffers: theory and methodology. Electrophoresis 1997;18:2145–53. [25] Bean SR, Bietz JA, Lookhart GL. High-performance capillary electrophoresis of cereal proteins. J Chromatogr A 1998;814:25–41. [26] Clegg JB, Naughton MA, Weatherall DJ. Globin chain separation by chromatography on carboxy methyl cellulose. J Mol Biol 1966;19:91–108. [27] Righetti PG, Gianazza E, Gianni AM, Comi P, Giglioni B, Ottolenghi S, Secchi C, Rossi-Bernardi L. Human globin chain separation by isoelectric focusing. J Biochem Biophys Methods 1979;1:47–59. [28] Comi P, Giglioni B, Ottolenghi S, Gianni MA, Ricco G, Mazza U, Saglio G, Camaschella C, Pich PG, Gianazza E, Righetti PG. Gg and Ag globin chain separation and quantitation by isoelectric focusing. Biochem Biophys Res Commun 1979;87:1–8. [29] Guerrasio A, Saglio G, Mazza U, Pich PG, Camaschella C, Ricco G, Gianazza E, Righetti PG. Electrophoretic separation of Ag and Gg human globin chains in Nonidet P-40. Clin Chim Acta 1979;99:7–11.