High-performance liquid chromatography of proteins

ANALYTICAL

BIOCHEMISTRY

103, l-25 (1980)

REVIEW High-Performance

Liquid

Chromatography

of Proteins

FRED E. REGNIER AND KAREN M. GOODING Department

of Biochemistry,

Purdue

University,

West Lafayette,

Indiana

47907

Received August 1, 1979 This paper presents a review of recent studies on high-performance liquid chromatography (hplc) of proteins in gel permeation, ion exchange, reversed phase, normal phase, and affinity modes. The discussion is generally oriented toward the nature of column packing materials and the effects of mobile phase composition, separation time, support pore volume, column length, column diameter, sample loading, and solute molecular weight on resolution. Applications of hplc techniques to the separation of a variety of peptides and enzymes of interest to basic research and clinical laboratories are also presented. The design and use of postseparation chemical reaction devices for the detection of peptides and enzyme activity is discussed in a section on detectors. The relative merits of packed bed, segmented flow, and capillary reactors is examined with respect to band spreading and reaction time. The review concludes with a discussion of future trends in hplc of proteins.

will examine the application of hplc techniques to the separation of proteins. Obviously, rapid chromatographic separations of proteins require that mobile phases be forced through microparticulate columns under pressure. This requires mechanical stability at high mobile phase velocity which cannot be achieved with most classical gel type support materials. A suitable column packing material for hplc columns must be semirigid to rigid. Additionally, the material must be macroporous to allow sufficient molecular penetration for gel permeation and have adequate surface area for high loading capacity in ion-exchange, reversed phase, and affinity chromatography. An attractive feature of new high-performance support materials would be that their elution protocols be similar to those of the classical supports. With the development of support materials that meet these criteria, the hplc of proteins becomes possible. A brief discussion of factors governing resolution of proteins and other molecules may be found in the Appendix.

I. INTRODUCTION

Chromatography has played a prominent role in the development of biochemistry. The introduction of paper, thin-layer, gas, gel permeation, ion-exchange, and affinity chromatography into the biochemical research laboratories of the world each brought a surge of new discoveries. In the past decade we have seen the development of still another type of chromatography, “high-performance liquid chromatography” (hplc).’ Actually hplc is not so much a new type of chromatography as a new way of looking at old chromatographic techniques. This paper ’ Abbreviations used: hplc, high-performance liquid chromatography; gpc, gel-permeation chromatography; hp-gpc, high-performance gel-permeation chromatography; CPG, contolled porosity glass; BSA, bovine serum albumin; CPK, creatine phosphokinase; LDH, lactate dehydrogenase; rplc, reversed-phase liquid chromatography; LRF, luteinizing hormone-releasing factor; hplac, high-performance liquid affinity chromatography; LADH, liver aldehyde dehydrogenase; CM, carboxymethyl; mplc, medium-performance liquid chromatography. 1

OOO3-2697/8O/O%OOl-25$02OO/O Copyright All rights

0 1980 by Academic Press, Inc. of reproduction in any form reserved.

REGNIER

AND

II. GEL PERMEATION

Theoretically, gel-permeation chromatography (gpc) is the simplest and most predictable chromatographic method. Solutes are separated by size with the large excluded molecules eluting first and the small, totally included molecules eluting last. In practice, however, many extraneous mechanisms such as adsorptive, hydrophobic, and ionic effects may affect the retention of a solute. Since the physical and chemical properties of proteins vary so widely, it is inevitable that some will interact with any given support. It is important, therefore, to identify the mode of interaction and either eliminate it or take advantage of it when useful. In gross mechanistic terms, proteins are resolved in gpc by partitioning between the mobile phase and the stationary liquid within the pores of a support. The total diffusion volume (V,) available to a small molecule such as glucose is V, + Vi. VO is the liquid volumn in the interstitial space between particles and Vi is the volumn contained in the support pores. Matrix-associated mobile phase is not available to solutes and therefore not included in VT. In true gpc the minimum elution volumn is V, and the maximum, V, + Vi. Solutes partially penetrating support pores have elution volumes (V,) described by the equation

III

V, = V, + K,Vi,

where KD is a distribution coefficient that may range from 0 to 1. More advanced treatments of the theory of aqueous gpc and the derivation of resolution equations to evaluate column efficiency are to be found in the works of Ackers (l), Scott (2), Kirkland (3,4), Giddings (5), and Unger (6). A. Surface

Modified

Inorganic

Supports

The controlled pore silica and glass supports developed in the past decade are the foundation of most high-performance gel-permeation chromatography (hp-gpc).

GOODING

Haller devised a method (7) to prepare controlled porosity glass (gpc) by an etching process that produces selected pore diameters from 40 to several thousand angstroms with less than 10% distribution in pore diameter. A second macroporous support, porous silica microbeads (Zorbax), is made by agglutination of submicron silica particles to form 5- to IO-pm supports (4,8-10). The pores, which range from 60 to 3500 A in diameter, are actually the spaces between fused microparticles. Unger developed a third type of controlled porosity silica (LiChrospher) by emulsion polymerization of polyethoxysilane (11). These particles normally have pores with diameters up to 300 A, but pores up to 30,000 A may be produced by calcining after mixing with NaCl. A fourth controlled porosity support, Porasil, is available in 60- 1500 A pore sizes. The process by which this material is produced is proprietary. A variety of other macroporous ceramics and silicas are commercially available but have not gained acceptance in gpc because their pore distributions are two broad or they are not available in appropriate pore and particle sizes. The direct use of inorganic supports for gpc of proteins has been of limited success. Negatively charged silanol groups on support surfaces either adsorb cationic species (12) or repel anionic molecules, excluding them from the pores (13). The use of mobile phases containing high salt concentrations only partially alleviate these problems. The deleterious effects of the inorganic surface have been overcome by surface modification with organosilane bonded phases. During the bonding reaction, surface silanols are derivatized and a neutral hydrophilic surface is created that imbibes water and neither adsorbs nor repels proteins. The two most widely used bonded phases are glycerylpropyl and N-acetylaminopropyl silane. The glycerylpropyl bonded phase support is commercially available on either glass (Glycophase-CPG)

HIGH-PERFORMANCE

LIQUID

CHROMATOGRAPHY

or silica (SynChropak GPC) (14- 16). Although the glycerylpropyl coating is also available from E. Merck on a microparticulate silica known as LiChrosorbDIOL, the manufacturer does not recommend this support for gpc. The N-acetylaminopropyl bonded phase supports (17) are not currently commercially available. These two bonded phase supports are comparable in terms of resolution and protein recoveries. The microparticulate glycerylpropyl supports have been used in the determination of bilirubin (18), food-related proteins, nucleic acids (19), and albumin-drug complexes (20). Further use of this bonded phase on macroparticulate glass has been reported in the analysis of keratin (21), denatured proteins (22), and intact proteins (14,23,24). Suflicient work has now been done in highperformance gel-permeation chromatography with various support materials that it is possible to predict how changes in mobile phase velocity, support pore volume, column length, column diameter, sample loading, viscosity, and solute molecular weight will affect resolution. A 15-min resolution of components in a protein mixture is seen in Fig. 1. The column used in this separation was packed with a microparticulate (10 pm) glycerylpropyl bonded phase silica having 100-A pores. Figure 2 shows the molecular weight calibration curve for both this column and a 500-A pore-diameter column. The linear relationship between the log of molecular weight and solute distribution coefficients (K,) in the region of the plot where proteins are penetrating the support pores establishes that the columns are operating in an exclusion mode. One of the central questions in the hplc of proteins is how fast separations may be achieved. As would be expected, the price that one must pay for high speed gel permeation analyses is loss in resolution. Figure 3 shows the relationship between mobile phase velocity and column efficiency. In this

COLUMN: MOBILE PHASE: FLOWRATE:

OF PROTEINS

Synthropoh 25 x I cm

GPC 100

0.1 M K,HPO,,

7

PH

I ml/min

L 4 am

I 0

I 5

I IO TIME

I I5

(mm)

FIG. 1. Resolution of proteins on a gel-permeation column. Column: SynChropak GPC 100, 25 x 1 cm. Mobile phase: 0.1 M K,HPO,, pH 7.0. Flowrate: 1.0 mUmin. DNA, Fib, BSA, CHYM, CYT, and GT represent deoxyribonucleic acid, fibrinogen, bovine serum albumin, chymotrypsinogen, cytochrome c, and glycyltyrosine, respectively.

case efficiency is expressed as theoretical plate height (H). Since mobile phase velocity in millimeters per second is more abstract than analysis time, the relationship between H and total elution time for VT on a 25-cm column is also provided in this figure. There are two important points to be observed. The first is that loss in resolution (increased H) is substantially greater for large molecules such as BSA than small (glycyltyrosine). This is due to the smaller diffusion coefficient of macromolecules and the effects of diffusivity on H (see Appendix). The second point is that mobile phase velocities of greater than 0.5 to 1 mm/s result in large losses in resolution with large molecules. Although it is possible to propel mixtures through 30-cm permeation columns in 2 min or less, only very crude separations are to be expected. A theoretical analysis of the effect of column length on resolution in fixed time analyses may also be obtained from Fig. 3.

REGNIER AND GOODING

0

.2

.4

.6

.8

1.0

0

KD

.2

.4

.6

.8

I .o

KD

FIG. 2. (a) Molecular weight calibration curve for proteins on SynChropak GPC 100 using 0.1 M KH,PO,, pH 7, as mobile phase with 0.5 ml/min flow rate. (b) Molecular weight calibration curve for proteins on SynChropak GPC 500 using 0.1 M KH,PO,, pH 7, as mobile phase with 0.5 mUmin flow rate.

Since this figure makes it possible to calculate the number of theoretical plates (N) in any length of column operating at a given mobile phase velocity and it is known that resolution

h

1

14

IO

ItI

76 5 ANALYSIS

I

&ME

L

(t??in)

2

FIG. 3. The plate height versus velocity curve for gel-permeation column. The column was 10 x 250 mm packed with SynChropak GPC 100 and operated with a mobile phase of 0.1 M phosphate buffer (pH 7.0).

is proportional to No.5 (see Appendix), the relative resolution of different length columns in a 15-min analysis was calculated and is presented in Fig. 4. It will be noted in this figure that increasing the column length from 25 to 50 cm would only produce a 7% increase in resolution of molecules similar to BSA. When these same calculations were done with a 7-min analysis time, the 50-cm column is actually less efficient than the 25-cm column in the resolution of large molecules. On the basis of resolution, separation time, cost, and ease of operation, it appears that 25- to 30-cm columns are optimum for proteins over 20,000 daltons. Column diameter also influences resolution. It has been our experience with lo-pm supports that column efficiency continues to increase as you go from 4- to lo-mm column diameters. This effect is due in part to wall effects. Knox et al. have shown in partition systems that when a solute approaches the column wall, serious band spreading results (25,26). The use of small particles, wide columns, and limited length can result in columns of “infinite diameter” where wall effects are negligible. Unfortunately, wide bore columns are more expensive in terms

HIGH-PERFORMANCE 60

LIQUID

-

70 -

60-

50 -

zz

40

-

30 -

20 -

IO -

I L km1

FIG. 4. The influence of column length on total theoretical plates in a 15min gel-permeation analysis. The data in this figure were derived from Fig. 3.

of both column hardware and packing material. Here again, cost must be weighed against resolution. Another variable to be considered in column operation is sample load. It has been determined experimentally with chymotrypsinogen (28) that loading capacity (C) in milligrams may be expressed by the formula rV4.4 = C where r is the radius in millimeters. Thus, a 4.2-mm column would have a loading capacity of 1 mg/injection. Greater loading results in a slow decline in resolution that requires the analyst to determine if the separation of components of interest is still sufficiently large to be of value. Column loading capacity also decreases with increasing solute molecular weight. This phenomenon is probably related to increases in intrinsic viscosity of large molecules. Other than separation time, the most notable difference between high-performance permeation chromatograms (Fig. 1) and those obtained on classical gel columns is the ratio of the total pore volume (Vi) to the column void volume (If,). A Sephadex G-200 column

CHROMATOGRAPHY

5

OF PROTEINS

might have a Vi/V0 ratio of 3 while those of hplc supports vary from 0.6 to 1.3. It is apparent that fitting the same number of peaks in the chromatogram between V,, and V, requires that the peaks be much sharper in the hplc column. Since the Vi/V, ratio of a permeation column is equal to the V,/V, ratio in resolution Eqs. [2] and [6] presented in the Appendix, it can be seen that an increase in the Vi/V0 ratio of a column results in increased resolution. The relationship between Vi/V, ratio and resolution is seen in Fig. 5. The implication of this figure is that if two columns have the same number of theoretical plates, the column with the greater pore volume will have the greater resolution. Since commercial support materials vary in pore volume, this is an important parameter to note in column selection. Mobile phase selection is always important in chromatographic systems. Surface-modified inorganic supports are very versatile for gpc in that they may be operated with buffers in the analysis of proteins or with organic solvents such as tetrahydrofuran in

I

I

1

,

I

I

0.4

0.6

1.2

1.6

2.0

2.4

1

2.6

Vi/V,

FIG. 5. The effect of column permeability (Vi/V,) on the number of plates (NJ required to given a resolution of 1.0 at an (Yvalue of 1.25 at KD = 0.5 in gel-permeation chromatography.

6

REGNIER

AND GOODING

the separation of polystyrenes and other synthetic polymers (27). The primary limitation on mobile phase selection is pH. Silica is degraded at pH values greater than 8 and column life is shortened at high pH. Halide ions should also be avoided when possible due to their corrosion of the stainless-steel parts of the liquid chromatograph. The most common nonpermeation contributions to retention on surface-modified inorganics are adsorption on surface silanols, ion exclusion, and partitioning with the bonded phase. There are always residual silanols, even on a well-coated support. As a column is used, the number of free silanols will slowly increase due to gradual erosion of the bonded phase. The use of 0.1 M buffer for proteins eliminates most of this adsorption (24). The use of a mobile phase conditioning column before the inlet can reduce and even replace the loss of bonded phase from the analytical column (28). Mobile phase additives such as ethylene glycol also diminish adsorption on aged columns (29). Occasionally when columns are eluted with low ionic strength buffers, ionic molecules elute in the void volume. This effect on ionic species is often due to ion exclusion (13). Ion exclusion is the repulsion of ionic solutes from pores containing like charge. This results in molecular weight measure-

-C

-C02CH&H202C-

1’

CH,

ments that are erroneously high because the molecules are totally excluded regardless of size. The use of a salt in the mobile phase eliminates this problem for proteins and other macromolecules. As noted above, the presence of an organic bonded phase presents the possibility of partitioning the solute between the stationary and mobile phases. Mori examined this phenomenon and found that when the stationary and mobile phases are of quite different polarity, solute partitioning may occur (30). In the case of the glycerylpropyl phase operating with an aqueous mobile phase (less than 0.5 M), partitioning has never been found with a protein but frequently occurs with aromatic amino acids and peptides (3 1). Partitioning or adsorptive effects, if not recognized, may yield deviant calibration curves (32). Alternatively, when n-propanol is used as the mobile phase, substantial partitioning of serum proteins occurred (33). B. Rigid Organic Gels

Two rigid organic gels, both of which are polymeric methacrylate ester matrices, have been successfully used in the gpc of water soluble polymers. One of these materials (Spheron), shown below, was developed in Czechoslovakia and is currently available in West Germany.

C CH, T-r

The dextran exclusion limits range from 103for Spheron P, to 107for the P1O,ooomaterial (34). The other methacrylate-based support (Shodex) is available in Japan in a variety of pore diameters and exclusion limits. The general chemical formula of this material is unknown, but it is thought to be an ester of glycerol (35). At the present time no

m

HZ-C-C01CH2CH20H I CH,

n

applications of the Shodex gel in protein separations have been published, but the material has been used in the resolution of aqueous soluble polymers. The Vi/V0 ratio of these materials is in the range of 1.1 to 1.3. The pressure limitations of these supports have not been published.

HIGH-PERFORMANCE

C. Inorganic-Organic

LIQUID

CHROMATOGRAPHY

Composites

5

Toya Soda Corporation of Japan has recently introduced a new support for gpc of water-soluble polymers including proteins (36,37). No information is available on either the chemical or the physical composition of these supports. However, based on combustion of the support, it contains large amounts of both organic and inorganic materials (38). Figure 6 shows good resolution on an 8 x 500-mm column. The fact that many compounds elute after the 45 ml totally included volume (V,) indicates that the column is not functioning in the exclusion mode with all compounds. Some solutesupport interaction must be occurring. III. ION-EXCHANGE

I

20

25 30 ELUTION

In gel-permeation chromatography, differential permeation of solutes into the support matrix is the basis for separation. Any interaction with the support surface is undesirable. Ion-exchange chromatography, on the other hand, is dependent on the ionic interactions between species on the support surface and charged groups in the protein. The capacity factor (k ‘) for a small molecule in an ion-exchange system is

PI where Kiec is the ion-exchange partition coefficient andA, is the total surface area of the support. AT may be further subdivided into external (A,) and internal (Ai) components in a porous material. It is apparent that with macromolecules which only partially penetrate support pores, the available ionexchange surface area (A,) is given the formula

k&c = Kiec

(Ae + AiKp) VIII .

[41

Since Ai is usually !90- 100 times larger than A,, penetration of the ion-exchange support becomes a significant factor in retention. It will be noted in Eq. [4] that differences in K, should make it possible to separate molecules that have identical K+c values. The penetration constant may also be shown to influence the ion-exchange capacity of a support. Solute ion-exchange capacity (I,) in milligrams per gram of support may be related to available surface area (A,) expressed in squared angstroms per gram and the surface area occupied by a single molecule (A,,,) through the equation I, =

A,M,

x 103

A,N

’

151

where M, is the molecular weight of the solute and N is the Avogadros number. By combining Eqs. [3] and [5] one obtains an equation,

[31

where K, is a penetration constant that indicates the fraction of Ai available to the molecule. Combining Eq. [2] and [3] it is seen that

50

35 40' 45 TIME (mid

FIG. 6. Separation of human serum. Column: G 3000 SW. Solvent: 0.02 M acetate buffer (pH 6.0) containing 0.15 M NaCI. Pressure: 90 kg/cm2. Flow rate: 1.0 mlimin. Charge: 30 ~1. Detector: UV 280 nm, 0.16 [ABS] (10 mV). Room temperature. Reprinted from K. Fukano, K. Komiya, H. Sasaki, and T. Hashimoto (1978) J. Chromatogr. 166, 47-54 by courtesy of Elsevier Scientific Publishing Co.

CHROMATOGRAPHY

Aa = A, + AiKp,

7

OF PROTEINS

I = (Ae + AiKp)

c

AIll

X-9

1094, N

[61

that relates ion-exchange capacity to K,. Since A, =+ Ai, it is obvious that the ion-

REGNIER

AND GOODING TABLE

ION-EXCHANGE

CAPACITIES

Inorganic support”

Particle size OLm)

Pore diameter (4

CPS CPG CPG CPS CPG CPG CPG CPG CPG CPG

20-30 37-74 37-74 37-14 37-74 74- 128 74- 128 74- 128 74- 128 74- 128

170 250 550 500 550 100 250 550 100 250

1

AND MICROANALYSIS

OF GLYCOPHASE

SUPPORTS

Surface

area W/g) 150 130 70 70 170 130 70 170 130

Stationary phaseb

Organic carbon (% by weight)

CM CM CM CM SP DEAE DEAE DEAE

6.69 5.71 4.09 4.21 4.03 6.95 5.97 4.36 6.37 4.26

QAE QAE

Ion-exchange capacityc (m&4 31 48 27 25 18 34 45 20 28 36

n CPS and CPG designate controlled porosity silica and glass, respectively. b The abbreviations CM, SP, DEAE, and QAE represent carboxymethyl, sulfonylpropyl, diethylaminoethyl, and quatemary ammonium exchange group, respectively. c The ion-exchange capacity is expressed in mg hemoglobin bound per cc of support.

exchange capacity of a support for molecules a series of proteins were nonspecifically adsorbed and denatured. Regnier later noted that do not penetrate is very small. Hemoglobin ion exchange capacities for a that the denaturation of chymotrypsin by series of supports are presented in Table 1 these alkylamine bonded phase supports (39). Although the 100-A pore support has could be eliminated by crosslinking adthe highest surface area and the 500-A the jacent amino groups with butadiene diepoxide least, the 250-A pore diameter support has (15). The properties of this crosslinked the highest ion-exchange capacity. These surface coating led to the development of data are easily explained in terms of K,. two different families of porous pellicular Due to the small penetration of hemoglobin ion-exchange supports. into the 100-A pore support, the macroThe first family of ion-exchange supports molecule could reach little of the total was prepared by copolymerizing epoxysurface. Conversely, the 500-A support monomers on the surface of glycerylpropyl allows good penetration but has too little silane bonded phase supports (19,39,41,42). A *. The 250-A pore support is intermediate A continuous film of epoxy polymer was in both KP and AT but is a compromise that formed that was linked to the surface produces maximum Z,. This phenomenon through the glycerylpropyl silane bonded will be true for all macromolecules with the phase. Stationary phases were either copore diameter of maximum Z, increasing polymerized into the coating or bonded later. with molecule size. Since the principle structural unit in the coatings was glycerol, the coatings were A. Surface-Modified Znorganic Supports called “Glycophases.” A general chemical Eltekov was the first to carry out protein formula for the coating as well as the separations on surface modified inorganic various P, groups that were used (diethylanion-exchange supports (40). Although the aminoethanol (DEAE), carboxymethyl, suly-aminopropyl silane bonded phase used in phonylpropyl, and quarternary ammonium exchangers) are presented below: these studies gave ion-exchange separations,

HIGH-PERFORMANCE

LIQUID

CHROMATOGRAPHY

OI = SiCH,CH,CH,(OCH,CH-CH,),

OF PROTEINS

OH I -(OCH,CHCH,),-(OP,), -OCH,CO,H,

where P, = -OCH,CH,N(CH,CH,),, -OCH2CH2CH2S03H,

-OCH2CH2ti(CH2CH3)2.

I CHs Application of the diethylaminoethyl (DEAE) anion-exchange support to the separation of tissue creatine phosphokinase (CPK) and lactate dehydrogenase (LDH) isoenzymes is seen in Figs. 7 and 8 (19). Enzyme activity was monitored with a postcolumn reaction detector to be described later. The separation of LDH isoenzymes by hplc has also been reported by Toren et al. (43-45) and the analyses of arylsulfatase isoenzymes by Bostick et al. (46).

The second family of ion-exchange supports developed for protein separations is unique in that the coatings are not covalently bonded to the surface of the support. The anion-exchange coating is prepared by adsorbing a layer of low molecular weight polyethylene imine to the support surface from organic solvent and then crosslinking the adjacent amine groups on the surface to form a skin or pellicle (47). Crosslinking Iroonzymu

SAMPLE:

CPK

PACKING:

DEAE

Glycophou/CPG s-top

25011 per.. 250

COLUMN:

4 x

!3OLVENf:

A = 0.05

mm M fris,

SS 0.05

M

NoCl , IO-’ Y Morcoptoothono~

,

pHa7.5 B=O.OS M Trir, 0.3 M NO’S, IO-’ M Morcoptoothono~. pli - 7.5 FLOW

RATE:

2500

DETECTOR:

340

PEAK IOENT,TY:

TIME

4mm/uc

PRESSURE:

(3mVmin)

psi nm

a = CPK, b = CPK, c = CPK,

(min.)

FIG. 7. Separation of CPK isoenzymes on DEAE Glycophase/CPG with a postcolumn enzyme detector. Reprinted from S. H. Chang, K. M. Gooding, and F. E. Regnier (1976)J. Chromatogr. 125, 103- 114 by courtesy of Elsevier Scientific Publishing Co.

10

REGNIER

AND GOODING 3AMPLE:

LDH

PACKING:

DEAE

COLUMN: SOLVENT:

Irornzymrr GlycoOhow

/CPG

2sOA Pm. S-IO/L 4x250 mm ss A = 0.025 M Trig. pH G = 0.025 M Trir. 0.2

8 0.0 M

NaCl , pH * 8.0 FLOW

RATE:

PRESSURE: DETECTOR: PEAK IDENTITY:

4 mm/arc

(3 ml/min)

2300 psi 340nm o b c d a

= = = = =

LDHJ LDH, LDH LDH’ LDH;

OOW

TIME

(min.)

FIG. 8. Separation of LDH isoenzymes on DEAE-glycophase/CPG with a postcolumn enzyme detector. Reprinted from S. H. Chang, K. M. Gooding, and F. E. Regnier (1976)J. Chromatogr. 125, 103- 114 by courtesy of Elsevier Scientific Publishing Co.

agents such as multifunctional oxiranes produce very stable pellicular matrices through carbon-nitrogen bond formation. The anion-exchange support is commercially available as SynChropak AX and provides separations similar to those obtained on DEAE-cellulose. Resolution of some of the components in a commercial Ovalbumin gelpermeation calibration standard are shown in Fig. 9 (28). An hplc instrument equipped with an anion-exchange column could replace electrophoresis for many analyses in the clinical laboratory. Isoenzyme profiles have been quantitatively determined in less than 15 min to detect myocardial infarction (40). Further use of this System for profiling the hemoglobin variants in blood (48) has been reported. The hemoglobin profile of a patient with p-thalassemia trait is seen in Fig. 10. Protein recovery is also an important issue with any new chromatographic technique. Although the question has not been examined

exhaustively, recovery of enzyme activity and protein from either of the families of ion exchange supports discussed above has never been reported to be less than 85% (41,47). Recoveries were equal to or greater than those obtained from DEAE-cellulose in all cases tested. Although they have not been used for protein separations, the Partisil SCX columns have been used in peptide fractionation (49). Unfortunately, the use of hplc ionexchange supports even in the fractionation of peptides has been given little attention. A systematic study of these materials is overdue. B. Rigid Organic supports Substitution of the methacrylate matrix of Spheron with ionic species has been used to prepare a series of supports for protein separations (50,51). Figs. 11 and 12 show the analysis of technical glucose

HIGH-PERFORMANCE

LIQUID

CHROMATOGRAPHY

SynChromh 2sOxOlmm

COLUMN MOBILE PHASE.

TIME

FIG.

OF PROTEINS AX 300 lb

A 0.02 M Tw AC OH 8 6~002MTwAc+ 0.5 Y NOAC on 0

FLOWRATE:

0.5 nl/min

SAMPLE:

OualBwnin Colibmtion

stonb4

hn)

9. Analysis of an ovalbumin calibration standard on a SynCbropak AX 300 column,

I

COLUMN: MOBILE PHASE:

SynChroook AX ?.O’Z 250x 4.1 mm ID A: .02M

Trir-Acelotr.

pli

8: .02 M Tris-Acelote .iM NoAcetote.

OH 8

FLOWRATE:

2.5

ml/min

PRESSURE:

1200

psi

SAMPLE:

Humon Blood A That. Troit

6

60 40

m o\’

20

TIME

(min)

FIG. 10. Analysis of the hemoglobin variants in the blood of a person with /3-thalassemia trait. The analysis was on a SynChropak AX 300 column. Reprinted from K. M. Gooding, K. C. Lu, and F. E. Regnier (1979) J. Chromatogr. 164, 506-509 by courtesy of Elsevier Scientific Publishing Co.

11

12

REGNIER AND GOODING

0

IO

20

30

F.N. FIG. 11. Chromatography of technical glucose on DEAE-Spheron. Load: 15 mg of preparation in 0.2 ml of buffer A. The ion exchange was equilibrated with buffer A. I, Buffer A without gradient. Linear gradients II (A + B) and III (B + C). IV, Buffer C without gradient. Flow rate, 2 ml/min; 4-ml fractions; temperature, 14°C; counter pressure, 3-7 atm: chart speed. 2 mm/min. Buffers: A. 0.01 M acetic acid + NaOH, pH 6.8; B, 0.3 M acetic acid + NaOH, pH 5.5; C, buffer B, 1 M in NaCl, pH 5.3. Broken line, glucose oxidase activity of effluent in Sarret units (S.U.; 1 S.U. corresponds to the consumption of 600 ~1 of oxygen at 30°C) with correction of shift of values. Reprinted from 0. Mikes, P. Strop, and J. Sedlackova (1978) J. Chromatogr. 148, 237-245 by courtesy of Elsevier Scientific Publishing Co.

oxidase on DEAE-Spheron and crude protease on CM-Spheron, respectively. Although these separations are somewhat slower than those obtained on the surfacemodified supports, it is probable that the full capabilities of these columns in terms of speed and resolution have not been totally exploited. The inherent resolution of ion-exchange chromatography, the speed with which separations may currently be achieved, the high recovery of enzyme activity and the ability of a few rather simple solvent systems to resolve a large number of proteins, suggest that high performance ionexchange chromatography has substantial potential in the future. III. REVERSED PHASE (HYDROPHOBIC) CHROMATOGRAPHY

Hydrophobic forces are ubiquitous in biological systems. Tanford (52) has proposed that the hydrophobic effect is a unique organizing force in nature that is based on solute repulsion by solvents. In the case of polypeptides, these solvophobic effects play

an important role in chain folding and the ultimate tertiary structure of proteins. Since amino acid side chains vary from polar to nonpolar and charged to neutral, the hydrophobicity of any peptide or segment of polypeptide will depend generally on the ratio of the various amino acids and pH. The net result of the thermodynamically controlled chain folding in polypeptides is often a compact globular molecule. Charged amino acid side chains are on the surface and hydrophobic groups are accumulated in surface pockets or buried in the interior. The variation in hydrophobicity of both peptides and proteins provides yet another means by which polypeptides may be fractionated. Chromatographic systems based on the partitioning of solutes between hydrocarbon stationary phases and polar mobile phases are widely used in the resolution of biological compounds. The early separations of fatty acids on hydrophobic supports (53) and the more recent application of alkyl substituted carbohydrate gels to the separation of proteins (54-56) are examples of hydrophobic chromatography. Howard and

HIGH-PERFORMANCE

LIQUID

CHROMATOGRAPHY

Martin first used the term “reversed phase” chromatography to distinguish hydrophobic chromatography from those partition chromatographies that used polar stationary phases and nonpolar mobile phases. It is unfortunate that this misnomer is now widely used in hplc. The most popular reversedphase hplc supports are porous silicas with organosilane bonded phases. The reader is referred to the literature for a more extensive discussion of the application of solvophobic theory to reversed-phase chromatography (57-60) and a review of the application of reversed-phase liquid chromatography (rplc) to biochemical separations (61). With the presence of hydrophobic amino acids in most peptides and proteins, it is not surprising that with aqueous mobile phases they are tightly adsorbed to reversed phase supports. Recent research on the hplc of peptides and proteins has been directed mainly toward the search for mobile phases that will allow selective desorption. Table 2 presents conditions for the separation of peptides from a selection of representative papers. No effort was made to include all of the literature on peptide separations in this table. A. Reverse Phase Supports

Surface modified inorganic supports used for reversed-phase chromatography are generally of three types: ethyl (C,)-, octyl CC,)-, or octadecyl (C&-bonded phases. The phenyl and nitrile-bonded phases have seen only limited use in peptide and protein separations. Early alkylsilane-bonded phase supports were prepared by reacting a trichloroalkyl silane with an inorganic support suspended in toluene. For steric reasons, residual surface silanols and halosilanes are left after this reaction. When the support was placed in water, a large number of these residual surface silanols became exposed. This problem was overcome by Roumeliotis and Unger through the use of monochlorodimethylalkyl silanes as outlined below (84):

I-

13

OF PROTEINS 44

MINUTES

-7%5z *+ 0.20 Id

0

0

IO

20

40

30

F.N

FIG. 12. Chromatography of crude protease on CMSpheron. The ion exchanger was equilibrated with buffer A. Load: 20 mg of enzyme in 0.3 ml of the same buffer. I, Buffer A without gradient. Subsequently linear gradients II (A + B), III (B + C), IV (C + D), and V (D -t E) were applied. VI, Buffer E without gradient. Flow rate, 4.07 ml/min; fractions, 4.24 ml; temperature 10°C; counter pressure, S-10 atm. Buffers: A, 0.005 M ammonia + formic acid, pH 4.5; B, 0.05 M ammonia + formic acid, pH 4.5; C, 0.25 M ammonia + acetic acid, pH 6.0; D, 0.5 M ammonia + acetic acid, pH 8.0; E, buffer D, 1.0 M in NaCI, pH 8.0. Full lines (---), record of effluent absorbance at 285; chart speed, 5 mm/min. The broken line (- --) shows proteolytic activity measured on 0.5-ml aliquots; the activity is given in absorbance units (280 nm) of trichloroacetic acid filtrate of digested haemoglobin (a proteolytic activity unit P.A.U. corresponds to absorbance 1.0). The position of the values was corrected to account for the retardation of fractions with respect to the record by a shift corresponding to the tubing volume + half the volume of the fraction. F.N. = Fraction No. Reprinted from 0. Mikes, P. Strop, and J. Sedlackova, J. Chromatogr. 148, 237-245 by courtesy of Elsevier Scientific Publishing Co.

silica = Si-OH

+ Cl-Si-R

-+

CH,

CHs silica = %-O-ii--R

+ HCl. I CHs

14

REGNIER

AND GOODING TABLE

CONDITIONS

FOR THE SEPARATION

Compound Actinomycins Bacitracin

2

OF SOME PEPTIDES

CHROMATOGRAPHY

Elution conditions (Mobile phase)

Column Pell.-f&” Pell.-C1s

BY REVERSED-PHASE

Ref.

Year

(62) (63)

1973 1974

64 165)

1975 1976

(66)

1976

Tetra- to hexapeptides

Pell.-Cl8 Pell.+

CH&N-HLO (1:l) A = 5% MeOH in phosphate buffer, pH 4.5 B = 46% MeOH, 28% CHsCN in phosphate buffer, pH 4.5 Same as Ref. (2) A = 15% acetone,O.O3% HCO,Nl& 0.01% thiodiglycol MeOH-H,O (1: 1) MeOH-H,O (70:30)

Vasopressin Angiotensin LRF Neurotensin (u-MSH Somatostatin

Cl8

20 to 33% CH,CN in0.01 M NI&OAc, pH 4.0

(67)

1976

Gramicidin Angiotensin a-MSH

Gd

A = 0.1 M phosphate buffer, pH 2.1 B = CH,CN

(68)

1977

Oxytocin Lypressin Omipressine Felypressine

C, and Ci8

17.5 to 20% CH,CN in phosphate buffer, pH 7.0

(69,70)

1977

Di- to decapeptides Di- to octapeptides

Pell.-4 GS

lo-70% CHsCN in Hz0 A = phosphate buffer, pH 2.0 B = MeOH

(71)

(72)

1977 1977

Several procedures Methanol/water (l/l) with either phosphoric acid, sodium hexanesnephonate, or sodium dodecyl sulfate

(73) (74)

1977 1978

(75)

1978

Bacitracin Oxytocin

Biologically active peptides Tri- to heptapeptides

Somatostatin LRF (luteinizing hormonereleasing factor)

Cl8 and CN’

1. A B 2. A B

= = = =

TEAP buffer, pH 2.5 40% A, 60% CH,CN TEAF buffer, pH 3.0 40% A, 60% CH,CN

Di- to decapeptides

Pell.-C,,

Many mobile phases with addition of phosphoric acid

(76)

1978

Insulin Glucagon ACTH pentaacetate

Many mobile phases with addition of phosphoric acid

(77)

1978

Di- to pentapeptides

MeOH-H,O (1:l) with either phosphoric acid, sodium hexanesulfonate, or sodium dodecyl sulfonate as a pairing agent

(78)

1978

Neurophysins

MeOH in acetate buffer, pH 5.7

(79)

1978

Oxytocin Diastereoisomers

NH,OAc,

(80)

1978

pH 4.01CHsCN (82: 18)

HIGH-PERFORMANCE

LIQUID TABLE

Compound

CHROMATOGRAPHY

15

OF PROTEINS

2-Continued

Elution conditions (Mobile phase)

Column

Ref.

Year

Insulin Cytochrome c Bovine serum albumin Catalase Ovalbumin

CM

A = 0.05 M phosphate buffer (pH 2), 2-methoxyethanol (95:.5) B = isopropanol-2-methoxy ethanol (pH 2) (955)

(81)

1978

Di- to pentapeptides

4

MeOH-HZ0 (1:l) with 2 mM tetralkylammonium acetate, pH 4, in the water

(82)

1979

Enkephalin Angiotensin Vasopressin cr-MSH Neurotensin Somatostatin Bombesin Glucagon Insulin Endorphin Calcitonin /3-Lipotrophin

G&l

A = phosphate buffer, pH 2.1 B = acetonitrile

(83)

1979

a Pell-C,, refers to a pellicular octadecylsilane-bonded phase support. b C,, specifies a porous microparticulate octadecylsilane-bonded phase support. e Pell-4 describes a pellicular phenylsilane-bonded phase support. d C, refers to a porous microparticulate octylsilane-bonded phase support. p 4 indicates the use of a porous microparticulate phenylsilane support. ’ CN designates a porous microparticulate propionitrile-bonded phase support.

This bonding chemistry results in heavier surface coverage and fewer residual surface silanols . Several manufacturers already use this technique and more will probably adapt the procedure for the C,- and C,,-bonded phases. There is still some question whether the C,-bonded phase is stable in aqueous systems. One of the first separations of peptides by rplc was that of actinomycins on Corasil/C,8 by Rzeszotarski and Mauger (62). This same pellicular support was used by Tsuji the following year to resolve the components in commercial bacitracin powders (63) and, subsequently, by a series of other investigators for the resolution of peptides. Although this support has been quite useful, the higher loading capacity and resolution of the totally porous 5- to lo-pm silica-based materials will probably lead to the displacement of the

pellicular supports. The increase in resolution gained through the use of porous microparticulate supports was demonstrated in the analysis of bacitracin powders (64). Approximately 70 times more theoretical plates/unit length were generated with the latter microparticulate supports. It has been reported that bonded phase chainlength has little effect on resolution of nonapeptides (69). This has been the general observation with many peptides; however, with higher molecular weight peptides and proteins, Rubinstein favors the use of C,-bonded phase because it is possible to elute solutes with less organic solvent (33). B. Mobile

Phase Selection

It may be generally stated that solute retention in reversed-phase systems is

16

REGNIER

AND

dependent on the hydrophobicity of the solute relative to the polarity of the mobile and stationary phases. Very polar mobile phases such as buffers favor solute retention while nonpolar solvents such as acetonitrile diminish it. To elute a hydrophobic solute from a reversed-phase column, the experimenter must either decrease the polarity of the mobile phase, increase the polarity of the solute by ion pairing, or both. Occasionally, solute retention is negligible, even in the most polar mobile phase. In this situation the solute may be made more hydrophobic by ion pairing. Thus, mobile phase selection is the proper balancing of these phenomena to achieve a controllable association of the solute with the stationary phase. It should be obvious that the ratio of water to organic solvent in the mobile phase has a major effect on k’ and that more hydrophobic solvents cause solute elution at lower organic solvent concentration. Rubinstein, in fact, favors n-propanol in the mobile phase for rplc separations of peptides and proteins because it is more hydrophobic and elutes proteins at a lower organic solvent concentration (33). For example, peptides requiring a 0 to 100% acetonitrile gradient for elution on Cl8 supports may be eluted with a 0 to 20% n-propanol gradient. In general, the larger polypeptides require higher organic solvent concentrations for elution. Rubinstein (33) states that in his studies with Lichrosorb RP-8 (C,) all proteins elute without precipitation with less than 40% n-propanol in the mobile phase. Combinations of solvents such as isopropanol and 2-methoxyethanol have also been used to elute proteins from C,, supports (81). These authors propose that the surface-active nature of 2-methoxyethanol is responsible for its utility in the mobile phase. Another reason for selecting a specific organic solvent in the mobile phase would be to control selectivity (a). However, studies with a series of peptides have shown that only minor changes in a! are

GOODING

achieved by changing organic solvents (69,83). Therefore, the choice of organic solvent may be based on secondary factors such as viscosity or protein stability. The influence of salt concentration on resolution has been examined at both pH 7 and 10 (69). In this study on nonapeptides, k’ decreased with increasing salt concentration at both pH values. This effect is thought to be similar to the “salting in” effect observed with proteins. The increasing solubility of peptides in the mobile phase at higher salt concentration is responsible for the decrease in k ’ values. Since the rate of decrease in k’ varied among peptides, conditions could be found that favor the separation of any two. Ionization of peptides also apparently has a strong influence on their retention properties in rplc. It is well known that the ionization of various groups in a peptide may be controlled by varying pH (85). The study of Krummen and Frei (69) with mobile phases in the pH range 5 to 12 showed that the k’ of tyrosine-containing peptides increased to a maximum at pH 8 and then slowly decreased with more basic conditions. The non-tyrosine-containing nonapeptide, felypressine, showed a continuing increase ink’ through pH 10. A possible explanation for the behavior of tyrpsine peptides is that ionization of the phenolic OH group in tyrosine (pK, = 10) could increase solubility in the mobile phase and decrease solubility in the stationary phase. A similar phenomenon is seen with anionic peptides. Under very acidic conditions, the carboxyls are protonated and solubility in the mobile phase decreases, increasing k ’ (76). A further elaboration of pH modification of k ’ will be discussed under ion pair formation. C. Ion Pairing

Agents

In the past few years, it has been found that the chromatographic behavior of an ionic solute may be modified by ion pairing

HIGH-PERFORMANCE

LIQUID

CHROMATOGRAPHY

with a counter ion or haeteron in the mobile phase (86-89). For example, a positively charged solute (S+) might ion pair with a negatively charged haeteron (Hi) to form a complex as shown below:

OF PROTEINS

17

a-M%

S+ + H, = S’H,. The complex will partition with the reversedphase support as an entity that has chromatographic properties different from those of the solute alone. Hydrophilic pairing agents will decrease the k’ of S+ in rplc while hydrophobic agents increase k’. Although they did not describe it as ion pairing, Molnar and Horvath (68) were the first to recognize the uniqueness of phosphate and perchlorate in the elution of peptides from reversed-phase columns. Their separation of (w-MSH peptides on a Lichrosorb RP-8 column with gradient elution from 0.1 M phosphate buffer (pH 2.1) to acetonitrile is shown in Fig. 13. It was later shown that the addition of 0.1% phosphoric acid to an acetonitrile-water mobile phase drastically reduces the k’ of peptides containing free amino groups (76). However, the addition of phosphoric acid produced the opposite effect when amine groups were blocked and all side chain functional groups protected. This apparently was due to a decrease in peptide solubility in the mobile phase with a concomitant increase in k’. The effect of phosphoric acid on basic peptides could not be accounted for simply on the basis of pH since k ’ also decreased with the addition of phosphate buffers at constant pH. Although the effect of phosphate is maximum under acidic conditions, it still causes a large decrease in the k ’ of linear antamanid at pH 7.0 (76). Rivier (90) has reported that triethylammonium phosphate is also a useful mobile phase in the elution of peptides from reversedphase columns. This mobile phase gave both high resolution and high recovery for closely related peptides such as luteinizing hormonereleasing factor (LRF), somatostatin, insulin,

1h

0

5

I I I I

10

TIME

15 (mm)

20

25

FIG. 13. Chromatogram obtained with melanotropin. Column: LiChrosorb RP-8 (5 pm octadecyl-silicia); Gradient elution from 0.1 M phosphate buffer, pH 2.1, with acetonitrile as thegradient former; gradient shape, see reference below; temperature 70°C; Row rate, 2 mbmin; initial AP, 150 atm. Reprinted from I. Molnar and C. Horvath (1977)J. Chromatogr. 142,623-640 by courtesy of Elsevier Scientific Publishing Co.

and cytochrome c. Hancock (78) has also used hydrophobic ion pairing agents to increase the k’ of peptides. The resulting increase in retention produced an increase in resolution. Complexes of anionic pairing agents (hexanesulfonate or dodfcylsulfonate) with the cationic groups (R-NHJ on peptides decrease the polarity of a peptide and increase its k ’ . The opposite strategy would be used on peptides with an excess of anionic groups. It is apparent that judicious selection of pairing agents will enable the experimenter to control both retention and selectivity for various peptides. A fourth technique using haeterons for the modification of peptide retention in rplc is to use a hydrophobic chelating agent and Zn (II). The metal ion plays a significant role in selectivity. Through the use of this technique it has been possible to resolve a series of dipeptides (91,92).

18

REGNIER

AND

D. General Comments It may be concluded from the rapid expansion of the scientific literature on peptide rplc that the technique is becoming a prominent analytical tool in the purification and analysis of peptides. However, the resolution of peptides from proteolytic digests for protein sequencing is still not widely reported. Apparently, the extreme hydrophobicity of the core peptides has presented a problem in elution. Future research will probably be directed toward determining the utility of phosphoric acidcontaining mobile phases in the resolution of a variety of proteolytic digests and the search for still other new mobile phases. The utility of reversed phase columns is somewhat more questionable for proteins. The present reversed phase supports require acidic conditions and rather large quantities of organic solvent before proteins are eluted. Many proteins are denatured under these conditions. The general utility of a chromatographic technique that uses structurally altering conditions for elution must be carefully examined in terms of the recovery of enzyme activity. Elution conditions that are less severe would be highly desirable. IV. NORMAL

PHASE CHROMATOGRAPHY

Rubinstein has recently found that proteins soluble in 80% n-propanol may be fractionated on Lichrosorb-DIOL columns by running a gradient from 80 to 50% (v/v) n-propanolsodium acetate (0.1 M) (33). All proteins eluted from the columns by the time the gradient reached 50% n-propanol. It was proposed that this technique be termed normal-phase chromatography of proteins. V. AFFINITY

CHROMATOGRAPHY

It has been suggested in a recent paper that the rapid separation of enzymes and antigens on high-performance affinity sup-

GOODING

ports be termed “high-performance liquid affinity chromatography” (hplac) (93). Using 60-A-pore-diameter silica with an adenosine monophosphate (AMP)-bonded phase, bovine serum albumin (BSA), liver aldehyde dehydrogenase (LADH), and lactate dehydrogenase (LDH) were separated in less than 3 min. After the elution of BSA with 0.1 M phosphate buffer (pH 7.5), LADH and LDH were desorbed by changing to a mobile phase of the starting buffer with 2 M NaCl (93). Approximately 90% of the enzyme activity was recovered from the support. Further use of this AMP support with an NADH gradient produced a separation of the H, and M., isoenzymes of LDH in 5 min. Through the use of immobilized anti-human serum albumin it was possible to obtain a rapid separation of human serum albumin from other serum components. This immunosorbent column was used through 20 cycles. The enormous selectivity and speed afforded by these techniques should make them a valuable analytical tool in the basic research and clinical environments. VI. ADSORPTION

CHROMATOGRAPHY

The deleterious effects of uncoated inorganic surfaces have been pointed out above. Contrary to these findings, there are also a number of reports of the fractionation of proteins on native inorganic supports. Mitzutani and Mitzutani recently reported that the anionic silanols on the surface of controlled porosity glass (CPG) gave fractions of rabbit serum and bovine parotid extracts similar to those obtained on CMcellulose (94). The adsorption of proteins on CPG was stronger than that on CM-cellulose but the separation was better on CPG. The order of eluting strength of buffers at pH 8.0 on the CPG were phosphate < Trishydrochloride < glycine. It was also reported that protein recoveries were as good as those on CM-cellulose.

HIGH-PERFORMANCE

**T

LD2

LIQUID

CHROMATOGRAPHY

19

OF PROTEINS

a

b

d

1

0

TIME,

MIN

36

6

I2

16 TIME

I

1

1

24

30

36

(mini

14. (a) Chromatogram of serum LD isoenzymes from a patient with an elevated level of LD activity. Total serum LD activity was 444 U/liter. The upper and lower traces were observed at detector 2 (downstream) and detector 1 (upstream), respectively, (b) Profile of serum LD isoenzymes resulting from the correction for background absorbance (a). The dispersion coefficient used in the computer was 0.46 s. The percentages of total area are: LD-5, 1.1; LD-4, 2.2; LD-3, 4.2; LD-3, 17.4; LD-2, 44.0; and LD-1, 31.0%. Reprinted from J. A. Fulton, T. D. Schlaback, J. E. Kerl, and E. C. Toren (1979) J. Chromarogr. 175, 283-291 by courtesy of Elsevier Scientific Publishing Co. FIG.

able chromatograms in Fig. 14. A similar system has been used for analyzing arysulfatase isoenzymes (46). A second procedure for detecting enzyme activity is by the use of packed bed reactors (19,112,113). Substrate is still mixed with analytical column effluent, but the mixture is forced through a column packed with nonporous particles. Such a reaction column provides increased reaction times with minimum peak broadening. No provision has been made for subtracting background absorbance in this system since the longer reaction time enhances the signalto-noise ratio. Analysis of a commercial alkaline phoshatase sample both at 254 nm and in the specific enzyme detector mode is shown in Fig. 15. It is seen that in the enzyme detector mode the detector produces a signal for only one protein.

A variation of the packed bed reactor is the use of immobilized enzymes (114). Since enzymes are often assayed by coupling their activity with other enzymes, the coupling enzymes must either be pumped continuously with substrates (19,112,113) or immobilized (114) to cut operational costs. An immobilized enzyme postseparation detector has been used to detect hexokinase isoenzymes and creatine phosphokinase isoenzymes after elution from a highperformance ion-exchange column (114). Although these systems are used as detectors, they are similar to the flow-through enzyme reactors described by Schifreen (115). A third type of postcolumn reaction detector which should prove suitable for enzyme detection is the air-segmented flow analyzer. These autoanalyzers (Technicon) have a long history of use in the clinical

20

REGNIER

VII. PEPTIDE-BONDED

AND

PHASE SUPPORTS

Grushka et al. bonded a series of tripeptides onto silica supports and used them to resolve dipeptides and amino acids. In the first study, the effects of the tripeptides, Gly-L-Val-L-Phe, L-Val-Ala-L-Val, and L-Val-L-Ala-L-Set-, as stationary phases were examined (95). The second paper examined the effects of mobile phase on elution characteristics for isomeric dipeptides (W). Using L-Val-L-Phe-L-Val on Partisil 10, it was found that changes in pH could reverse the elution order for dipeptide pairs. VIII. SPECIFIC

DETECTION

SYSTEMS

The detectors generally used in hplc are refractive index, uv-visible spectrophotometric, electrochemical, conductivity, and fluorescence monitors. Although they have a broad range of selectivities and sensitivities, none is specific for the peptide bond or enzyme activity. Consequently, selective determination of peptides, proteins, and specific enzymes in the presence of other compounds is not possible. To increase specificity, a chemical reaction can be performed on the eluents and the products can be monitored. This has been accomplished in postcolumn reaction detectors coupled in series with analytical columns. A general review of reaction detectors in hplc has recently appeared which discussed the relative merits of capillary, air-segmented flow, and packed bed reactors (97). Theoretical discussions of band broadening and detector design can also be found for packed bed (98- 100) and flow-segmented (101) detectors. In general, capillary tubes are satisfactory for very fast reactions where band broadening cannot become sigificant. Reactions that require longer incubation times should be done in the packed bed or flow-segmented reactors which better maintain the integrity of the analytical column’s resolution.

GOODING

A. Peptide and Protein

Detectors

Udenfriend et al. found that fluorescamine yields a fluorophor when reacted with amino acids, peptides, and other primary amines (102). Peptides and proteins eluted from a column can be mixed with a fluorescamine solution and the fluorescent derivatives detected ( 103 - 105). Optimization of the mixing chamber (105) and reaction conditions (104) have been described by Frei. The apparent shortcoming in the use of fluorescamine is its spontaneous hydrolysis in water. Roth found that o-phthalaldehyde which is stable in water is also a useful reagent for peptide detection (106). It has been used to detect peptides from the mapping of proteins such as bovine serum albumin (107) and cytochrome c (108). Sensitivity is excellent since 300 pg of digested albumin gave good maps (107). B. Enzyme Detectors The catalytic nature of enzymes makes them well suited for reaction detectors. By detecting the products of the enzyme reaction after an incubation period, the signal from the enzyme is highly amplified. Toren et al. were able to detect lactate dehydrogenase isoenzymes by adding lactate and NAD to the effluent from a DEAE-glycophase column and monitoring the NADH formed after passage through capillary incubation columns (45,109,llO). One of the problems with nonkinetic enzyme assays is that it is difficult to detect background absorbance or fluorescence. The background problem has been attacked in the capillary postcolumn reactor systems either by splitting the stream of enzyme incubation mixture and incubating at two different temperatures or by taking absorbance measurements at two different points in the reactor flow stream and subtracting (45,111). An example of computer-aided background subtraction with the latter system produced the remark-

HIGH-PERFORMANCE

LIQUID

CHROMATOGRAPHY

OF PROTEINS

ICO(A)

-I m 3

o0

+ IO

I 30

20 TIME

*

(min)

0

IO TIME

20 (min)

FIG. 15. Fractionation of a commercial calf intestinal alkaline phoshatase sample with and without the enzyme detector. Operational parameters are as follow: Sample: (A) 8 mg/ml calf intestinal alkaline phosphatase, (B) 0.4 mg/ml calf intestinal alkaline phosphatase. Buffer: initial buffer, 0.05 M Tris (pH 8); final buffer, 0.05 M Tris (pH 8), 0.3 M NaCl. Analytical column: 0.05 X 50 cm, 37-74 pm particle size, 250 A pore diameter, DEAE-glycophase/CPG. Postcolumn: 0.5 x 60 cm, 40 pm particle size, glycophase G/nonporous sodium silicate. Substrate: 8 mMp-nitrophenylphosphate in initial buffer. Flow rate: (A) 1.37 ml/min (analytical pump), (B) 0.35 mUmin (substrate pump). Detector: (A) 280 nm, (B) 410 nm. Temperature: room temperature. Reprinted from S. H. Chang, K. M. Gooding, and F. E. Regnier (1976) J. Chromatogr. 125, 103- 114 by courtesy of Elsevier Scientific Publishing Co.

laboratory for quanitating enzyme activity. It is possible that sometime soon someone will couple this to a column to monitor resolved isoenzymes. IX. FUTURE TRENDS

It may be concluded from the data presented above that high-performance liquid chromatography is an emerging new technology that is of value in protein fractionation. Separation times are usually in the range of 10 to 60 min and resolution is equivalent or superior to that of the classical modes of chromatography. Elution protocol on hplc columns is usually comparable to that of the classical column techniques so that new mobile phases do not always have to be selected to carry out separations. The growth of protein fractionation by hplc may be expected to be very rapid in the next 5 years. Several factors that will

influence the growth of this technique are the cost of preparative columns, the existence of alternative analytical methodology and the ease of automation. For example, the current cost of analytical hplc columns ($300-$400) is expensive but tolerable. However, most protein fractionations carried out today are with columns of 1 to 4 cm in diameter where the separation is preparative in nature. The cost of columns of this diameter packed with lo-pm high-performance support would range from $2000 to $4000 apiece. It is questionable whether many laboratories can justify expenditures of this magnitude. For this reason, we believe that a new generation of intermediate or medium-performance liquid chromatography (mplc) supports must be developed.2 These materials will be of larger 2 The term medium-performance raphy (mplc) is a trademark.

liquid chromatog-

22

REGNIER

AND

particle size (37-74 pm) and easier to pack (118). Although separation times will be in the range of 30 min to 3 hr, their substantially lower cost and ease of packing will make them more cost effective. A second area of major growth will be in those areas where there are no competitive techniques. The use of reversed phase liquid chromatography for the purification of peptides is such a case. The speed and resolution of hplc will revolutionize peptide separation. Although they have not currently been identified, it is probable that several other types of protein fractionation will be found that may only be accomplished with hplc. The third and most obvious area of growth will be in the strictly analytical application of the technique. In addition to the advantages of speed and resolution, hplc can easily be automated. These are attractive attributes in quality control, federal monitoring, pharmaceutical, and clinical laboratories. We would predict that in the next decade, 70-80% of the liquid chromatographic separations in biochemical laboratories will be achieved with medium or high-performance systems. X. APPENDIX Resolution In chromatographic systems, resolution of components is achieved by partitioning between two physically distinct phases that share a common interfacial boundary. Separation of solutes is achieved by moving one of these phases, the mobile phase (P,), relative to the stationary phase (P,). The distribution of solute between these two phases is a constant known as the partition or distribution coefjkient (K,). This constant is expressed as KD =

weight of solute in PJvolume weight of solute in P,/volume

of P, of P,

GOODING

The number of moles of solute in the stationary phase (n,) divided by the number of moles in the mobile phase (n,) is equal to a constant called the capacity factor (k ’ ). The following equalities may be deduced through Eq. [l], k’=n,=C,l=KD+, PI nm Gnvm m where V, and V,,, are the volumes of P, and P,, respectively. It is seen that capacity factor (k’) is a convenient means of relating migration velocity or retention to basic properties of the system (K,, V,, and V,). When a solute does not partition with the stationary phase, its KD is zero. Thus, the capacity factor (k ‘) for nonretained peaks is zero. It is apparent that resolution of two components requires that KD, not equal The relative difference in these K,. constants is termed the separation factor (a) and is expressed as [31

The separation factor specifies the relative difference between the peak maxima of the two components but not the degree of resolution. To assess resolution requires that one also consider the peak width, or efficiency of the column. The gross resolving power of chromatographic systems is usually expressed in terms of theoretical plates (N) where the equation 2

N=16& t

i

141

relates solute retention time (t) and peak width (At) to N. Column efficiency is then expressed in height equivalent to a theoretical plate (H): H = LIN,

[51

where L is column length. A relationship between the resolution (R,) of two com-

HIGH-PERFORMANCE

LIQUID

CHROMATOGRAPHY

pounds, their partition coefficients, retention properties and theoretical plates in a column is given by the equation

R, = f(+)(A)

.

[6]

It is a further goal that a given resolution be achieved as quickly as possible. After the mobile phase, stationary phase and column temperature have been selected, cr and k’ for components are fixed. The only variable in Eq. [6] that is related to analysis time is N. An analysis of those factors that control N explains the difference between high-performance liquid chromatographic systems and the classical gravity-fed columns. As mobile phase velocity increases, N decreases and its companion term, H, increases. Decreasing analysis times decreases the efficiency of the separation system as the spreading of chromatographic bands worsens. Band spreading is caused by factors in the column which promote eddy diffusion, longitudinal diffusion and poor mass transfer (115,116): [71

The A or eddy diffusion term is caused by flow through a packed bed. As such, it is an indication of how well a column is packed and is independent of the linear velocity (v) of the mobile phase. A is proportional to the particle diameter (d,) and is therefore minimized by good packing of microparticulate supports. The B or longitudinal diffusion term is proportional to the diffusion coefficients (D) of the solute in the mobile and stationary phases. In liquid chromatography, where diffusion coefficients are very small, B approximates zero. The C or mass transfer term is dependent on both the particle size and the diffusion coefficients: C = c,d;lD,

+ c,d:lD,

+ ca.

PI

In this equation, d, is the depth of the stationary phase, and cl, c2, and c3 are the mass transfer coefficients in the mobile

23

phase, stationary phase, and stagnant pools, respectively. REFERENCES Ackers, G. K. (1970) Advan. Protein Chem. 24, 343-447. Scott, R. P. W., and Kucera, P. (1976) J. Chromatogr.

125, 251-263.

Yau, W. W., Kirkland, J. J., Bly, D. D., and Stoklosa, H. J. (1976) J. Chromatogr. 125, 219-230.

4. Kirkland, J. J. (1976) J. Chromatogr. 125, 231-250. 5. Giddings, J. C., Bowman, L. M., and Myers, M. N. (1977) Macromolecules 10, 443-449. 6. Unger,K.K.,Kern,R.,Ninou,M.C.,andKrebs, K.-F. (1974) J. Chromatogr. 99, 435-443. 7. Haller, W. (1971) U. S. Patent 3,549,524. 8. Kirkland, J. J. (1974) U. S. Patent 3,782,075. 9. Iler, R. K., and McQuestion, H. J. (1974) U. S. Patent 3,855,172. 10. Kirkland, J. J., and Antle, P. E. (1977) J. Chromntogr.

Sci.

15, 137- 147.

11. Unger, K., Schick-Kalb, J., and Krebs, K.-F. (1973) J. Chromatogr. 83, S-9. 12. Mizutani, T., and Mizutani, A. (1975) J. Chromatogr.

111, 214-216.

13. Stenlund, B. (1976) Advnn. Chromatogr. 14, 37-74. 14. Regnier, F. E., and Noel, R. (1976)J. Chromatogr. Sci.

H = A + B/v + Cv.

OF PROTEINS

14, 316-320.

IS. Regnier, F. E. (1976) U. S. Patent 3,983,299. 16. Chang, S. H., Gooding, K. M., and Regnier, F. E. (1976) J. Chromatogr. 120, 321-333. 17. Englehardt, H., and Mathes, D. (1977) J. Chromatogr.

142, 3 11-320.

18. Lu, K. C., Gooding, K. M., and Regnier, F. E. (1979) C/in. Chem. 25, 1608- 1612. 19. Chang, S. H., Gooding, K. M., and Regnier, F. E. (1976) J. Chromatogr. 125, 103- 114. 20. Sebille, B., Thuaud, N., and Tillement, J. -P. (1978) J. Chromatogr. 167, 159- 170. 21. Barford, R. A., Kupec, J., and Fishman, M. L. (1977) J. Water Polk. Control, May 1977, 764-767. 22. Blagrove, R. J., and Frenkel, M. J. (1977) J. Chromatogr.

132, 399-404.

23. Persiani, C., Cukor, P., and French, K. (1976) J. Chromatogr.

24. Crone,

Sci.

14, 417-421.

H. D., and Dawson,

J. Chromatogr.

R. M. (1976)

129, 91-96.

25. Bristow, P. A., and Knox, J. H. (1977) Chromatographia

IO, 279-289.

26. Knox, J. H., and Parcher, J. F. (1969) Anal. Chem.

41, 1599-1606.

27. Noel, R. J. (1977) Ph.D. Thesis, Purdue University, Lafayette, Ind.

24

REGNIER

AND GOODING

28. Gooding, K. M., Lu, K. C., Vanecek, G., and Regnier, F. E. (1979) 4th International Symposium on Column Liquid Chromatography. 29. Regnier, F. E., Gooding, K. M., and Lu, K. C. (1978) 29th Pittsburgh Conference Paper No. 654. 30. Mori, S. (1978) Anal. Chem. 50, 745-748. 31. Fisher, L. J., Thies, R. L., and Charkowski, D. (1978) Anal. Chem. JO, 2143-2144. 32. Klein, J., and Treichel, K. (1977) Chromatographia 10, 604-610.

J. Chromatogr.

160, 301-305.

37. Fukano, K., Komiya, K., Sasaki, H., and Hashimoto, T. (1978) J. Chormatogr. 166, 47-54. 38. Rokushika, S., Ohkawa, T., and Hatano, H. ( 1978) U. S.-Japan Seminar on Advanced Techniques in Liquid Chromatography, Boulder, Colo. 39. Chang, S. H., Noel, R., and Regnier, F. E. (1976) Anal. Chem. 48, 1839-1845. 40. Eltekov, Y. A., Kiselev, A. V., Khokhlova, T. D., and Nikitin, Y. S. (1973) Chromatographia 6, 187- 189. 41. Regnier, F. E., Gooding, K. M., and Chang, S. H. (1977) Cont. Top. Clin. Anal. Chem. 1, l-48. 42. Chang, S. H., and Regnier, F. E. (1977) U. S. Patent 4,029,583. 43. Kudirka, P. J., Schroeder, R. R., Hewitt, T. E., and Toren, E. C. (1976) Clin. Chem. 22, 471-474. 44. Schroeder, R. R., Kudirka, P. J., and Toren, E. C. (1977) J. Chromatogr. 134, 83-90. 45. Fulton, J. A., Schlabach, T. D., Kerl, J. E., and Toren, E. C. (1979)5. Chromatogr. 175,283-291. 46. Bostick, W. D., Dinsmore, S. R., Mrochek, J. E., and Waalkes, T. P. (1978) Clin. Chem. 24, 1305-1316. 47. Alpert, A., and Regnier, F. E. .I. Chromatogr., in press. 48. Gooding, K. M., Lu, K. C., and Regnier, F. E. (1979) .I. Chromatogr. 164, 506-509. 49. Radhakrishnan, A. N., Stein, S., Licht, A., Gruber, K. A., and Udenfriend, S. (1977) J. Chromatogr.

132, 552-555.

50. Mikes, O., Strop, P., and Sedlachova, J. (1978) J. Chromatogr.

148, 237-245.

51. Mikes, 0.. Strop, P., Zbrozek, J., and Coupek, J. (1976) 1. Chromatogr. 119, 339-354. 52. Tanford, C. (1978) Science 200, 1012-1018. 53. Howard, G. A., and Martin, A. J. P. (1950) Biochem.

Chromatogr.

J. 46, 534-544.

142, 213-232.

58. Tanaka, N., and Thornton, E. R. (1977) J. Amer. Chem.

33. Rubinstein, M. (1979) Anal. Biochem. 98, l-7. 34. Mikes, 0. (1975) in Liquid Column Chromatoggraphy, (Deyl, Z., Macek, K., and Janak, J., eds.), Elsevier, Amsterdam. 35. Nakamura, S., Kaiho, I., and Ishiguro, S. (1978) U. S.-Japan Seminar on Advanced Techniques in Liquid Chromatography, Boulder, Colo. 36. Hashimoto, T., Sasaki, H., Aiura, M., and Kato, Y. (1978)

54. Hofstee, B. H. J. (1976) J. Macromol. Sci. AlO, 111-117. 55. Hofstee, B. H. J. (1975) Biochem. Biophys. Res. Commun. 63, 618-624. 56. Nagasawa, T., Fujinori, H. N., and Heinrich, P. C. (1979) Eur. J. Biochem. 94, 31-39. 57. Scott, R. P. W., and Kucera, P. (1977) J.

Sot.

99,730O.

59. Horvath, C., Melander, W., and Molnar, I. (1976) 1. Chromatogr. 125, 129- 156. 60. Tanford, C. (1973) The Hydrophobic Effect, Wiley-Interscience, New York. 61. Karger, B. L., and Giese, R. W. (1978) Anal. Chem. 50, 1048A-1073A. 62. Ryeszotarski, W. J., and Mauger, A. B. (1973)J. Chromatogr.

86, 246-249.

63. Tsuji, K., Robertson, J. H., and Bach, J. A. (1974) J. Chromatogr. 99, 597-608. 64. Tsuji, K., and Robertson, J. H. (1975) J. Chromatogr.

112, 663-672.

65. Gruber, K., Stein, S., Brink, L., Radhakrishnan, A., and Udenfriend, S. (1976) Proc. Nat. Acad. Sci. USA 73, 1314-1318. 66. Hancock, W. S., Bishop, C. A., and Heam, M. T. W. (1976) FEBS Lett. 72, 139-142. 67. Burgus, R., and Rivier, J. (1976)in Peptides 1976: Proceedings of the 14th European Symposium on Peptides, pp. 85-94. 68. Molnar, I., and Horvath, C. (1977)J. Chromatogr. 142, 623-640.

69. Krummen,

K.,

Chromatogr.

70. Krummen, Chromatogr.

and Frei, R. W. (1977) J. 132, 27-36.

K.,

and Frei, R. W. (1977) J. 132, 429-436.

71. Hansen, J. J., Greibrokk, T., Currie, B. L., Johansson, K. N.-G., and Folkers, K. (1977) J. Chromatogr.

135, 155-164.

72. Month, W., and Dehnen, W. (1977)J. Chromatogr. 140, 260-262.

73. Udenfriend, S., and Stein, S. (1977) in Peptides: Proceedings of the 5th Annual Symposium, pp. 14-26, Wiley, New York. 74. Hancock, W. S., Bishop, C. A., Meyer, L. J., Harding, D. R. K., and Heam, M. T. W. (1978) J. Chromatogr. 161, 291-298. 75. Rivier, J., Spiess, J., Perrin, M., and Vale, W. (1979) in Biological and Biomedical Applications of Liquid Chromatography, (Hawk, G., ed.), Dekker, New York. 76. Hancock, W. S., Bishop, C. A., Prestidge, L., Harding, D. R. K., and Hearn, M. T. W. (1978) J. Chromatogr. 153, 391-398. 77. Hancock, W. S., Bishop, C. A., Prestidge, R. L., Harding, D. R. K., and Hearn, M. T. W. (1978) Science 200, 1168-1170. 78. Hancock, W. S., Bishop, C. A., Meyer, L. J.,

HIGH-PERFORMANCE

LIQUID CHROMATOGRAPHY

Harding, D. R. K., and Hearn, M. T. W. (1987) J. Chromatogr.

161, 291-298. (1978)5. Chromatogr.

79. Glasel, J. A. 145,469-472. 80. Larson, B., Viswanatha, V., Chang, S. Y., and Hruby, V. J. (1978) J. Chromatogr. Sci. 16, 207-210. 81. Month, W., and Dehnen, W. (1978) J. Chromatogr.

147, 415-418.

82. Hancock, W. S., Bishop, C. A., Battersby, J. E., Harding, D. R. K., and Heam, M. T. W. (1979) J. Chromatogr. 168, 377-384. 83. O’Hare, M. J., and Nice, E. C. (1979) J. Chromatogr. 171, 209-226. 84. Roumeliotis, P., and Unger, K. K. (1978) J. Chromatogr. 149, 21 l-224. 85. White, A., Handler, P., Smith, E. L. (1959) Principles of Biochemistry, McGraw-Hill, New York. 86. Schill, G., Modin, R., and Persson, B. A. (1977) in Handbook of Derivatives for Chromatography, Chap. 14, (Blau, K., and King, G. S., eds.), Heyden, London. 87. Krommen, J., Fransson, B., and Shill, G. (1977) J. Chromatogr. 142, 283-297. 88. Knox, J. H., and Jurand, J. (1976)5. Chromatogr. 125, 89- 101. 89. Schill, G., Modin, R., Borg, K. O., and Persson, B. A. (1977) in Handbook of Derivatives for Chromatography (Blau, K., and King, G. S. eds.), Chap. 14, Heyden, London. 90. Rivier, J. E. (1978) J. Liq. Chromatogr. 1, 343-366. 91. LePage, J. N., Lindner, W., Davies, G., Seitz, D. E., and Karger, B. L. (1979) Anal. Chem. 51, 433-435.

92. Cooke, N. H. C., Viavattene, R. L., Wong, W. S., Davies, G., and Karger, B. L. (1978) J. Chromatogr. 149, 391-415. 93. Ohlson, S., Hansson, L., Larsson, P. 0.. and Mosbach, K. (1978) FEBS Left. 93, 5-9. 94. Mizutani, T., and Mizutani, A. (1979) 1. Chromatogr. 168, 143- 150. 95. Kikta, E. J., and Grushka, E. (1977) J. Chromatogr.

135, 367-376.

96. Long, G. W. K., and Grushka, Chromatogr.

E. (1977) J.

142, 299-309.

97. Frei, R. W., and Scholten, A. H. M. T. (1979) J. Chromatogr.

Sci.

17, 152-160.

OF PROTEINS

25

98. Deelder, R. S, Kroll, M. G. F., and Van Den Berg, J. H. M. (1976) J. Chromatogr. 125, 307-314. 99. Deelder, R. S., Kroll, M. G. F., Beeren, A. J. B., and Van Den Berg, J. H. M. (1978) J. Chromatogr.

149, 669-682.

100 Huber, J. F. K., Jonker, K. M., and Poppe, H. (1980) Anal. Chem. 52, 2-9. 101. Snyder, L. R. (1976) J. Chromatogr. 125, 287- 306. 102. Udenfriend, S., Stein, S., Bohlen, P., Dairman, W., Leimgruber, W., and Weigele, M. (1972) Science 178, 871-872. 103. Bohlen, P., Stein, S., Stone, J., and Udenfriend, S. (1975) Anal. Biochem. 67, 438-445. 104. Frei, R. W., Michel, L., and Santi, W. (1976) J. Chromatogr.

126, 665-677.

105. Frei, R. W., Michel, L., and Santi, W. (1977) J. Chromatogr.

142, 261-270.

106. Roth, M. (1971) Anal. Chem. 43, 880-882. 107. Benson, J. R. (1976)Anal. Eiochem. 71,459-470. 108. Creaser, E. H., and Hughes, G. J. (1977) Chromatogr.

.I.

144, 69-75.

109. Schroeder, R. R., Kudirka, P. J., andToren, E. C. (1977) J. Chromatogr. 134, 83-90. 110. Schlaback, T. D., Fulton, J. A., Mockridge, P. B., and Toren, E. C. (1979) Clin. Chem. 25, 16001607. 111. Fulton, J. A., Schlabach, T. D., Kerl, J. E., Toren, E. C., and Miller, A. R. (1979) J. Chromatogr. 175, 269-281.

112. Schlabach, T. D., Chang, S. H., Gooding, K. M., and Regnier, F. E. (1978) J. Chromatogr. 134, 349-364.

113. Schlabach, T. D., Alpert, A. J., and Regnier, F. E. (1978) C/in. Chem. 24, 1351- 1360. 114. Schlabach, T. D., and Regnier, F. E. (1978) J. Chromatogr.

158, 349-364.

115. Schifreen, R. S., Hanna, D. A., Bowers, L. D., and Carr, P. W. (1977) Anal. Chem. 49, 1929- 1939. 116. Giddings, J. C. (1965) Dynamics of Chromatography Part I, New York. 117. Grushka, E., Snyder, L. R., and Knox, J. H. (1975) J. Chromatogr. Sci. 13, 25-37. 118. Snyder, L. R., and Kirkland, J. J. (1974) Introduction to Modern Liquid Chromatography, p. 189, Wiley-lnterscience, New York.