Pro. Natl. Aad. Si. USA Vol. 77, No. 3, pp. 1632-1636, Marh 1980 Medial Sienes Predition of peptide retention times in high-pressure liquid hromatography on the basis of amino aid omposition (lipophiliity/separation tehniques) JAMES L. MEEK Laboratory of Prelinial Pharmaology, National Institute of Mental Health, Saint Elizabeths Hospital, Washington, D.. 20032 ommuniated by Brue Merrifield, Deember 17,1979 ABSTRAT Analysis of peptides by reverse-phase highpressure liquid hromatography would be simplified if retention times ould be predited by summing the ontribution to retention of eah of the peptide's amino aid side hains. This paper desribes the derivation of values ("retention oeffiients") that represent the ontribution to retention of eah of the ommon amino aids and end groups. Peptide retention times were determined on a Bio-Rad "ODS" olumn at room temperature with a linear gradient from 0.1 M NaIO4, ph 7.4 or 2.1, at 0 min to 60% aetonitrile/0.1 M NaIO4 at 80 min. The NalO4, a haotropi agent, was added to improve peak shape and to minimize onformational effets. Retention oeffiients for the amino aids were omputed by using a Hewlett-Pakard 9815A alulator programmed to hange the retention oeffiients for all amino aids sequentially to obtain a maximum orrelation between atual and predited retention times. orrelations of 0.999 at ph 7.4 and 0.997 at ph 2.1 were obtained for 25 peptides inluding gluagon, oxytoin, [Metlenkephalin, neurotensin, and somatostatin. This high degree of orrelation suggests that, for peptides ontaining up to 20 residues, retention is primarily due to partition proesses that involve all the residues. Although steri or onformational fators do have some effet on retention, the data suggest that under the above hromatographi onditions the retention of peptides ontaining up to 20 residues an be predited solely on the basis of their amino aid omposition. This possibility was tested by using data taken from the literature. The possibilities for separating and isolating small peptides have been markedly improved by the introdution of reverse-phase high-pressure liquid hromatography (HPL) (1-5). This tehnique depends upon the hydrophobi interations between a hydroarbonaeous olumn and the peptides to be separated: the more hydrophobi (lipophili) the ompound, the stronger its retention by the olumn. To elute strongly retained ompounds, aqueous solutions (the "mobile phase") ontaining a large amount of organi solvent must be pumped through the olumn. hoie of the optimum mobile phase and hromatographi onditions for given peptides an be found by trial and error after qualitatively examining the balane of hydrophobi and hydrophili amino aids present in the peptide. However, as noted by Molna'r and Horvath (1), it should be possible to obtain quantitative estimates of the hydrophobiity of the amino aids ontained in a peptide, whih will reflet their retention on the reverse-phase olumn. Estimates of hydrophobiity based on otanol/water partition oeffiients exist for many but not all amino aids. By using suh values, O'Hare and Nie (2) noted that the retention order for small peptides was generally orrelated with the sum of the values for the most hydrophobi residues of the peptides. The many deviations they found between the observed order of elution and the lipophiliity estimates presumably derive from the fat that retention The publiation osts of this artile were defrayed in part by page harge payment. This artile must therefore be hereby marked "advertisement" in aordane with 18 U. S.. 1734 solely to indiate this fat. on otadeylsilyl silia gel is a quite different proess from otanol/water partition and from the unavailability of hydrophobiity data for terminal groups of the peptides. It is the purpose of this paper to show that "retention oeffiients" an be derived diretly from HPL data for all amino aids and end groups suh that the retention time of a peptide an be predited from the sum of the retention oeffiients for eah amino aid and end group. MATERIALS AND METHODS Peptides were obtained from Sigma and Peninsula Laboratories (San arlos, A). HPL grade aetonitrile was obtained from Fisher. The HPL system was assembled from modular omponents: a one-hamber glass gradient maker, a pump (Milton Roy, Riviera Beah, FL), and a sample valve (Rheodyne, Berkeley, A). The olumn effluent was passed in series through a variable wavelength photometer (Altex, Berkeley, A) and a filter fluorometer (Farrand, Valhalla, NY). The mobile phase gradient normally used was from 0.1 M NaIO4/0% aetonitrile at 0 min after injetion to 0.1 M Nal04/60% aetonitrile (vol/vol) at 80 min. For separations at ph 7.4, the starting buffer ontained 5 mm phosphate buffer, ph 7.4. For operation at ph 2.1, both starting and final buffers ontained 0.1% phosphori aid (3). Linear gradients from 100% A to 100% B are generated if a solution A is plaed in a mixing hamber and then B is added at 1/2 the rate at whih the mixture is withdrawn (6). The gradient maker onsisted of a 3 X 13 m glass ylinder with two Teflon three-way stopoks and a magneti stirrer. To generate the gradient, the hamber was filled with 40 ml of starting buffer. Final buffer was then added to the hamber at 0.5 ml/min, while the mixture was pumped into the olumn at 1.0 ml/min. Peptides were deteted by absorbane'at 200 or 220 nm or by fluoresene [after the primary amino groups of the peptides had reated with fluoresamine (7)]. With a mobile phase of ph 7.4, the olumn effluent ould reat diretly with fluoresamine, beause the ph optimum for many peptides is near neutral (8). The fluoresamine (10 mg/100 ml of aetonitrile) was added to the effluent at 0.1 ml/min. ontinuous neutralization of the ph 2.1 mobile phase was ahieved by adding the organi base imidazole (1.0 M final onentration) to the fluoresamine/aetonitrile mixture. The retention oeffiients were omputed by repetitive regression analysis: values for eah amino aid were suessively hanged by 0.2 min until maximum orrelation between atual and predited retention times was obtained. Starting values for the retention oeffiients of the neutral and hydrophili amino aids were initially assumed to be zero. Starting values for the lipophili amino aids were obtained by plotting the retention Abbreviation: HPL, high-pressure liquid hromatography. 1632
Medial Sienes: Meek times of oligomers (e.g., diphenylalanine, triphenylalanine, tetraphenylalanine, et.) vs. the number of residues; the slope of the plot equals the retention per residue. To ompute the retention oeffiients, a Hewlett-Pakard 9815A alulator was programmed to store these starting retention oeffiients for the 26 amino aids and end groups, to store the atual retention times for the 25 peptides studied, and then to alulate predited retention times for these peptides by summing the retention oeffiients for eah amino aid ontained. After alulating the orrelation oeffiient between predited and atual retention times, 0.2 min was added to the retention oeffiient for an amino aid; the predited retention times and orrelation oeffiient were then again alulated. If the orrelation had been improved by adding 0.2 min to the retention oeffiient, the hange was kept; otherwise the value was returned to that previously used, and 0.2 min was added to the next amino aid and so on. After heking all amino aids to see whether inreasing the retention oeffiient ould inrease the orrelation, 0.2 min was sequentially subtrated from eah amino aid in turn and orrelations were again alulated after eah subtration. At the end of these two yles, the slope of the plot of predited vs. atual retention times was alulated beause the predited and atual times should be equal, not merely be orrelated (proportional). Therefore, if the slope was greater than 1.0, all retention oeffiients were multiplied by 0.99; if the slope was less than 1.0, the values were multiplied by 1.01. These yles were repeated until a near maximum orrelation had been obtained. RESULTS Table 1 lists the 25 peptides used for alulation of retention oeffiients and their retention times at ph 7.4 and 2.1. As expeted, small neutral peptides (triglyine and pentaalanine) Table 1. Retention times of peptides separated by HPL Retention time, min ompound ph 7.4 ph 2.1 1. Triglyine 2.0 3.0 2. Pentaalanine 4.6 8.1 3. Divaline 6.9 14.5 4. Dimethionine 10.5 21.0 5. TRF 11.5 11.2 6. Tuftsin 11.7 12.0 7. Trityrosine 19.5 29.7 8. [Met]Enkephalin 27.5 38.0 9. Trileuine 28.0 36.8 10. [Leu]Enkephalin 29.3 42.0 11. Ditryptophan 31.6 44.3 12. Angiotensin II 32.2 47.5 13. a-endorphin 32.3 47.7 14. aerulein 34.2 42.2 15. Oxytoin 36.4 37.9 16. Gastrin-(12-15) 36.5 42.4 17. Neurotensin 39.0 48.0.18. Physalaemin 41.0 43.0 19. Triphenylalanine 41.6 49.5 20. LHRF 42.8 49.5 21. a-melanotropin 46.2 46.2 22. Bradykinin 48.0 45.0 23. Eledoisin-related peptide 53.0 44.0 24. Gluagon 53.6 60.0 25. Somatostatin 57.5 55.0 Peptides were hromatographed at room temperature on a Bio-Rad ODS olumn with a linear aetonitrile gradient (0.75%0/min). TRF, thyrotropin-releasing fator; LHRF, luteinizing hormone-releasing fator; Eledosin-related peptide, Lys-Phe-Ile-Gly-Leu-Met-NH2. Pro. Natl. Aad. Si. USA 77 (198 1633 were only slightly retained. Small peptides with lipophili side hains (e.g., triphenylalanine) and most of the larger biologially ative peptides required muh higher onentration of aetonitrile for elution. Aidifiation of the mobile phase inreased the retention of peptides with free terminal arboxyl groups ([Metlenkephalin and angiotensin II) or with aidi residues [gastrin-(12-15)]. Most peptides with masked arboxyl or arboxyl and amino groups (thyrotropin-releasing fator, oxytoin) had similar retentions at both phs. Peptides ontaining the basi residues lysine or arginine (luteinizing hormone-releasing fator, eledoisin-related peptide) exhibited a dereased retention at lower ph due either to inreased ionization of the amino group or to formation of an ion pair with the perhlorate in the mobile phase. Fig. 1 shows the separation at ph 2.1 of various peptides. The sharpness of the peaks with minimal tailing demonstrates the high resolution possible with HPL. Fig. 1 also demonstrates the differene in detetor seletivity between measurement of absorbane at 220 nm (top trae) and measurement of fluoresamine-indued fluoresene (bottom trae). ompounds suh as thyrotropin-releasing fator (peak 5) without a free amino-terminal group give little or no fluoresene. Fig. 2 shows that there is an approximately linear inrease in retention time for phenylalanine oligomers as phenylalanine residues are added to diphenylalanine. This finding indiates that, with the linear gradient used in these experiments, the addition of eah phenylalanine residue adds approximately the same retention time to the peptide. The slope thus equals the retention added per side hain and peptide bond, but does not inlude the ontribution of the terminal amino or arboxyl groups. Extrapolation of the line to 0 residues gives a positive value, whih represents the ontribution to retention of the end groups. The retention for phenylalanine itself 0.05' D o.0 0-0 Q) : IL).1) LL F I1 2 b~~~~. ~~~ _V 0 10 20 30 40 50 60 Retention time, min FIG. 1. Separation of peptides by reverse-phase HPL on a Lihrosorb RP18 olumn. (Upper) Absorbane at 220 nm (0.05 absorbane units full sale). (Lower) Fluoresamine-indued fluoresene. A 40-gl sample ontaining 300-600 ng of eah peptide was hromatographed at ph 2.1 at room temperature with a flow rate of LO ml/min and an aetonitrile gradient of 0.75%/min. 7 I 10 25 24
1634 Medial Sienes: Meek E 0w a) a1) 1 2 3 4 5 Number of phenylalanine residues FIG. 2. Linearity of retention time with the number of phenylalanine residues. Phenylalanine oligomers were hromatographed by using a linear aetonitrile gradient. The slope of this plot indiates the HPL retention per phenylalanine residue. was not plotted beause the pks for amino aids differ onsiderably from those of peptides and the extent of ionization markedly affets retention. Table 2 lists the retention oeffiients alulated by repeated regression analysis. Amino aids with aromati or aliphati side hains have a marked positive ontribution to retention whih hanges relatively little with ph. Residues with aidi side hains have a marked negative ontribution to retention whih inreases in magnitude as ionization inreases. Basi and neutral residues have little effet on retention. Fig. 3 shows the remarkably high orrelation obtained when plotting the atual retention times from Table 1 vs. the times predited by summing the retention oeffiients listed in Table 2 for eah peptide. It should also be possible to predit retention times when using various other hromatographi onditions. The retention times for several di- and tripeptides (numbers 1, 2, 3, 9, and 19 of Table 1) were determined with several gradient rates and with olumns of several manufaturers. With a gradient rate of 1.5% aetonitrile per min (twie the usual rate), all retention times were 70% of normal. With a gradient rate of 0.5%/min, retention times were inreased to 120% of normal. Retention times for these ompounds obtained with the Bio-Rad olumn were similar to those seen with olumns from Waters Assoiates (10-,m partile size), Lihrosorb (Rheodyne) (5,mn), and Dupont (5,um). That there are minor differenes in retention with different olumns an be noted by omparing Table 1 (Bio-Rad olumn) with Fig. 1 (Lihrosorb olumn). Thyrotropin-releasing fator (peak 5) was shifted about 2 min relative to pentaalanine and divaline; luteinizing hormone-releasing fator was shifted about 2 min relative to [Met]- and [Leu]- enkephalin. TEST OF PREDITIVE ABILITY OF RETENTION OEFFIIENTS To examine how well these retention oeffiients are able to predit times for other peptides and other hromatographi onditions, data were taken from the literature for omparison Pro. Natl. Aad. Si. USA 77 (198 Table 2. Retention oeffiients of amino aid residues Retention oeffiients Amino aid (N) ph7.4 ph 2.1 Tryptophan Phenylalanine Isoleuine *Leuine Tyrosine Methionine Valine Proline Threonine Arginine Alanine Glyine Histidine ystine Lysine Serine Asparagine Glutamine Asparti aid Glutami aid Amino- -OOH -Amide Pyroglutamyl- Aetyl- Tyrosine sulfate (7) (13) (9) (11) (9) (1 (7) (13) (2) (8) (6) (3) (19) (17) (8) (1) (1) 14.9 13.2 13.9 8.8 6.1 4.8 2.7 6.1 2.7 0.5 0.0-3.5-6.8 0.1 L2-4.8-8.2-16.9 2.4-3.0 7.8-1.1 5.6 10.9 18.1 13.9 11.8 10.0 8.2 7.1 3.3 8.0 1.5-4.5-0.1-0.5-2.2-3.2-3.7-1.6-2.5-2.8-7.5-0.4 6.9 5.0-2.8 3.9 6.5 N, number of peptides used for alulation of retention oeffiients that ontained this amino aid. Retention oeffiients (in min) were determined by reverse-phase HPL on a Bio-Rad ODS olumn. The predited retention time for a peptide equals the sum of the retention oeffiients for the amino aids and end groups plus to (the time for elution of unretained ompounds). In these experiments, to = 2.0 min. with predited times. The best available ompilation of retention times for peptides of biologial interest was made by O'Hare and Nie (2). These authors used an aetonitrile gradient at ph 2.1 (as in the present study), although the inorgani ion added (0.1 M phosphate), Hypersil ODS olumn, and triphasi linear gradient differed from the onditions used here. They hose a gradient onsisting of a rapid rate of hange for about the first 6 min, then a 40-min slower rate of hange, and a final 5-min rapid rate. The retention harateristis of ompounds eluting during the seond phase of their hromatographi system should be omparable to those used in the present study, although retention of ompounds eluting between 10 and 15 min might be overestimated due to the influene of the initial rapid gradient. To examine the omparability of the two hromatographi systems, the atual retention times of phenylalanine oligomers was plotted (Fig. 4). Least squares analysis gave a line with a slope of 36 (representing the differene in gradient rates) and a y interept of -12.3 min [due to the rapid initial gradient used by O'Hare and Nie (2)]. The high orrelation for these three points (0.9983) indiates that, for these ompounds at least, the hromatographi onditions an be ompared. Table 3 lists the atual retention times of ompounds tested by O'Hare and Nie of 31 amino aid residues or less, exluding insulin hains A and B. These latter ompounds ontain ysteine for whih no retention oeffiient had been alulated beause ysteine was not ontained in any of the peptides tested in Table 1. Predited retention times were alulated for the ompounds in Table 3 by summing the retention oeffiients, multiplying
Medial Sienes: Meek Pro. Natl. Aad. Si. USA 77 (198 1635 610.E_ w 30 4-540 2 25 ) 1 7 1 6 14 12 1 21 7+ ".2 2 10 4.56 3~~ 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Atual retention time, min FIG. 3. orrelation of atual retention times vs. times predited by summing retention oeffiients for the amino aids and end groups. Numbers adjaent to the data points indiate the peptides listed in Table 1. (Left) ph 7.4; orrelation = 0.9996. (Right) ph 2.1; orrelation = 0.9970. by the slope from Fig. 4 (36), and adding the y interept (-12.3 min). As seen in Table 3, the predited and atual retention times agree reasonably well, espeially onsidering the differenes in olumns, mobile phases, and gradients. For all peptides up to 20 residues, the average error was 4 min. DISUSSION The basi premise of this study was that it should be possible to derive values that reflet the hydrophobiity (positive or negative) of the amino aids in peptides and that it is the sum of these values that might primarily determine the extent of retention on reverse-phase HPL olumns. Amino aid omposition annot be the only fator determining the extent of retention beause it is possible to separate position isomers with the same omposition (e.g., Gly-Trp and Trp-Gly) (1) and stereoisomers. However, it seemed likely that for small linear peptides, sequene and onformation should have relatively little effet on retention. As shown by these data, this premise appears justified for peptides up to about 20 residues long. There is no evidene that the size of these peptides has any effet on retention. The high exlusion limit of these olumns I E z 0 _. az Phe 2, 20 30 40 50 Retention time in NaIO4, min FIG. 4. omparison of retention times for phenylalanine oligomers in two hromatographi systems. Data of O'Hare and Nie (2) were obtained with a triphasi aetonitrile gradient ontaining NaH2PO4 on a Hypersil ODS olumn. Data in this paper were obtained with a linear aetonitrile gradient ontaining NalO4 on a Bio-Rad ODS olumn. 60 (30,000 for 100-A pore size partiles) makes it unlikely that steri exlusion has a signifiant role in separation of these small peptides. Quite large peptides and even small enzymes an be hromatographed with high resolution by HPL (2), although it is likely that these ompounds are in a partly folded onfiguration. In theory, even higher resolution might be obtained if onditions ould be found in whih all the residues in the moleule were available for interation with the surfae of the stationary phase. Table 3. omparison of data from the literature with predited retention times No. of Retention time, min Error, ompound residues Predited Atual* min [Met]Enkephalin 5 18.4 19.0-0.6 [Leu]Enkephalin 5 20.9 22.0-1.1 ATH-(5-1 6 11.9 17.0-5.1 ATH-(34-39) 6 27 31.0-4.0 ATH-(4-1 7 12.4 20.5-8.1 ATH-(4-11) 8 33.1 30.0 3.1 Angiotensin II 8 27.4 23 4.4 Substane P-(4-11) 8 33.1 30.0 3.1 Oxytoin 9 27.4 23.0 4.4 [Arg]Vasopressin 9 9.0 14.0-5.0 [Lys]Vasopressin 9 10.2 13.0-2.8 [Arg]Vasotoin 9 7.2 12.0-4.8 Substane P 11 33.3 29.0-4.3 a-msh 13 27.3 26.0 1.3 Neurotensin 13 28.9 24.5 4.4 Somatostatin 14 25.5 32.0-6.5 Bombesin 14 22.6 26.0-3.4 Gastrin-1 17 26.8 28.5-1.7 ATH-(18-39) 22 20.2 30.5-10.3 ATH-(1-24) 24 34.9 21.5 13.4 Melittin 25 61.3 46.0 15.3 Gluagon 29 39.7 36.0 3.7,B-Endorphin 31 41.5 34.0 7.5 * Data were from O'Hare and Nie (2). Preditions of retention times were made by summing retention oeffiients for eah peptide and adding to (2 min). To orret for the different gradient rates used in these studies, the sum of the oeffiients was multiplied by 36 (the slope in Fig. 4). To orret for the biphasi gradient used by O'Hare and Nie, they interept of Fig. 4 (-12 min) was then added. ATH, ortiotropin; MSH, melanoyte-stimulating hormone.
1636 Medial Sienes: Meek HOIE OF HROMATOGRAPHI ONDITIONS The ideal mobile phase should allow sharp peaks with minimal tailing, have low absorbane at 200 nm to simplify high-sensitivity detetion, be easy to neutralize to failitate reation with fluoresamine, and have only volatile omponents to allow onentration of the effluent for subsequent analysis. Early reports in whih peptides were hromatographed with methanoli mobile phases ontaining no inorgani salts showed broad peaks of little use onsidering the omplexity of real samples. Molnar and Horvath (1) showed that extremely high resolution ould be obtained with aetonitrile gradients ontaining 0.6 M H104 or 0.1 M phosphate, ph 2.1. However, both these solutions are diffiult to neutralize ontinuously. Use of the strong phosphate buffer at neutral ph also gave exellent results, but the low solubility in aetonitrile of sodium phosphate limits its usefulness. Rubinstein et al. obtained exellent results with an isopropanol gradient and volatile pyridine buffers, but these buffers prelude the use of ultraviolet detetors. The present experiments were performed with NalO4 beause it permitted both reation with fluoresamine and detetion at 220 nm. Unfortunately, NalO4 is not volatile. Preliminary experiments showed that many small peptides gave sharp peaks with only the dilute phosphori aid reommended by Hanok et al. (3) or very dilute sodium phosphate buffer, ph 7.4. However, one ompound (substane P) gave very broad peaks unless Nal04 was added. It may be that the perhlorate bloks adsorption to some exposed sites on the silia beads of the olumn. However, a theoretial reason for using NalO4 is that it is a strong "haotropi" agent-i.e., an inorgani ion that favors the transfer of nonpolar groups to water by altering water struture (9). Suh agents inrease the solubility of lipophili polymers and are likely to break down seondary and tertiary strutures of peptides. It may be that high onentrations of NalO4 will aid in hromatography of very large peptides. The hoie of a linear gradient, started at the time of injetion, was ditated by several onsiderations: (i) it provides optimal peak shape in all regions of the hromatogram, (ii) it is reproduible in all laboratories regardless of the equipment, and (iii) it was hoped that linear gradients would permit estimation of retention times from amino aid omposition. Molbrfr and Horvaith (1) reported that in an isorati (nongradient) Pro. Natl. Aad. Si. USA 77 (198 separation a plot of log k' (the adjusted retention time/to) vs. the number of alanine residues in alanine oligomers yielded a straight line.,snyder and Kirkland (1 reported that with linear gradients on hemially bonded phases log k' was linear with vol% of B over the range 20-90% B. It therefore follows that a linear gradient should give a linear inrease in retention time as residues are added to a homologous series. That this is approximately true is shown in Fig. 3. omplete gradients might not be required for routine separations of a few ompounds; a near optimal gradient program ould be readily designed by prediting the retention times, alulating the aetonitrile onentration at that point (0.75%/min), and hoosing a slightly lower onentration of aetonitrile to bring the retention time into the desired range for k' of 2-10 (1. Although HPL has so far not proven useful in the diret hemial estimation of peptides in rude tissue samples, HPL appears to be the most powerful tool urrently available for the preparative separation and isolation of small peptides and seems to be the ideal method to hek the speifiity of measurements made by site-speifi tests suh as radioimmunoassays or bioassays. Hopefully, the ability to readily predit retention times and to simplify hoie of hromatographi onditions will aid suh studies. 1. Molnar, I. & Horvath,. (1977) J. hromnatogr. 142, 623-640. 2. O'Hare, M. J. & Nie, E.. (1979) J. hromatogr. 171, 209-226. 3. Hanok,-W. S., Bishop,. A., Prestidge, R. L., Harding, D. R. K. & Hearn, M. T. W. (1978) Siene 200,1168-1170. 4. Rubinstein, M., Stein, S. & Udenfriend, S. (1977) Pro. Nati. Aad. Si. USA 74,4969-4972. 5. Rivier, J. (1978) J. Liquid hromatogr. 1, 343366. 6. Lakshmanan, T. K. & Lieberman, S. (1954) Arh. Biohem Biophys. 53, 258-281. 7. Udenfriend, S. (1972) Siene 178, 871-872. 8. Frei, R. W., Mihel, L. & Santi, W. (1976) J. hromatogr. 126, 665-677. 9. Hatefi, Y. & Hanstein, W. G. (1969) Pro. Nati. Aad. Si. USA 62, 1129-1136. 10. Snyder, L. R. & Kirkland, J. J. (1974) in Introdution to Modern Liquid hronatography (Wiley, New York), p. 466.