2012 A . Bossi and P. G . Righetti Electrophorrriv 1997. 18, 21312-2018

Aiessandra Bossi Pier Giorgio Righetti

Generation of peptide maps by capillary zone electrophoresis in isoelectric iminodiacetic acid

University of Verona, Department of Agricultura1 and Industrial Biotechnologies, Verona, Italy

Capillary zone electrophoresis in stationary, isoelectric buffers is a novel method for generating peptide maps of protein digests. The buffer system developed is composed of iminodiacetic acid (IDA), whose physico-chemical parameters were found - by theoretically modeling and experimental verifica- tion - to be: pf 2.23 (at 100 mM concentration), pK, = 1.73 and pK, = 2.73 (no attempts were made at measuring the pK of the primary amino group, since such a low pf value would be compatible with any pK value of the basic group, down to as low as pK 5.5). IDA is compatible with most hydro-organic sol- vents, including trifluoroethanol (TFE), up to at least 40% v/v, typically used for modulating peptide mobility. In naked capillaries, a buffer comprising 50 mM IDA, 10% TFE and 0.5 O/o hydroxyethylcellulose (HEC) allows genera- tion of peptide maps with high resolution, reduced transil limes and no inle- raction of even large peptides with the wall. However, the best background electrolyte was found to be a solution of 50 mM IDA in 0.5% HEC and 6-8 M urea, one of the best solubilizers of proteins and peptides known. In this last electrolyte system, peptide maps of 0-casein digests (known to contain also very large peptides, up to 6000 Da) could be generated with excellent resolu- tion and half the transit times as compared with the standard buffer adopted in peptide analysis (80 mM phosphate buffer, pH 2.0). IDA thus appears to be another valid isoelectric buffer system, operating in a different pH vvindow (pH 2.33 in 50 mM IDA) as compared to the other amphotere previously adopted (50 mM Asp, pH 2.77) for the same kind of analysis.

1 Introduction

Isoelectric buffers are gaining a wider acceptance in capillary zone electrophoresis (CZE) due to the fact that, in virtue of their extremely low conductivities, they are compatible with high voltage gradients (800-1000 V/cm), thus favoring high resolution and allowing much reduced transit times. Perhaps one of the earliest reports on the use of isoelectric buffers was that of Mandecki and Hayden [l], who adopted isoelectric histidine as the sole buffering species in gel slab electrophoresis of oligo- nucleotides. However, the use of such buffers outside their natural use, i.e. for creating pH gradients in isoelec- tric focusing [2], was largely ignored in electrokinetic sep- arations, with the exception of the preparative free-flow electrophoretic applications by Bier e? al. [3-51, where cycloserine was used as a buffer. Only in 1995, Hjertén e? al. [6] reproposed such buffers for CZE analysis. This group demonstrated excellent fractionations of proteins and other small M, analytes, in a time scale up to 100 s . Such separations could also be modeled theoretically by Branco et al. [7] with close agreement of simulated data with the experimental ones. At the same time, Gelfi e? al. [8] proved that in 100 mM His (pH = pZ= 7.47) anal- ysis of antisense oligonucleotides could be performed in as little as 4-5 min with superior resolution as compared

Correspondence: Prof. P. G . Righetti, L.I.T.A., Room 3.16, Via Fratelli Cervi No. 93, Segrate 20090 (Milano), Italy (Tel/Fax: +39-2-2642364; E-mail: righetti@imiucca.csi.unimi.it)

Nonstandard abbreviations: 8-CN, B-casein; CZE, capillary zone elec- trophoresis; HEC, hydroxyethyl cellulose; IDA, iminodiacetic acid; TFE, 2,2,2-trifluoroethanol

Keywords: Isoelectric buffers / Peptide maps / Casein

with the standard Tris-borate-EDTA buffers (pH 8.3) nor- mally adopted for DNA separations. In a subsequent report, Stoyanov e? al. [9] explored such oligonucleotide separations in two types of stationary buffers: single isoelectric amphoteres, such as His, Lys and focuised car- rier ampholytes. A novel, fundamental parameter for eva- luating the performance of such isoelectric species was proposed: not just the absolute buffering power (B; by which a buffer such as Lys would be much superior to His, owing to the favorable A pK value) but the Blh ratio, i.e., the ratio between the B power and its conductivity, by which His offers a better performance due to its p l being close to neutrality, where bulk solvent conduc- tivity is negligible. In this paper, a novel concept for oli- gonucleotide separation was described, namely zone electrophoresis against a stationary pH gradient, by which a mechanism similar to the classical disc electro- phoresis for proteins could be activated, leading to a sep- aration driven simultaneously by the charge and mass differences of the DNA analytes.

Soon the concept of isoelectric buffers was also adopted for separation of peptides. Peptide mapping for many years has been used for the characterization of the pri- mary structure of proteins, ever since Ingram firsí report- ed, in a now classical series of papers [lo-121, tlie tech- nique of fingerprinting and discovered that the differ- ence between normal human adult and sickle cell hemo- globins was the replacement of a glutamic acid wilh a valine residue. By and large, the experimental parame- ters adopted in developing peptide maps by CZE have been based on the use of acidic buffers (typically phos- phate, formate, glycinate) at pH values ranging firom 1.9 up to 2.8, in concentrations from 25 to 100 mhf (for a review, see [13]). The use of acidic separation buffers for

0 WILEY-VCH Vcrlag GmbH. 69451 Weinheim. 1997 0173-0835/97/1111-2012 $17 50+.50/0

Generation of peptide maps by C Z E 2013 Elecrroohoresis 1997, IR. 2012-2018

peptide mapping by CZE presents some distinct advan- tages. First of all, at pH values well below the pK of the free silanols on the fused-silica wall (assessed as pK = 6.3, with a wall neutralization at pH 2.3) [14], their disso- ciation will be significantly suppressed and, in principle, the negative charge on the silica surface abolished. The electroosmotic flow (EOF) should thus be negligible and the peak capacity greatly increased. Additionally, the low pH value ensures that the vast majority of peptides will bear a net positive charge, thus allowing their migration at the cathodic end of the capillary, past the detector window. As a third bonus, the negligible negative charge on the capillary wall will also mean that the coulombic interaction between the fused-silica surface and the pep- tides will be markedly diminished, thereby removing the most important reason for peak broadening and skewing [ 15, 161. However, satisfactory separations are not always obtained using acidic buffer conditions. As the charge of the peptides plays a pivotal role in their separation by CZE, other pH values might be necessary in order to avoid comigration of peptides having cross-over points at a given pH value in their titration (pH/mobility) curves. Additionally, at low-enough pH values, effective- ly minimizing peptide-wall interaction (Le., at or below pH 2), the buffer and bulk water conductivities are so high that only low voltage gradients can be adopted (typically not higher than 200 V/cm), thus greatly leng- thening the analysis times (in many reports, up to 60-70 min are required for full development of complex peptide maps). During such long runs, peptide zones considerably broaden and peak resolution worsens. In view of these shortcomings, Righetti and Nembri [17] recently suggested the use of isoelectric aspartic acid as a background electrolyte, operating at pH = p I = 2.77 (at 25°C). These authors could produce peptide maps of casein in only 10-12 min (as opposed to 80 min in standard phosphate buffer, at 2.0) at voltage gradients as high as 800 Vlcm, with much increased resolution. Adsorption of some larger peptides to the wall was completely eliminated by addition of 0.5 Yo hydroxyethyl- cellulose (HEC) and 5% trifluoroethanol (TFE) to the background isoelectric Asp buffer.

Although the method has been successfully used for generating peptide maps of a- and 0-globin chains from tryptic digests of human adult hemoglobin 1181, and has been shown to be highly competitive in respect to ana- logous separations by RP-HPLC, it suffers from some drawbacks. One of them is the limited solubility of Asp at pH = pl: already at 50 mM concentrations this buffer tends to precipitate if the room temperature drops below 20°C. Additionally, Asp is incompatible with a number of organic solvents typically adopted for peptide separa- tions: among them, trifluoroethanol (TFE), which has been found to the quite useful as peptide solubilizer [19] and as a modulator of peptide mobility [20]. As an addi- tional drawback, its pZ (2.77) is somewhat too high to prevent peptide interaction with the wall silanols in uncoated caoillaries. In search of an even better back-

aware of any use of this peculiar buffering ion in any electrokinetic methodology, this compound is well- known in the field of immobilized metal afinity chroma- tography (IMAC), where it is covalently bound to resin beads (via its amino group) for chelating a variety of metals 1211.

2 Materials and methods

2.1 Reagents

IDA and TFE were obtained from Fluka Chemie (Buchs, Switzerland). Fused-silica capillaries (75 pm ID, 375 pm OD) were from Polymicro Technologies (Phoenix, AZ, USA) and were used as such, without inner coating. The Centricon 30 membranes were from Millipore (Bedford, MA, USA). Trypsin (TPCK-treated, type XIII) and B-casein were purchased from Sigma (St. Louis, MO, USA). Hydroxyethylcellulose (HEC, number-average molecular mass, M, of 27 000) was from Aldrich (Mil- waukee, WI, USA).

2.2 Capillary electrophoresis

Capillary zone electrophoresis (CZE) was carried out with the Waters Quanta 4000E unit equipped with auto- mated Millenium software. Fifty cm long and 75 pm ID uncoated capillaries were used. Severa1 types of back- ground electrolytes were tested: (i) the standard 80 mM phosphate buffer, pH 2.0, commonly adopted in peptide separations (in this case, the capillary was first precondi- tioned by washing with buffer containing 1.75% liquid linear polyacrylamide); (ii) 50 mM isoelectric IDA (pH = pZ= 2.30 at 25°C); (iii) same as (ii), but supplemented with 0.5% HEC M, 27 O00 and 10 or 15% TFE; (iv) same as (ii) but supplemented with 0.50/0 HEC and 6 M urea. In al1 cases, the sample and standard were loaded by hydrostatic pressure for 15 s . Separations were performed at 150 V/cm in buffer (a) and up to 500 V/cm in buffers (ii)-(iv). Ultraviolet absorbance was moni- tored at 214 nm.

2.3 pH determinations

In order to assess potential pH changes with dilution of isoelectric IDA, the following potentiometric titration was performed. Doubly distilled, degassed water was pre- pared to contain 10 mM KC1. The solution was thermos- tated at 25°C and supplemented with IDA, first up to 5 mM, then up to 10 mM, and subsequently at 10 mM in- crements up to 100 mM. With each IDA addition, the pH was carefully assessed with a pHM64 Research pH Meter, equipped with a GK2401C combination electrode from Radiometer (Copenhagen, Denmark). The apparent p l changes as a function of increasing amounts of organ- ic solvents (such as TFE and urea) were measured in the same manner.

ground electrolyte, we report here the use of iminodia- cetic acid (IDA) as a unique isoelectric buffer, possess- ing a remarkably low pZ(2.23 at 100 mM concentration) and an extreme solubility both in neat water and in a number of hydro-organic solvents. Although we are not

2.4 Buffering capacity and conductivity measurements

In order to asseSS the p power of free IDA (from which one could derive the two pKs of the neighboring carbox-

2014 A. Bossi aiid P. ti. Righetti Elrcimphor~.\rr 1997, 18, 2012-2018

yls active at pH = pZ) the following measurements were carried out. A 50 mM solution of IDA (20 mL), thermos- tated at 25"C, was left to equiiibrate in presence of the pH electrode, until constant readings of pH values were attained. This solution was then rendered 5 mM iri NaOH (from a carefully titrated 1 N NaOH standard) and the pH variation measured. Two subsequent addi- tions of NaOH were made and the three different pH readings averaged out (as we are working in a linear por- tion of the titration curve, the three readings were in fact identical). By definition, 6 = dB/d(pH), i.e., it is given by the amount (in equivalents) of a strong base added to the solution of the protolyte, divided by the resulting increase in pH occurring in said soiution [22]. The resulting value of fi power was additionally assessed with the differential pH meter [23]. Conductivity of neat IDA solutions, or in presence of various additives (TFE, urea) was measured at 25°C with an Orion conductivity meter fitted with a 1 cm cell.

2.5 0-Casein digestion

Tryptic peptide maps of P-casein were obtained as fol- lows: 10 mg/mL p-casein, dissolved in 50 mM CAPS buffer, pH 8.2, were added with trypsin to a trypsin/B- casein ratio of 2% w/w. HydFolysis continued for 4 h at 50°C. The reaction was stopped by adding acetic acid to pH 4.0. The peptides were separated from excess trypsin by centrifugation through Centricon membranes.

3 Results

3.1 Physico-chemical characterization of IDA solutions

As we could not find any report on the utilization of IDA as a buffering ion in electrophoresis, we first set out to measure its physico-chemical parameters, since we were unable to find data even on its plvaiue. Figure 1 shows the dependence of the pH and conductivity from the molarity of IDA solutions. It is well known, from lheoretical considerations of Rilbe [24] and from our experimental work [17, 181, that indeed the p l of an amphotere is a limit value, which can vary between two extremes: at high-enough concentration, it will approach the true p.l value, at infinite dilution it will reach the p l of distilled water (p1 = 7.0). It is in fact seen that, at rather low concentrations ( 5 mM) the apparent p í of IDA is 2.70, whereas at high-enough levels (100 mM) it tends to plateau at a value of 2.23, which we assume to be the real p l value of IDA. Concomitantly, as the pH decreases at increasing concentrations of IDA, also the overall con- ductivity of the solution increases. For practica1 pur- poses, we then adopted a constant concentration of IDA of 50 mM (apparent pí: 2.33).

Since selectivity in peptide separations can be markedly manipulated by working in hydroorganic solvents (espe- cially in TFE), it was of interest to measure the same parameters (pH and conductivity, which greatly condi- tion the voltage gradients applicable to the capillary sepa- ration cell) in presence of these additives. Figure 2 gives the variations of DH and conductivitv as a function of

4-17 h

O E E E v

O

I I I I 2.20

O 25 50 75 100 125 IDA concentration (mM)

Figure 1. pH and conductivity profiles as a function of IDA conceritra- tion in solution. IDA was varied from 5 to 100 mM. Note that in this concentration interval the apparent pí decreases from pH 2.70 to pH 2.23, with a concomitant increment of conductivity from 0.5 to 3.5 mmhos.

3 , 0 0 1 h o 5 2.00

2 o

E .- c v

>

1.00

0.00

o 1244 D

o

O

2.40

5 U

2.36

O

O 20 40 % Trifluoroethanol

Figure 2. pH and conductivity profiles as a function of addition of tri- íluoroethanol at constant IDA concentration of 50 mM in solution. TFE was varied from O to 40% v/v. Note that, although there is a marked decrement of the mixed hydroorganic solvent conductivity, the variation of apparent p l is minute.

whereas the conductivity markedly decreases (from 2.4 to 0.5 mmhos), the apparent p l of IDA is hardly affected (it varies only from 2.33 up to 2.43 in 40% TFE). On the contrary, both parameters are strongly affected in pres- ente of urea (Fig. 3). In 6 M urea the conductivity drops from 2.4 down to 1.1 mmhos and the apparent pf in- creases from 2.33 to as high as 3.13. This marked de- crease of conductivity is probably due to the fact that urea coordinates severa1 molecules of water by means of weak interactions. There is thus a decrease of free water

addition of TFE,-from 5 to 40% viv. It is seen that, molecules in the solvent, which corresponds to a change

Elecrrophuresis 1997. IR. 2012-2018 üeneration of peptide maps by CZE 2015

2 4

o

O

1 2

O

O

O

O

o

l , I l I l , r

3 60

3 20

I , 2 80

O

2 40

2 O0

O 2 6 8 U d (M)

Figure 3. pH and conductivity profiles as a function of addition of urea at constant IDA concentration of 50 rnM in solution. Urea was varied from O to 6 M. Note the marked decrernent of the mixed hydro- organic solvent conductivity and the large increase in apparent p l of IDA.

in the dielectric constant of the IDA-urea solution, which leads to a marked change in pK (and thus pi) values of the ionogenic groups of IDA, with concom- itant increment in the apparent pH of the IDA solution and thus a decrement of conductivity.

3.2 Generation of B-casein peptide maps

We next assessed the behavior of the IDA background electrolyte in resolving peptides from a tryptic digest of B-casein, as compared with standard protocols. We have adopted B-casein since this system was well character- ized by us in previous work, both in CZE [17] and in an immobilized enzyme reactor [25]. In addition, P-casein lysates contain some large fragments, reaching as high as 5000-6000 Da, liable to interact with the free silanols of the silica wall. Thus, this digest would represent the tou- ghest possible test to the novel method here proposed. Figure 4 shows the CZE separation of the tryptic digest of B-casein in the standard 80 mM phosphate buffer, pH 2.0. Between 20-25 major and minor peaks are distin- guishable in the electropherogram. Among them, three major peaks could be precisely mapped by subjecting them to IPGs, followed by band excision, microse- quencing, and mass spectrometry. They are: peak 1, p í 6.1, fragment j3-CN (114-169); peak 2, p l 6.93, frag- ment 8-CN (49-97); and peak 3, p13.95, fragment B-CN (33-48). This separation, however, suffers from some inconveniences: first, the analysis time is too long, since ca. 75 min are required for complete map development. Second, probably due to the long running time, not al1 peaks are resolved to base-line and some zones appear to the envelopes of more than one band, since shoulders are clearly visible in some of them. Additionally, the peak widths are unacceptably large: e.g., the pl3.95 frag- ment, one of the last eluting peaks, has a width at 4 (J of 2 and 'í2 min. Due to the relatively large-bore capillaries utilized (75 vm), needed for increased sensitivity, it was

40.00

30.00

h c d r N 20.00 8 L %

C m

2 10.00

0.00

1 I 1 I

Time (min)

1 I I

O 20 40 60 80

Figure 4. CZE of tryptic digests of B-casein. Conditions: 75 pm, 50 cm long capillary, bathed in 80 mM phosphate buffer, pH 2.0. Sample application: by hydrostatic pressure for 5 s. Separations were per- formed at 150 V/cm (current: 85 LA) and detection was at 214 nm. The three major peaks are: (1) p l 6.1, fragment B-CN (114-169); (2), p l 6.93, fragment B-CN (49-97) and (3), p l 3.95, fragment P-CN (33-48). Note that the total running time is > 70 min.

not possible to perform the CZE run at voltage gradients higher than 150 V/cm, since even at this relatively modest field strength the current is already close to 90 pA, The situation is ameliorated in isoelectric buffers. As shown in Fig. 5A, in 50 mM IDA, in presence of 15% TFE, the analysis time is reduced to 50 min at a voltage gradient of 300 V/cm and to 40 min at 400 V/cm in 50 mM IDA, 0.5% HEC and 10% TFE (Fig. 5B). Never- theless, neither of the two latter buffer systems appears to be ideal, since some of the peaks are only partially resolved and some tailing is evident, suggesting still some interaction with the wall. The system that finally gave us the best results is shown in Fig. 6: 50 mM IDA, in 0.5% HEC and 6 M urea, could afford base-line reso- lution of most peptides with markedly improved peak sharpening and was compatible with a 500 V/cm field strength. In this last case, we also performed an analysis on the reproducibility of the elution profile. The transit times of the five major peaks in Fig. 6 were measured in a series of six sequential runs and statistically analyzed; the average relative standard deviation was found to be 0.7%.

4 Discussion

4.1 On the physico-chemical parameters of IDA

IDA represents a new buffer which could have some interesting applications in CZE of peptides. However, we could not find much data in the literature on this com-

A. Bossi and P. G. Righeiti

10 20 30 40 Time (min)

50 60

10 ?O 30 40 Time (min)

Figure 5. CZE of tryptic digests of B-casein in a 75 pm ID, 50 cm long capillary, bathed in (A) 50 r n M isoelectric IDA and 15% TFE or (B) 50 mM IDA, 10% TFE and 0.5% HEC. (A) Run at 300 V/cm (current: 10 VA); (B) 400 V/cm (current: 15 VA). Detection at 214 nm.

pound, neither on its p l value nor on the pK values of the two carboxyls active at the pl. Only two compounds relatively close to IDA have been reported by Hjertén et al. [6]: N-cyciohexyliminodiacetic acid (pI 1.92, pK, = 1.62 and pK, = 2.22) and N-( 1-carboxycyclohexy1)imino- diacetic acid (pl 2.11, pK, = 1.63, pK2 = 2.59). The first compound would have been ideal, probably, for peptide separations, due to its extremely low pI and unique ApK value (a ApK of 0.6 guarantees and extraordinary @ power; according to Rilbe [24] such compounds might not even exist!). Neither compound, however, couid be

8

6 h c r c! o 4

e (u

K m O u>

9 2

O

Elccrrophoresrs 1997. IR. 2012-2018

5 10 15 20 25 30 35 40 Time (min)

Fiigur? 6. CZE of tryptic digests of Ij-caseiti in a 75 prn ID, 50 cm long capillary, bathed in 50 mM isoelectric IDA together with O 5% HEC and 6 M urea. Run at 500 V/cm (current: 16 wA). Detection at 214 nrn.

found in chemical catalogues. On the contrary, IDA is easily available from a number of suppliers and quite inexpensive. We thus set out to provide its physico- chernical characterization. According to our data (Fig. 1) the true p l value should be close to 2.23. However, it was still necessary to find the pK values of the íwo car- boxylic groups. As shown in Fig. 7, these could be de- rived as follows: a theoretical @ power has been calcu- lated, with the aid of the cornputer program of GiafTreda et al. [26], by plotting the progressive p decay (at con- stant 50 mM IDA concentration) as a function of pro- gressively larger ApK values, from a theoretical ApK minirnurn of 0.6 up to ApK= 4.4. The calculated 0 power also takes into account the buffering capacity of bulk solvent (at a pH = pZ= 2.33, the molarity of free protons in the solvent is 6.4 mequiv. L-’). By confronting this theoretical p power with the one experimentally ob- tained by direct titration (p = 54 mequiv. L-‘ pII-’) one derives a value of ApK = 1. Thus, since the p l of 2.23 of IDA is the arithmetical mean of the pKs of the two car- boxyls active at its p l value, it is then a simple niatter to attribute a pK, of 1.73 and a pK2 of 2.73. Note tha.t such a compound would still have an extraordinary B power at its pZ, not so easily matched by other existing ampho- teres.

After a long search through chemical catalogs for com- mercially available “good” carrier ampholytes, S vensson- Rilbe could compile, in 1962, his classical table lkting 40 such compounds [27]. Of those, only the dipeptide His- His (pí7.30) matches IDA, with a ApK= 1 , and only one fares better: the dipeptide Asp-Asp (pI 3.04, ApK 0.64). It should be borne in mind, however, that dipeptides cannot be used as background eiectroiyte for generating peptide maps: at the concentrations needed for ensuring adequate buffering power, Asp-Asp would have a strong absorbance at 214 nm, thus obliterating the signal of the

Elecfrophoresis 1997, 18, 2012-2018 tieneration of peptide maps by CZE 2017

70 -

60-

h

';p 5 0 - i .- >

E &

3 40-

Y

2 30- R o> .- g 20- m .

analyte. The advantage of IDA, arnong others, is that it is completely transparent at al1 wavelengths. Another unique finding emerges from Fig. 7: it is seen that, as ApK progressively widens, and the hypothetical ampho- lyte (still supposed to exhibit a plof 2.33) is transformed from a "good" to a "poor" species, its apparent pl keeps increasing, up to pl2.83 for a ApK of 4.4, and this while maintaining its concentration constant at 50 mM throughout the simulation. This is quite unexpected and had not been predicted by al1 previous theories dealing with the properties of carrier ampholytes in solution. In fact, according to Rilbe [24] and to our data [17, 181, the apparent p l should only change upon dilution of the amphotere in soiution. This paradox can, however, be explained by considering the position of the p l value along the pH scale: at such acidic pZ, maintaining a con- stant concentration of ampholyte, but allowing for pro- gressively higher ApK vaiues, is equivalent, for al1 prac- tical purposes, to diluting a "good" carrier ampholyte (c$ Fig. 1). If one were to plot the progressive decay of B power of IDA upon dilution in Fig. 1, one would have the same B power decay plotted here in Fig. 7 upon pro- gressive widening of ApK. As additional proof of that, we have plotted in Fig. 8 the variations of P power and apparent p l for an ideal ampholyte having a pZ of 7.0 in a similar ApK intervai (from 0.6 to 5.0). It is seen that, although the decay law for power is the same in both cases, the apparent p l remains constant, due to the fact that, at pH 7.0, the fi power of water is negligible. It is thus seen that the apparent p l of an amphotere is far from being a constant: it varies not only as a function of its concentration in solution, but also as a function of its ApK, especially when the p l is situated in quite acidic or basic values of the pH scale.

4.2 On the generation of peptide maps by CZE

As shown in this and in previous papers [17, 181, CZE in isoelectric buffers could become a formidable tool in analysis of complex peptide maps. In the case of a- and P-globin tryptic digests, we have shown [ 181 that CZE in isoelectric Asp, with analysis times not exceeding 10 min, compares favorabiy with RP-HPLC, which typically takes up to 60 min for complete gradient elution of al1 peptides. Additionally, we have shown that, for dificult separations of peptides exhibiting a cross-over point at the p l of 50 mM Asp (2.77), one could, by diluting the amphotere, move the pH along the titration curve of the analyte, up to pH 3.1, where resolution of such peptides could be elicited [18]. Thus, isoelectric buffers appear to be unique tools, in that their selectivity can be manipu- lated by exploring a smail pH window (up to 0.4 pH units) while still using a single buffering compound. Although isoelectric Asp performed very well in our hands, it still had some problems, connected with its poor solubility at room temperature (50 mM being the highest practica] concentration) and with its incompati- bility with organic solvents (concentrations of TFE > 10% elicited massive precipitation). From this point of view, IDA offers some unique advantages: it is cornpat- ible with most hydro-organic solvents and it offers a more acidic pH window for peptide separations. At such acidic pH values (2.23 to 2.33), interaction of the pep- tides with the wall should be minimized, allowing the

. . . . O . . O . O . O

. O

O . a O .

O 0. O

O lo i o

O

0

O

O

O

1 3'00

1 2.40

- a F! I

a a O

O O o o 2.20 O I I I , , , , , I

O 1 2 3 4 5 A PK

Figure 7. Variation of buffering power (B) and apparent p l for an ideal ampholyte having a starting p l value of 2.33 (at 50 mM concentration) and progressively higher ApK values, from 0.6 to 4.4. Note that, as the ApK progressively widens and B diminishes, the apparent p l is shifted to higher values on the pH scale, although the concentration of the amphotere is kept constant. Simulation performed with the program of Giaffreda et al. 1261.

" O F 1"'

9 --+ C) -1 .-

8'6' 1 1 4 ' 1 ' 8 ' 2 ' 2 2'6 3 3 4 ' 3 ' 8 ' 4 2 ) d 6 4 PK

+ l 3 * p l

Figure 8. Variation of buffering power (B) and apparent p l for an ideal ampholyte having a starting p l value of 7.00 (at 50 mM concentration) and progressively higher ApKvalues, from 0.6 to 5.0. Note that, as the ApK progressively widens and B diminishes (with the same decay law as in Fig. 7). the apparent p l remains constant. Simulation performed with the program of Giaffreda et al. [26].

use of naked capillaries, as adopted in the present report. Additionally, we should like to emphasize that by far the best solvent for peptide separation appears to be a plain solution of 50 mM IDA in 0.5% HEC in presence of 6 (or 7) M urea. As shown in Fig. 6, this buffer offers the best separations of P-casein peptides, apparently without any binding of such peptides to the silica wall (when this occurs, a dramatic worsening of the elution profile results; see Fig. 2 in 17). In addition, theoretical model- ing and experimental evidence has shown that there are two major indicators in the electropherogram upon adsorption of the analyte to the wall: peak tailing and lack of return of the zone profiles to the original base- line [28, 291. These two indicators seem to be absent in the elution profile of Fig. 6. It is curious that, arnong the many reports on peptide separations by CZE, utilizing al1 sorts of additives (as reviewed in [IS]), there does not

2018 A. Bossi and P. G. Righetti Electrophoresis 1997, 18, 2012-2018

appear to be any mention of the use of 6 M (or 8 M) urea as a solvent for such peptides. Yet urea is the ideal solvent for proteins and peptides alike: it is a classical hydrogen bond breaker, which unfolds proteins while keeping them in solution. The major drawback of urea is the potential carbamylation of free amino residues, which readily occurs at the alkaline pH values typical of zone electrophoresis. However, in the case of peptides, since these separations are typically carried out at very acidic pH values, the danger of carbamylation simply does not exist.

P.G.R. is supported by grants from Consiglio Nazionale deile Ricerche (CNR, Roma, Progetto Strategico Tecnoiogie Chimiche Innovative No. 96.05076.ST74 and by Comitato Ecnologico, No. 96.0I895.CTII), by Agenzia Spaziale Ita- liana (ASI, Roma, grant No. ARS-96-214) and by MURST.

Received May 12, 1997

5 References

[ l ] Mandecki, W., Hayden, M., DNA 1988, 7, 57-62. 121 Righetti, P. G., Isoelectric Focusing: Theory, Methodologv and Appli-

cations, Elsevier, Amsterdam, 1983. [3] Bier, M., Egen, N. B., Twitty, G. E., Mosher, R. A., Thormann, W.,

in: King, C. J., Navratil, J. D. (Eds.), Chemical Separations: Princi- pies Litarvan Literature, Denver 1986, pp. 133-151.

[4] Bier, M., Long, T., J . Chromarogr. 1992, 604, 73-83. [SI Bier, M., Ostrem, J., Marquez, R. B., Electrophoresis 1993, 14,

16) Hjertén, S., Valtcheva, L., Elenbring, K., Liao, J. L., Electro-

[7] Blanco, S., Clifton, M. J., Loly, J. L., Peltre, C., Elecfrophoresis

1011-1018.

phoresis 1995, 16, 584_594.

1996, 17, 1126-1133.

181 Gelfi, C. , Perego, M., Righetti, P. G., Electrophoresis 1996, 17,

191 Stoyanov, A. V., Gelfi, C., Righetti, P. G., Electrophoresis 1997, 18, 1470-147s.

717-723. [lo] Ingram, V. M., Biochim. Biophys. Acta 1958, 28, 539-545. [ll] Hunt, J. A., Ingram, V. M.. Biochim. Biophys. Acta 1958, 28,

[12] Ingram, V. M., Biochim. Biophys. Acta 1959, 36, 402-411. [13] Castagnola, M., Messana, I., Rossetti, D. V., in: Righetti, P. G.

(Ed.), Capillaiy Electrophoresis in Anaiytical Biotechnology, CRC Press, Boca Raton 1996, pp. 241-275.

[14] Bello, M. S., Capelli, L., Righetti, P. G. , J . Chromatogr. A 1994,

[15] Bello, M. S . , Zhukov, M. Y., Righetti, P. G., J. Chromatog,r. A 1995,

[16] Ermakov, S . , Zhukov, M. Y. , Capelli, L., Righetti, P. G., J . Chroma-

(171 Righetti, P. G., Nembri, F., J . Chromatogr. A 1997, 772, 203-211. [18] Capelli, L., Stoyanov, A., Wajcman, H., Righetti, P. G. , J . Chroma-

[19] Bossi, A., Righetti, P. G., Riva, E., Zerilli, L., Electrophoresis 1996,

[ZO] Castagnola, M., Cassiano, L., Messana, I., Paci, M., Rossetti, D. V., Giardina, B., J. Chromatogr. A 1996, 735, 271-281.

[21] Arnold, F. H., Bio/Technologv 1991, 9, 151-156. [22] Van Slyke, D. D., J. Biol. Chem. 1922, 52, 525-536. [23] Luzzana, M., Dossi, G. , Mosca, A,, Granelli, A,, Berger, D.,

Rovida, E., Ripamonti, M., Musetti, A., Rossi-Bernardi, L., Clin. Chem. 1983, 29, 80-85.

[24] Rilbe, H., p H and Buffer Theory: A New Approaci'i, Wiley, Chichester 1996, pp. 47-49.

[25] Righetti, P. G., Nembri, F., Bossi, A., Mortarino, M., Biotechnoi. Progr. 1997, 13, 258-264.

1261 Giaffreda, E., Tonani, C., Righetti, P. G., J . Chromatogr. 1993, 630,

[27] Svensson, H., Acta Chem. Scand. 1962, 16, 4 5 6 4 6 6 . 1281 Conti, M., Gelfi, C., Bianchi-Bosisio, A., Righetti, P. G., Electro-

[29] Zhukov, M. Y., Ermakov, S. V., Righetti, P. G., J. Chromatogr. A

5 4 6 5 4 9 .

684, 311-320.

693, 113-130.

togr. A 1995, 699, 297-313.

togr. A 1997, in press.

17, 1234-1241.

3 13-327.

phoresis 1996, 17, 1590-1596.

1997, 766, 171-185.