Peptide stability in drug development: A comparison of peptide reactivity in different biological media

AmA August 1992 Volume 81, Number 8

JOURNAL OF PHARMACEUTICAL SCIENCES A puMicatin of

the American Pharmaceutical Association

ARTICLES

Peptide Stability in Drug Development: A Comparison of Peptide Reactivity in Different Biological Media MICHAELF. POWELL'*', HOWARD GREY', FEDERICO GAETA*,ALEXSElTE',

AND

SONIA COL6N' Received Ma 17. 1991, from the 'CyW Corporation, 3525 John Hopkins Court, San Diego, CA 92121. Accepted for publication October 9, l&l. *Present address: Genentech, Inc., 460 R. San Bruno Blvd., South San Francisco, CA 94080. A M 0 Degradation kinetics for several peptides that bind to the major histocumpatability complex on antigen-presenting cells were determined in both human serum (HS; 25%) and synovial fluid (SF;25%) from patients with rheumatoid arthritis to test whether therapeutic intervention of rheumatoid arthritis by direct intrasynovial injection is feasible (at least in terms of peptide stability). Controls consisted of enzymaticallyimmature 10% fetal calf serum and peptidase-rich 5% liver homogenate (all diluted with RPMI-1040 tissue culture medium). Peptide half-lives ranged from -4 to >10 OOO min, with most peptides showing half-lives of -10-100 mln. These studies show that, even though the populations of inflammatory and other cell types in SF and HS are different (and may, therefore, generate different peptidase profiles),the observed peptide stabilities in SF and HS are similar. This finding indicates that the effect of SF on peptide stability is similar to that of HS.

The blockade of the major histocompatibility complex (MHC) on antigen-presenting cells by peptide antagonists constitutes a novel therapeutic treatment for rheumatoid arthritis, type I diabetes, and other autoimmune diseases.1 To this end we have previously defined some requirements for peptide binding to MHC molecules, as well aa addressed selected pharmaceutical aspectaof peptide antagonist design.2 Specifically,such peptides must exhibit sufficient in vivo stability to "load up" the MHC on the surface of antigen-presenting cells before degradation by serum peptidasea (aswell aa hepatic and renal clearance)takes its toll. Because the rates of peptide binding to both purified MHC and cellular MHC are particularly slow (half-lives of several h o r n with micromolar amounts of peptide under pseudo-fireborderreaction conditions),the rapid, in vivo degradation by serum peptidaaes of an unstable peptide, even if it is an excellent MHC binder, may prevent its usefulness aa a therapeutic agent. Generally, there are two methods for determining an enzymatic profile in biological fluids: blotting analysis and dimcbstability studies of the substrate(s). Enzyme blotting works well for known peptidases and proteinasea but may not reflect the enzyme activity of interest if the desired antibodies are not available. Alternatively, peptide-stability studies m a y 0022-3549/92.'~-0731%02.50/0

0 1992, Amenican PharmaceuticalAssociation

provide information on peptidase type and specificity, aa well as actually measure peptide half-lives, which are of interest for pharmaceutical development. We chose the latter method to probe the relative peptidase levels in human serum (HS)and synovial fluid (SF). Peptide stability in biological media, with a plethora of reaction conditions, has been previously reported, but few studies have s d c i e n t data for a comparison between different media. The most studied media are HS and plasma, with several reports of other serum types,- serum with cells, (animal) liver homogenate, SF, gastric fluid, brush-border homogenates, and isolated peptidases.7-9 To shed some light on peptide stability in biological media, particularly that in SF, we determined the stability of several peptides in the following media: 25%HS,25% SF, 10% fetal calf serum (FCS), and 5% mouse liver homogenate (MLH). The latter two media were included aa controls, because FCS is enzymatically immature (and is also a commonly used constituent of tissue culture media) and MLH is known to be rich in peptidases. The peptides studied herein have been used elsewhere to probe binding to MHC and are good model compounds for therapeutic treatment of autoimmune disease with peptides. In general, peptides exhibit a wide range of reactivity in biological media. For example, the biologically active peptides shown in Table I have half-lives ranging over several orders of magnitude in serum (or plasrna).S.6.6J~18Peptide modification to include synthetic amino acids, d-amino acids, or amide isosteres sometimes affords peptide entities of remarkable stability, with half-lives in serum of >loo0 min (e.g., ref 13). Herein we also report peptide stability constants ranging over several orders of magnitude, which are sometimes associated with subtle changes in peptide molecular structure, as well aa a good correlation between peptide stability in HS and that in SF from patients with rheumatoid arthritis.

Experimental Section Materials-Peptides were synthesized with a 430A peptide syn. thesizer or with solid-phase methods wing N-tert-butoxycarbonyl (t-Boc)-protectedamino acids.10 Cleavage of peptides from a phenyl-

Journal of Pharmaceutical Sciences I 731 Vol. 81, No. 8, August 1992

Table CPeptide Stablilty in HS (or Plasma) at 37 "C Peptide Thymopentin (TP5)

L x i l : 5 : Ac[Pro2]TP-5-NH2 A-Sleep inducing peptide [Leu5]enkephalin (&AlaZ)[Leu5]enkephalin LHRHb Nafarelin Dermorphin DEEd CCK 4' CCK 7' CCK 84 CCK 8-S03H CCK 9 analogueh CCK 10' des[Gly-NH,gvasopressin OxytEin derivative TRW TRH-OH GRH'

Sequence

Half-life, min

RKDW RkDW Ac-RPDW AC-RPDW-CONH, WAGGDASGE YGGFL YaGFL pyro-EHWSYGLRPG-CONH, pyro-EHWSYXZRPG-CONH, YaFGYPS-CONH, TSEKSQTPLVTL WMDF-CONH, YMGWMDF-CONH, DYMGWMDF-CONH, D(Y-SO3H)MGWMDF-CONHZ RD(Y-SO,H)MGWMDF-CONHZ DRD(Y-SO,H)MGWMDF-CONH, CYFQNCPR PLG-CONH, pyro-EHP-CONH, pyro-EHP YADAIFTNSY RKVLGQLSARKLLLQDIMSRQQGESNQERGARARL-CONH,

1

30 40 Stable 5 5,6' 63" 590 >1440 >180 20 13 5 18 50 3 30 >720 Stable 9 27 17

Reference 10 10 10 6 11 12,13 13 14 15 12 3 16 16 16 16 16 16 6 5 17 17 18

Corrected for 100% HS. Luteinizing hormone releasing hormone. X, 2-Naphthyl-dalanine. des-Enkephalin-yendorphin. 'Cholecystokinin-4. Cholecystokinin-8. Cholecystokinin-9 analogue. ' Cholecystokinin-10. Thyrotropin releasing hormone. Growth hormone releasing hormone. a

'Cholecystokinin-7.

acetomidomethyl resin (PAM;BaChem, Switzerland) was accomplished by HF cleavage with the appropriate scavengerslg; peptide purification was carried out by semipreparative, reversed-phase HPLC. Peptide identity was verified by amino acid analysis and, when required, by fast-atom bombardmentmass spectrometry. All other chemicals (reagent or HPLC grade) were purchased commercially (Sigma or Aldrich) and used without further purification. Pooled HS (type AB, not inactivated by heat; Irvine Scientific) was delipidated by centrifugation before use. All dilutions (v/v)of medium were made with RPMI-1040 tissue culture medium (J.R. Hazelton Biologics). SF from four patients with rheumatoid arthritis (obtained from the International Institute for the Advancement of Medicine) was pooled, filtered (0.22-pm filter) to remove particulates, and then diluted with RPMI-1040 tissue culture medium. Protein levels in HS and SF were determined by bicinchoninic acid (BCA) analysis. FCS was obtained from J. R. Hazelton Biologics, inactivated with heat a t 56 "C for 35 min to remove complement, filtered, and diluted with RPMI-1040 to make the final 10% FCS (v/v) solution. (Removal of complement is a standard practice in preparing tissue culture media and does not affect peptidase levels noticeably.) MLH was prepared by collecting mouse livers (-25) from nonfasted mice, carefully removing the gall bladders (another source of enzymatic activity), washing three times with ice-cold isotonic phosphate buffer, and grinding the tissue with a hand-held type A manual homogenizer. The pulverized m o u e livers were diluted with RPMI-1040 to make the final 5% (wlv), preparation and then spun a t 10 000 rpm (Sorval) for 20 min (4 "C) to remove debris and lipids. The supernatant was removed and frozen at -25 "C until needed. Reaction KineticeTypically, 1 mL of reaction solution in a 1.5-mL Eppendorftube was temperature equilibrated at 37 ? 1"C for 15 min before addition of 5 p L of peptide stock solution [ l o mg/mL in dimethyl sulfoxide (DMSO)]to make the final peptide concentration of 50 pg/mL. (This peptide concentration is high enough for easy UV detection a t 214 nm, without being too high to cause enzyme saturation.) The initial time was recorded, and a t known time intervals, 100 pL of reaction solution was removed and added to 200 pL of either 6% aqueous trichloroacetic acid (TCA) or ethanol. The cloudy reaction sample was cooled (4 "C) for 15 min and then spun at 14 000 rpm (Eppendorfcentrifuge)for 2 min to pellet the precipitated serum proteins. The following controls were carried out for each run. (1)Each set of stability experiments included a degradation study on a reference peptide, carried out in separate solution. Peptide 28 was used as a reference peptide because of its rather slow degradation in all the biological media studied. (2) Peptide stability was also determined in 732 I Journal of Pharmaceutical Sciences Vol. 81, No. 8, August 1992

'

precipitation supernatant containing 24% TCA to ensure that the peptides did not undergo acid-catalyzed degradation while awaiting HPLC sample analysis. (3) Sample recovery aRer precipitation of serum proteins with TCA or ethanol was determined by comparison of the peptide peak area at time zero with that of a peptide stock solution of known concentration in DMSO:H20 (1:l).Generally, 6% TCA showed the highest peptide recovery. However, when 0.98 (eight points).

Table ICPepNde Half-llverm

Source

Number 1 2 3 4 5

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

Chicken ovalbumin 370-81 Lambda repressor 12-26 Tetanus toxin 1272-84 Mouse hemoglobin HIV gp13 97-112

Hemoglutinin 302-13 Hemoglutinin 187-206 Pertussis toxin 64-75 HIV gp17 33-48

Myelin basic protein 75-98 Tetanus toxin 830-43 Tetanus toxin 830-43 Synthetic Tetanus toxin 947-69 Mywbacterium leprae Pertussis toxin 151-64 Mouse 82 microglobulin 31-46 Mywbacterium leprae Tetanus toxin 830-43 Tetanus toxin 830-43 Hemoglutinin 307-19 Tetanus toxin 83043 Tetanus toxin 830-43 Tetanus toxin 830-43 Tetanus toxin 830-43 Tetanus toxin 830-43 Tetanus toxin 830-43 Hernoglutinin 307-19

Peptide Sequence HIATNAVLFFGR YLEDARRLKAIYEKKK NGQIGNDPNRDILY VITAFSDGLK

QKQEPIDKELYPLTSL CPKYVRSAKLRM RTLYQNVG'TWSVGTSTLNK DNVLDHLTGRSC IVWASRELERFAVNPG GRTQDENPWHFFKNIVTPRTPPP QYIKKNSKFIGITE QYlKANSKElGlTE ARRLKAIYARRLK

FNNFTVSFWLRVPKVSASHLEQY YTLLQAAPALDKLKLTGDEATGANI R ILAGALATYQ

HPPHIEIQMLKNGKKI

TLLQAAPALDKY

pyro-QYIKKNSKFIGITE QY IKANSKFIGKTE Y PKFVKQNTLKLAT

QYlKANSKFlGlFE TYlKANSKFlGlTE QYlKANQKFlGlTE

QElKANSKFlGlTE QYIKANSKFKGITE QY IKANSKFIG ITE PKWKQNTLKLAT

HS 4 3 32 2 12 3 114 913 4 99 17 7 1 47 948 3 74 9 224 21 28 22 37 13 25

Calculated Half-life (min) in: SF FCS MLH 4 2 43

3 10 2 136 2760 4 211 11 3 2 29 458 2 78 24 41

10 37 10 11 12 14 5 6 23

8 9 27' ~

* Calculated half-lives in

3 19 120 4 14 31 136 704 4 91 37 16 11 203 600 3

90 9 240 12 26 21 85 21 21 15 15 51 ____

1 1 1 0.4 25 -b

11 18 1 0.3 4 2 0.1 1 2 1 24 1 2 3 1 3 2 2 17

2 1 1 ~

was obtained by dividing the observed half-life by the dilution factor. Not determined. The &amino acid-protected peptide pKYVKQNTLKLAt had a calculated half-life of >180 min (or >720 min in 25% HS). 100% medium

Results and Discussion Peptide degradation was studied in 25%HS, 25%SF, 10% FCS, and 5% MLH. These media were selected for the following reasons: (1) the lower concentrations (5-25%) of medium, because they increase peptide recovery and retard reaction kinetics to a manageable rate (the rates of peptide degradation a t low peptide concentration also vary proportionately with serum concentration, as shown later); (2) HS, for ita general applicability to drug development; (3) SF, because our peptide antagonists may be administered clinically by intrasynovial injection; (4) FCS, as an enzymatically immature control serum and because it is widely used in tissue culture media; and (5) MLH, as a control medium rich in peptidases and proteinases. The peptides studied have molecular weights ranging from 1090 to 2768 (10-25 amino acids in length) and showed no indication of solid-state degradation when stored for several months a t -20 "C. All peptides were soluble in DMSO a t 10 mg/mL and in aqueous solution (pH 7, isotonic) at 50 pg/mL without evidence of peptide aggregation or liquid-crystal formation. The peptide sequences (Table 11)show that some of the peptides are structurally distinct in that they arise from a number of antigenic sources. On the other hand, some sequences are quite similar, representing single amino acid substitutions that were synthesized to probe for MHC-binding specificity (not reported herein) and to determine if different enzymatic degradation pathways are followed. The degradation of these peptides was followed with stability-specific, reversed-phase HPLC (a sample set of HPLC chromatograms is shown in Figure 1). No attempt was made to determine the specific degradation pathways for these peptides (a monumental task), although the experiments (vide infra) and the literatures indicate that a predominant degradation pathway is exopeptidase cleavage. (For example,

peptide 28 has a half-life of -2 h in 25% HS, whereas its analogue, with d-amino acids at the N and C termini, has a half-life of >12 h under the same conditions.) Integration of the parent peptide peak areas to give pseudo-first-order plots (as shown, for example, in Figure 2) affords determination of the peptide half-lives by linear least-squares analysis [halflife (tllz)= 0.69/k,where k is the rate constant]. These half-lives have been corrected for the dilution factor (for example, dividing the half-life in 25% HS by 4 gives the calculated degradation half-life in 100% HS; Table 11).This method of calculating half-lives in 100%medium is supported by the linear dependence of degradation rate on serum concentration (Bee, for example, reference peptide 28 in Figure 3). When a single peptide was studied several times (for example, reference peptide 281, the value reported in Table 11 is the calculated average; rarely did the measured rate constants differ by >15%. In these experiments, peptide degradation was due solely to enzymatic activity; all peptides were quite stable in phosphate-buffered saline, with observed half-lives usually >20 000 min at 37 "C. Peptides with an N-terminal glutamic acid were converted to the pyroglutamic acid derivative in phosphate-buffered saline, with measured rate constants ranging from 5 x to 10 x s-l (tllz 7000-14 000 min) a t 37 "C.These chemical degradation rates were slow enough to consider as negligible their contribution to the overall degradation rate in biological media. The overall rate correlations for the data of Table 11can be seen by comparing the logarithm of half-life for the different media versus the logarithm of half-life for 25%HS, which was chosen arbitrarily as the standard (Figure 4).Peptide stability in HS correlates fairly well with that in SF from patients with rheumatoid arthritis (Figure 4A).For most peptides, the difference in reactivity was approximately twofold or less (either way), with the maximum variation being approxi-

-

Journal of Phamaceuticel SciencesI 733 Vol. 8 1, No. 8, August 1992

3 0

I

(u

I

I i

1

I

!

i

I

1 A

!/1 25

0

50

75

100

125

150

175

200

limo (mln)

J

Flgure2-Pseudo-first-order kinetic plots for degradationof 28 in (m) 10, (0)25, and (A)75% HS at 37 "C.

-c i

i

I

B

i

i 0 0 /

I

I

1

20

40

I

I

I

60 80 100 % Human Serum

-Effect of HS concentration on enzymatic degradation of *tide 28. All peptides studied in this manner (datanot shown)exhibited a linear dependence between degradation rate and medium concentration (with zero intercept; y = 0.042 + 0.025~;r = 0.998). Flaure

I

1

C

i

I

I

jI~

-

i

Flgure1-Reversed-phase HPLC chromatogramsfor the degradation of 28 in 25% H S at 37 "C: (A) 4 min; (B) 102 min; (C) 165 min. For most peptides, the degradation products are not alwaysobserved, presumably because of multiple reaction pathways (and thereforedifferent products) available to peptides of this size. The peak at 20.4 rnin is 28; the other

peaks are serum components.

mately fourfold. Although the protein contents of HS (-75 mg/mL) and pooled SF (37 k 14 mg/mL) are similar, the different cell populations in HS and SF still could lead to different peptidase profiles and, therefore, M e r e n t rates of peptide degradation. However, this does not occur for the peptides studied herein. Further, there was no correlation between the stability of 28 in the individual SF samples with the protein levels in SF (data not shown). Because peptide stabilities in pooled HS and pooled SF from patients with rheumatoid arthritis are quite similar, peptides administered 734 I Journal of Phannaceuficel Sciences Vol. 81, No. 8, August 1992

directly to the synovia should be as stable as those administered by other parenteral routes. Several of the calculated rates of peptide degradation in FCS are somewhat slower than those in HS (Figure4B), presumably because of the reduced peptidase levels in FCS.FCS also has lower antibody levels than HS (or mature bovine serum);thus, FCS is superiorfor culturing human cellsthat might be otherwise overwhelmed by a cytotoxic antibody response. Thus, tissue culturemedia prepared with FCS instead of HS may be preferred when peptide stability is important (as may be the case for cellular assays of inhibition of antigen presentation to define peptide antagonists for MHC blockade). Finally, peptide degradation was determined in a peptidaae-rich tissue homogenate, MLH. Most of the peptides studied herein were degraded more rapidly in MLH than in HS, even though no special precautions were taken (other than freezing) to prevent homogenate autodegradation. Also, enzyme cofactora were not added, as has been done in many other liver homogenate assays. As expected, there was little correlation of the degradation rates in MLH with degradation rates in the other biological media tested. Many of the peptides in Table I1 have calculated half-lives of