Protein Formulation
`and Delivery
`Second Edition
`
`11
`
`Edited by
`Eugene J. McNally
`Gala Biotech, a Catalent Pharma Solutions Company
`Middleton, Wisconsin, USA
`Jayne E. Hastedt
`ALZA Corporation
`Mountain View, California, USA
`
`informa
`
`healthcare
`
`New York London
`
`Bausch Health Ireland Exhibit 2010, Page 1 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`lnforma Healthcare USA, Inc.
`52 Vanderbilt Avenue
`New York, NY 10017
`
`© 2008 by Jnfomm Healthcare USA, Jnc.
`lnforma Hcolthcnre is an lnfomHl business
`
`No claim lo original U.S. Government works
`Printed in the United States of America on acid-free paper
`10 9 8 7 6 5 4 3 2 l
`
`International Standard Book Numbcr-10: 0-8493-7949-0 (Hardcover)
`lntemational Standard Book Number-I 3: 978 -0-8493-7949-9 (Hardcover)
`
`This book contains information obtained from au thentic nnd highly regarded sources. Reprinted materilll is
`quoted with permission. i.tnd 1'0urces arc indica1ed. A wide variety of references are listed. Rcasonable e(fons
`have been made to publish reliable data and information. buc the au(hor and lhc publisher cannot assume
`responsibility for lhe validity of all materials or for the consequence of their use.
`
`No pan of this book mny he rcpn nted, n:produced, transmilled. or utilized in any form by any elecu·onic, mocltan(cid:173)
`lcal. or other means. n w knn1Y11 orltcrt,,ner invented, including photocopying, microfilming, ond n:oording. or
`in any infonnmion sLora_ge or relricvol system. without wriuen permission from the publishers.
`
`Vor pcmu,-ton w phu1ot•upy or u~c mureriul clc.:ll\1111c,11ty l'roon thh 1vork. plea~c r,ecc..,, www.uop1'righ1.
`crnn (h11p://www.copyrip hl """'" , r cc,nlur< lh~ Copyri; ht Cl~ar1,ncc Center. Inc. (C C) 222 R11>cwood
`Dctv ·, D,u,vcr;, MAU IY-3. Q7g.7~0. 400, ( '
`,, " not-for,p111fil or~n11i,;11,vn ihnt pru,•ldc, Ii ·cn,r, and
`' ·t1,11 Jtiou I'll• ,, VIII ic()· ur u, " · ror ur8UUl~auon, thul l1uv' "'Cl' gnmtc,1 II 1•hotocopy li,rn,c by tltb cc,
`Q scparDle system of payment has bcon arranged.
`
`Trademark Nolicc: Product or corporate names may be trademarks or registered trademarks. and arc used
`only for identification and explanation wi1hou1 in(ent lo infringe.
`
`Library of Congress Calaloging-in-Puhllcalion Data
`
`Protein formulation und delivery/ edited by Eugene J. McNally, Jayne E. Hnsted1. - 2nd ed.
`p.; c.m. --
`(Drugs and the pharmaceutical sciences ; 175)
`Includes hihliographical references and in<lex.
`!SBN-13: 978-0-8493-7949-9 hardcov•r : olk. paper)
`ISBN-I 0: 0-8493-7949-0 (hardcover : alk. paper\
`I. Protein drugs--Dosagc forms. 1. McNally. Eugene J., 1961 - II. Hastedt. Jayne E.
`Ill, Series: Drugs and the pharmaceutical sciences; v.175.
`[DNLM: I. Protein Conformation. 2 Drug Delivery Sys1ems. 3. Drug Design.
`4. Drug Stabi lity. 5. Pro1eins--adm1nistration & do.s•8•· WI DR893B v.175 2007
`I QU 55.9 P9667 20071
`
`RS43 I.P75P77 2007
`6l5' .19- dc22
`
`Visit the lnforma Web site at
`www,infonna.crnn
`
`and the lnforma Healthcare Web sire at
`www.i nformahenlthcere.com
`
`2007023435
`
`Bausch Health Ireland Exhibit 2010, Page 2 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`2
`
`Chemical Considerations in Protein
`and Peptide Stability
`
`Paul M. Bummer
`University of Kentucky, Lexington, Kentucky, U.S.A.
`
`0EAMI0ATION
`
`Introduction
`
`The dearnidation reactions of asparagine (Asn) and glutamine (Gin) side-chains
`are among the most widely studied noncnzymatic covalent modifications lo
`proteins and peptides ( 1- 7). Considerable research efforts have been extended
`to elucidate the detaiJs of the dcamidation reaction in both in vitro and in vivo
`systems, and a number of well-written, in-depth reviews are available (1- 5,8,9).
`This work touches on ly on some of the highlights of the reaction and on the roles
`played by pH. temperature. buffer, and other formulation components. Possible
`deamidation-associated changes in the protein structure and state of aggregation
`also are examined. The emphasis is on Asn deamidation, si nce Gin is significantly
`less reactive.
`
`Reaction Mechanism
`
`The primary reaction mechanism for the deamidation of Asn in water-accessible
`regions of peptides and proteins al basic or neutral conditions is shown in Figure I.
`For the present, discussion is confined to the intramolecular mechanism, uncom(cid:173)
`plicated by adjacent amino acids at other points in the primary sequence. Under
`alkaline conditions, the key step in the reaction is lhe formation of a deprotonated
`amide nitrogen, which cruTies out the rate-determining nucleophilic attack on the
`side-chain carbonyl, resulting in a tetrahedral intermediate and finally the formation
`of the five-member succinimide ring. For such a reaction, the leaving group must be
`
`7
`
`Bausch Health Ireland Exhibit 2010, Page 3 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`8
`
`Bummer
`
`~NH2
`
`R..,_N.y~,.,
`
`0
`Asparaglnyl Residue
`
`R, ~ :
`"N~ -~R1
`0
`
`Reactive f nion
`
`R..._ / ~ N::R NHYN 1
`
`0
`
`Aspartyl Residue
`
`lsoaspartyl Residue
`
`Figure 1 PmpuseJ reucllou 111cchanism for deamidation of asparaginyl residue. Note the
`format ion ot the ,11ccinhnillyl 1111ermcllifite and the two possible final products.
`
`easily protonated, and in this case, it is responsible for the characteristic formation of
`ammonia (NH3) . The succinimide ring inte1mediate is subject to hydrolysis, result(cid:173)
`ing tn either the corresponding aspartic acid or the isoaspartic acid (l3-aspartate).
`Often, the ratio of the products is 3: 1, isoaspartate to aspartate (l 0-12). In the ca~e
`of acid catalysis (pH < 3), a tetrahedral intermediate is also formed, but breaks down
`with the loss of NH3 without going through the succinimjde (I 3- 17). The reaction
`also appear!; to be sensitive to racemization at the a.-carbon, resulting in mixtures of
`o- and ,.-isomers ( \0, 13-15). The rate of degradation of the parent peptide in aque(cid:173)
`ous media often follows pseudo-first-order kinetics (16,17).
`
`Bausch Health Ireland Exhibit 2010, Page 4 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`Chemical Considerations in Protein and Peptide Stability
`
`9
`
`A number of other altemative reactions are possible. The mosr prevalent
`1·eactio11 appears to be a nuclcophilk :JLtuck of the Asn sicle-chain amide nitrogen
`on tile peptide carbonyl. resulting in main-chain cleavngc ( I 0, 16, 18). Thii. reac-
`11011 (Fig. 2) is slower than thol of cyclic i.mjde formation :t11d L~ mc,st frequently
`observed when Asn is followed by proli11e. :i residue incapable of forming an inn•
`ii.ed peptide-bond nitrogen.
`
`pH De.pendence
`
`Under conditions of strong acid (pH l-2), deamidation by direct hydrolysis of I.he
`amide side-chain becomes more favorable than formation of cycl.ic imidc (] 6, 19).
`Under these extreme conditions, the reaction is often complicated by main-chain
`cleavage and denaturation. Deamidation by this mechanism is not likely to pro(cid:173)
`duce isoaspartate or signiticant racemization ( 16).
`Under more n1oder:1te conditions, the effect of pH is the result of two
`opposing reactions: (i) deprotonacion of the peptide-bond oi trogen, promoting
`
`R 1 = Amino end of protein
`
`R2 "' Carboxyl end of protein
`
`Figure 2 Pmposed reaction mechanism for mam-chain cleavage by asparaginyl residues.
`
`Bausch Health Ireland Exhibit 2010, Page 5 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`10
`
`Bummer
`
`the reaction and (ii) protona.tion of the side, chain-leaving group, inhibiting the
`reaction. Tn deamidation reactions of short chain peptides uncomplicated by
`structural alterations or covalent dimcri:r.ation (20), the pH-rate profiles exhibit
`the expected "V" shape, with a minimum occurring in the pH range of 3 to 4 (16).
`Computation studies by Peters and Trout (21) have been helpful in shedding light
`on the effect of pH . These authors have suggested that under mildly acidic condi(cid:173)
`tions (3 < pH, 4), the rate-limiting step is the attack of the deprotonated nitrogen
`on the side-chain. The rate-limiting step at neutral pH is the hydrogen transfer
`reaction, while under basic conditi ons (pH > 7), it is the elimination of NH2 from
`the tetrahedral intennediate. Experimental studies have shown that the increase
`in rate on the alkaline side of the minimum docs not strictly correlate with the
`increase in deprotonation of the amide nitrogen, indicating that the rate of reac(cid:173)
`tion is not solely dependent on the degree of the peptide-bond nitrogen deprot(cid:173)
`onation ( 16, 19). The pH minimum in the dcamidation reaction measured in vitro
`for protein s may (22) or may not (23) fall in the same range as that of irnple
`peptides. Overall pH-dependent effects may be modified by structure-dependent
`factors , such as dihedral angle flexihility, water accessibility, and proximity of
`neighboring amjno acid side-chains (see section Peptide and Protein Struct\ire).
`
`Effect of Temperature
`
`The temperature dependence of the deamidation rate has been studied in a variety
`of simple peptides in solution (16,24,25). Small peptides are easily designed to
`avoid competing reactions-, such as oxidation and main-chain cleavage, and arc
`thus useful to isolate attention directly on the dcamidation rate. Jn solution, deam(cid:173)
`idation of small peptides tends to follow an Arrhenius rclarionship. Activation
`energies of the reaction do tend to show pH dependence, and a discont inuity in the
`Arrhenius plot is expected when the mechanism change from direct hydrolysis
`(acid pH) to one of cyclic imide (mildly acidic to alkaline pH).
`The deam.idation rate of proteins also hows temperature dependence
`(23,26,27) under neutral pH. For deamidation reactions alone, temperature(cid:173)
`associated rate acceleration in proteins may be due lo enhanced flexibility of the
`molecule, allowing more rapid formation of the cyclic imide (28). or it may occur
`by cataly i by side-chains brought into the vicinity of the deamidation site (5).
`The availability of water appears tu be an important determinant in
`temperature-associated effects. In studies of lyophilized formulation of Val-Tyr(cid:173)
`Pro-Asn-Gly-Ala, the deamidation rate constant was ob •erved to increase about an
`order of magnitude between 40°C and 70°C (29). ln contrast, in the solid state, the
`Arrhenius relationship was not observed. Further, the deamidation in the solid state
`showed a marked dependence upon the temperature when the peptide was lyophi(cid:173)
`liz.ed from a soluti on of pH 8, while little temperature dependence was observed
`when lyophili7,ation proceeded from olutions at either pH 3.5 or pH 5. The authors
`related this temperature difference to changes in the reaction mechanism that may
`occur a a function of pH.
`
`Bausch Health Ireland Exhibit 2010, Page 6 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`Chemical Considerations in Protein and Peptide Stability
`
`11
`
`Adjuvants and Excipients
`
`The influence on deamidation by a variety of buffer ions and solvents has been
`examined. As pointed out by Cleland et al. (4) and reinforced by Tomizawa et
`al (13), many of these additives are unlikely to be employed as pharmaceuti(cid:173)
`cal excipienis for formu latjon, but they may be employed in protein isolation
`and purification procedures (30) . Important clues to stabilization strategies can
`be gained from these studies. In the follow ing, it is fruitful to keep in mind the
`importance of the attack of the ionized peptide-bond nitrogen on the side-cbain
`carbonyl and the hydrolysis of the cyclic im ide (Fig. I).
`
`Buffers
`
`Bulfer catalysis appears to occur in some but not all peptides and proteins studied
`(5). Bicarbonate (16) and glycine ( 12) buffers appear to accelerate deamidatioo. On
`one hand, the phosphate io n has been shown to catalyze deam idation, both in pep(cid:173)
`tides and in proteins (12, 13,16,31-34), generally in the concentration range ofO to
`20 mM , Capasso et al. (35,36) observed the acceleration of deamidalion by acetate,
`carbonate, Tris, morpholine, and phosphate buffers only in the neutral to basic pH
`ranges. On the other hand, Lura and Schrich (37) fo und no influence on the rate of
`deamidation of Yal-Asn-Gly-Ala when buffer components (phosphate, carbonate, or
`imidazole) were varied from Oto 50 mM. A general acid-base mechanism by which
`the phosphate ion catalyzes deamidation was challenged in 1995 by Tom.izawa et al.
`(l 3), who found that the rate of lysozyme at I 00°C did not exhibit the ex pected Li near
`relationship of deamidation rate on phosphate concentration. Although not linked to
`deamidation, it is worthwhile to note that at pH = 8 and 70°C, tris(hydroxymethyl)
`aminomethane buffer (Tris) has been shown to degrade to liberate highly reactive
`formaldehyde in forced stability rudies of peptides (38).
`
`Io nic Strength
`
`The effects of ionic strength appear to be complicated and not open to easy gen(cid:173)
`eralizations. Buffer and ionic strength effects on deamidation are evident in pro(cid:173)
`teins al neutral to alkaline pH (5). In ·elected peptides and proteins, the catalyt i
`activity of phosphate has been shown lo be reduced moderately in the presence
`of salts NaCl, LiCI, and Tris HCI (12, 13). Of these salts, NaCl showed the least
`protective effect aga inst dcam idation ( 13).
`Jn the peptide Gly-Arg-Asn-Gly at pH 10, 37°C, the half-life t,n of dearni(cid:173)
`dat ion dropped from 60 hours to 20 boucs when the ionic strength was increased
`from 0.1 to 1.2 (22). However, in the case of Val-Ser-Asn-Gly-Val at pH 8, 60°C,
`there was no observable difference in the ,,,2 of deamidation when solutions with(cid:173)
`out salt were compared to those containing I M NaCl or LiCI ( 12). Interestingly,
`for lysozyme at pH 4 and 100°C, added sail showed a protective effect against
`deamidation, but only in the presence of the phosphate ion (13) .
`In reviewi ng the data above, Brennan and Clarke (17) tentatively attributed
`the promorion of deamidation by elevated levels of ions to enhanced stabilization
`
`Bausch Health Ireland Exhibit 2010, Page 7 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`12
`
`Bummer
`
`of the ionized peptide-bond nitrogen, promoting attack on the side-chain amide
`ca rbonyl. Other mechani. ms would include di. n.1p1ion of t r1i:1ry . 1mcrure in pro(cid:173)
`teins that may have tabilized Asn residues. in some as-ye1 unknown fn8hi n. Thul
`promotion of deamidation is observed ins me cases of peptide .• and inhibiliull io
`others does !-ugges1 rather complex und competing effects. Clearly, the stabilizing
`effects, when observed at all, are often at levels of salt too concentrated for most
`pharmaceutical formulations.
`
`Solvents
`
`The effect of various organic solvents on the rate of dcamidation has not received
`much attention; it would be expected, however, that in the presence of a reduced
`dicle tric 1111:llium. the peplide-b od nitrogen would b ·le -~ likely to i nizc. Since
`rhc anionic peptide-bond nitrog n is necessary in the formation or the cyclic
`imide, u J w dielectric mediwn would retm·d the progress ul' the rem::tion and be
`reflected in the free energy difference for ionization of the peptide-bond nitrogen
`(17). F 1llowing thi1- hypolh sis, Brennan and
`larke CW) unnly7,ed succi nimidc
`formation of rh pcplide Vnl-Tyr-Pru-Asn-Gly-Ala lrhc ~ame peptide cmpl yed
`by Patel and BllrchaJdt ( 16) in studies o pH effccls in aqueous solution] as a !'unc(cid:173)
`tion of organic co. olvent (ethanol, glyt:erol, and dioxin) al constant pH and ionic
`strength. The lower dielectric constant media resu lted in a significantly lower rate
`of deamidalion, in agreement wilh 1he hypothesis. II wtis argued 1h01 1hc ·imilar
`rates of deamidation for different cosolvenl systems 1>f' th same effective dielec(cid:173)
`tric constant indicated that changes in visc<r ity und Wtller content of the medium
`did not play a sigoificant role.
`The effect of organic cosolvenls on deamidation in proteins is even less
`well characterized than that of peptides. Trifluoroethanol (TFE) inhibits deami(cid:173)
`dation of lysozyme at pH 6 and l00°C ( 13), and of the dipeptide Asn-Gly, but
`does not inhibit the deamidation of free ami no acids. The mechanism of protec(cid:173)
`tion is not clear; direct interaction of the TFE with the peptide bond was postu(cid:173)
`la1i:d . i-.u1 noi demonstrated. An 31ferna1ive hyp(11he. i~ i. 111111 TFE induce. gre:ner
`tructural rig[dily in th pro1ei11 , produ ·ing ti strutlurc ·omcwhat resistant to the
`format io11 of 1he cyc lic imiue in1cr111cdio1e. Orher. pharrnaceu1ically ncceprnble
`solvent5, ethanol and glycerin, did not exhibit the same protective effects as TFB
`on lysozyme.
`Of course, in dosage form design. organk solvents such as TFE are not use(cid:173)
`ful as pharmaceutical adjuvants. The effect~ of low dielectric may still supply a
`rationale for the solubih1.11tion of peptides in aqueou · surfactant systems, where
`the b.ydrophobic region of a micelle or llposome could potentially enhan e the
`stabilization of theAsn residues from deamidation. As pointed out by Brennan and
`Clarke ( 17), the results of experiments in organic solvents can have implications on
`the prediction of points ol' d amidulion in proteins as weU. For Asn re: idues near
`the urface of the protein, where the <lieleclric constant is expected to approach that
`of water. the deamidarion rate would be expected to be high. For Aso residues bur(cid:173)
`ied in more hydrophobi regions of the protein, where poli'lfitics arc thought 10 be
`
`Bausch Health Ireland Exhibit 2010, Page 8 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`Chemical Considerations in Protein and Pepride Stability
`
`13
`
`more in Line with that of ethanol or dioxane (40), reaction rates would be expected
`to be coasidcrably slower:
`Compuuttional studies on the effects or solvent on tht: reaetiun were carriei.l
`oul rece111ly by Catak el uL (4 1 ). They rC[Jllr1 !hat, in 1he absence ol' water, the
`overall activ~tion energy barrier iN on rbe on,h, r of50 kcal/mol, :111d that lhis drops
`Lo a value of about 30 kcal/mo I in the presence of water. In all, about three water
`molecules participate directly in the reaction, assisting in hydrogen transfer and in
`the cyclization outlined in Figure I ,
`
`Po lymers and Sugars
`
`Considerable interest has developed in the swbilization of proteins and peptides in
`solid matrices, eitber polymeric or sugar based. In most solid polymer matrices,
`the primury role is Lo improve phurm11cokinet1c amJ pharmcodynmnic propenies
`of the active by modifying release characterislics nnd most s tudies arc desigr1cd
`with this intention i11 mi nd (42). Sugars nre us uuUy employed as an aid 10 lyophi(cid:173)
`lization of proteins, with the intent of maintaining the tertiary structure and pre(cid:173)
`venting aggregation (43).
`The state of the polymer and the activity of water appear to be critical factors
`in the stabilization of the peptide against deamidatio11. In general, the observed
`degr:1dution rate constsnts exhibit the fo llowing rank order: solution > ru bbery
`polymer> glassy polymer (38,44-46), However. this observation does 1101 appear
`to be valid in every case (44). It has been pmposcd that up 10 30% of a peptide
`may bind to polyvinylpyrrolidone (PVP) in the solution state, complicating the
`kinetic analysis (47),
`Peptide siability in polymer n1au·ices that a,·e them~elvcs also undergoing
`dcgr:1d:11ion provides a unique challenge. Por example. it has been observed that
`PVP may ftlm1 at.ltlue1, wiLh lhe N-terminu, nf re pt ides (48). Sy~1emnlic,1udie1, of
`1be dean1idation of u 111odel 11eptidc iu film!> or the copolymer polylactic-glycolic
`acid (PLGA) have shown that the reaction is the primary route of degradution
`only ofter longer storage limes ut higher water content (49). The delay in the
`onset o f deamidation of peptide in PILGA may be reluled to the time nece.~$ary to
`establish an "acidic microclimate" that arises from the hydrolysis of the polymer
`(50). In support of this acid-catalyzed deamidation hypothesis in PLGA films, the
`reaction product. ison~partutc was not found.
`Computational studies may bUpply additional insight. Computer simula•
`tions of the mobility of pe ptide . wn1er. NH,, and polymer in PVP matrix have
`been carried out by Xiang and Anderson (51). They observed that the diffusiv(cid:173)
`ity of water, NH3• and peptide were hetween two and three orders of magnitude
`slower in PVP compared to aqueous solution. Importantly, the con formational
`dynamics of the peptide in the g la~sy polyme.r exhibited a higher energy ban-iet
`between states than seen for the peptide in water. Thus, two of the critical events
`in the process of deamidation, the conformational chonges necessary to form the
`cyclic intermediate in the glassy polymer and the diffusion away of the NH3 after
`release, are both slowed considerably in the solid state-.
`
`Bausch Health Ireland Exhibit 2010, Page 9 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`14
`
`Bummer
`
`The effect of sugars on the deamidation of a model peptide has been exam(cid:173)
`ined. At pH= 7, a solution of peptide in a 5% sucrose or mannitol reduced the
`deamidation rate to about 16% of that found in the ab ·encc of sugars (52). When
`stored in the solid state, the rate of reaction was even slower, although sucrose
`appeared to s tabilize the peptide m , grcuter ex tent than did mnnn.itol. Lt was
`observed that sucrose rema.med amorphou. during the ttist period whi le mannitol
`crystallized, complicating the interpreta1io11 of th daw (53). Cle land et al. (43)
`determined rhat 360: I wa the optimal sugar-antibody molar ratio necessary to
`inhibit aggregation and deamidation over a three-month period. Sugars sucrose,
`trehalose, and mannitol were able to stabilize the protein so long as less than 8.4%
`moisture was present.
`Our understanding of the stabilization of peptides and proteins in polymer
`and sugar matrices is far from complete, and additional insight into the molecular
`mechanism might benefit from the bount y of studies c:11Tied out with small mol(cid:173)
`ecules in sl milnrsy ·tems. E ·perimenls must be designed ciu·efully and interpreted
`With caution so as to clearly separate the solvent effects or water and perhaps even
`NH3 on the reaction from the plastisizing effects on the matrix.
`
`Peptide and Protein Structure
`
`·n r Gin rtsiduc
`1'he :1bility to identify whi h
`r
`in II thempeuti protein
`p ptid m;1y be ulnerable 10 dean id;11i n w uld have gr :1t pra ·ti al appli :1-
`1 011 in preformulution and lormulutioo s tudies. The effc ·ts of' various level of
`tl'ucrure.-primary, secc ndary, and wrtiary-un:: believed to be mnpleK aucl
`varied. At present, only primary structure effects have been chanu:tcrizcd in a
`systematic man ner.
`
`Primary Sequence
`
`The primary seq uence of amino acids in a peptide or protein is often the first
`piece of structural data presented lo the formulation scientist. Considerable effort
`has been spent to elucidate the influence of flanking amino acids on the rates of
`deamidation of Asn and Gin residues. The potential effects of flanking amino
`acids are best elucidated in simple peptides, uncomplicated by side reactions or
`secondary and tertiary structure effects.
`
`In an extended series of
`Effect of amino acids preceding Asn or Gin:
`early studies, Robinson and Rudd (24) examined the influence of primary sequence
`on the deamidation of Asn or Gin in the middle of a variety of pcntapeptides. Mild
`physiologic conditions (pH 7.5 phosphate buffer at 37°C) were employed. A few
`general rules can be extracted from this work:
`
`1.
`
`In practically every combination tested , Gin resi dues were less prone
`to deamidation than Asn. For the two residues placed in the middle of
`otherwise identical host peptides, the half-life of the reactions differed
`hy a factor ranging from two- to threefold.
`
`Bausch Health Ireland Exhibit 2010, Page 10 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`--
`
`Chemical Considerations in Protein and Peptide Stability
`
`15
`
`2.
`
`In peptides Gly-X-Asa-Ala-Gly, steric hindrance by unionized X side(cid:173)
`chains inhibits dcamidation. The rank order of deamidation rate found
`was Gly >Ala> Val> Leu> Ile, with the t 1n ranging from 87 to 507
`days. It remains unclear why bulky residues inhibit the reaction, but
`reduced Aexibi lity of the sequence may be a factor. A similar effect was
`noted when Gln replaced Asn. In this case, 1112 ranged from 418 to 3278
`days, in accordance with the diminished reactivity of Gln,
`3. For the same host peptide, when the X side-chain was charged, the deam(cid:173)
`idation rate of Asn followed the rank order of Asp> Glu > Lys > Arg.
`
`Effect of amino acids followingAsn or Gin: Early experiments on di pep(cid:173)
`tides under extreme conditions indicated a particular vulnerability of the Asn-Gly
`sequence to deamidation (54). More recent studies of adrenocorticotropic hor(cid:173)
`mone (ACTH)-like sequence hexapeptide Val-Tyr-Pro-Asn-Gly-Ala under physi(cid:173)
`ologic conditions (55) have velified that deamidation is extremely rapid (t,n of 1.4
`days at 37°C). The formation of the ~uccinimide intermediate is thought to be ihe
`basis for the sequence dependence ( 10) of deamidation. ll is generally believed
`that bulky residues following Asn may inhibit sterically the formation of the suc(cid:173)
`cinirnide intermediate in the dearnidation reaction.
`Stcric hindrance of the cyclic imide formation is not the only possible
`gen~..sib of sequence-dependent deamidation. The resistance to cycllc irnide
`fonn::ulon in the presence of a carboxyl-fln.nking pro line peptide may be related
`to the inability of the prolyl amide nitrogen to attack the Asn side-chain (10).
`The computational studies of Radkiewicz et al. (56) suggest that tl1e effect of
`the adjacent residue may largely be attributed to electrostatic/inductive effects
`influencing the ability of the peptide nitrogen atom lo ionize (as seen in Fig. I).
`In the case of glycine, the inductive effect is insufficient to explain the results,
`and the authors argue that the ability of glycine to sample more conformational
`space compared to other amino acids may help stabilize the nitrogen anion.
`Experimentally, the replacement of the glycyl residue with the more bulky
`leucyl or prolyl residues resulted in a 33- to 50-fold (respectively) decrease in
`the rate of deamidation (10). Owing to the highly flexible nature of the dipep(cid:173)
`tide, the deamidation rate observed in Asn-Gly is thought to represent a lower
`limit.
`In more recent studies. deamidation of Val-Tyr-X-Asn-Y-Ala, a peptide
`sequence derived from ACTH, was examined with different residues in both
`flanking positions (57). When X was histidine (and Y is glycine), no acceleration
`of deamidation was found relative to a peptide where X is proline. Placing a His
`following the Asn was found to result in similar rates of deamidation when X was
`phenylalanine, leucine, or valine. The rate when X was histidine was slower than
`that of alanine, cysteine, serine, or glycine. These results indicate that histidine
`does not have unique properties in faci litating succinimide formation. Of inter(cid:173)
`est was the observati.on that histidine on the carboxyl side of the Asn did seem to
`accelerate main-chain cleavage products.
`
`Bausch Health Ireland Exhibit 2010, Page 11 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`16
`
`Bummf:r
`
`Some of the general rules for peptides may also show higher levels of
`depenJcnce on primary sequence.1'yler-Cmss :md Schrich ( 12) m 1died the influ(cid:173)
`ence or different amiao acids on lhe udj acem amino end of 1he pcntapcptide Val(cid:173)
`X-Asn-Ser-Val at pH 7 .3. For X = His, Ser, Ala, Arg, uni.I Len, deamidnLion rates
`were essentially constant and approximately seven times slower than the Val(cid:173)
`Ser-Asn-Gly-Val standard peptide. Of special interest to the investigators was the
`observation of no difference in deamidation rates betWeen those amino acids with
`and without ~-branching (Suell as voJine for glycine). This Is in direcr contrnSL 10
`the findings of Rohinson and Rudd (24) o f l<Hold diffcrem:es in deamidat1on
`for Valine .~11hsti1ution for g lycine in Gly-X-Asn-Ala-Gly, ~howo e,1rJier. Under
`the mild alkal ine conditions of Patel ood Borchardt ( I 6J, V~I-Tyr-X-Asn-Y-Alu,
`no difference in the deamidation rate constants was observed when praline was
`substituted for glycine in the X position.
`
`Data mining: Data-mining approaches have been employed to formulate
`a semiquantitative means of predicting the effect of primary structure on rate of
`deamidation. Capasso (58) proposed the extrathermodynamic relationship shown
`in Equation I :
`
`( I)
`
`Herc k1 is 1he observed rate constant for deumidntion, Xr is the average contribu(cid:173)
`uon of the speC'ific amino acid 1hn1 precede~ Asn. Yv i~ the average contribution
`due 10 l11e amino acid that follows A~n. and ZA.,, is the value when hoth the preced(cid:173)
`ing and following amino acids nre glycin(!. Over 60 peptides were included in 1he
`dambase. A~ expected, the grcate~t influence 011 1hc deamlclation rate in peptides
`wa~ found Lo arise from the identity of the following nmino acid. Some of the
`values for Y,. arc listed in l'.1ble I . A~ sugge~ted previously. relntive 10 the effect
`or g lycine. bulky hydrophobic :11111110 acids such 3S v111lne, leucine, and isoleucinc
`
`Table 1 Rate Constants Reported for the Reaction' of
`OH' with the Side Chains of Selected Amino Acids (IO I )
`
`Amino acid
`
`Cysteine
`Tyrosine
`Tryptopban
`Histjdine
`Methionine
`Phenylalanine
`Arginine
`Cystine
`Serine
`Alanine
`
`k {L/mole-s)"
`
`4.7 X 1010
`1.3 X 1010
`1.3 X IQIO
`5 X 10'
`8.3 X 109
`6.Sx 10'
`3.5 X 109
`2.1 X 109
`3.2)(10ti
`7.7 X 101
`
`•Mos I values dctennined via radiolysis.
`''The pH values of mnny of these studies have not been listed.
`
`Bausch Health Ireland Exhibit 2010, Page 12 of 38
`Mylan v. Bausch Health Ireland - IPR2022-01102
`
`

`

`r
`
`Chemical Considerations in Protein and Peptide Stability
`
`17
`
`appear to show the slowest deamidation rate, while the smaller, more polar histi(cid:173)
`di ne and serine show a rate closer to that of glycine. There do appear to be some
`discrepancies between these results and those mentioned earlier ( 12, 16.57), in
`particular with respect to the experimentally observed effect of histid ine. Clearly,
`different databases may give different results. At best, Equation l may be viewed
`as a first approximation for estimating deamidation rates in fonnulation scudies.
`Robinson et al. (59-61) have taken a different approach to mining by
`including means to account for the three-dimensional structure of the side-chains
`a nd by avoiding the use of data gathered in the presence of the known catalyst,
`phosphate buffer. A method ha been proposed to estimate the deamidation reac(cid:173)
`tion half-life at 37°C and pH= 7.4 based on the primary sequence (61). The
`extent to which this method may accurately predict the deamidation rate of pep(cid:173)
`tides in pharmaceutical systems has not yet been rigorously tested experimen(cid:173)
`tally, but if proven valid, it would supply a rather useful tool in guiding early
`formulation studies.
`
`Secondary and Tertiary Structure
`
`X-ray or nuclear magnetic resonance (NMR) data can provide a detailed map of
`the three-dimensional structure of the protein or peptide. The role of secondary
`and tertiary structures in intramolecular <leamidation of proteins has been dis(cid:173)
`c us ed by Chazin and Kossiakoff (62). It ls beyond the scope of this work to
`present a comprehensive review of the details of deamidation reacti.ons in spe•
`cific proteins. Excellent reviews of a variety of specific proteins exist (6). For the
`most pan, detailed mechanisms relating the secondary an<l tertiary structures of
`proteins to enhancement of rates of deamidation are not yet avai lable.
`Clear differences in the deamidation rates of some proteins are evident when
`native and denatured states are compared ( 13,63). Denaturalion is thought to enhance
`main-chain flexibility and water accessibility (62), Sufficient conformational flex(cid:173)
`ibility is required for the Asn peptide to assume the dihedral angles of <I> =- l 20°C
`and 'I'=+ 120°C necessary for succinimide formation . ln as much as such angles
`tend to be energetically unfavorable (64) in native proteins, it may be expected that
`Asn residues in the midst of rigid secondary structures, such a helices, may be
`resistant