FEBS 17787
`
`
`
`
`
`FEBS Letters 397 (1996) 225-229
`
`Conservation of amino acids in multiple alignments: aspartic acid has
`
`
`
`
`
`unexpected conservation
`
`
`
`Andras Fisera,b, Istvan Simonb, Geoffrey
`J. Bartona,*
`
`"University of Oxford, Laboratory of Molecular Biophysics, The Rex Richards Building, South Parks Road, Oxford OXJ 3QU, UK
`
`
`
`
`
`
`
`
`
`
`
`
`
`6 Institute of Enzymology, Biological Research Center, Hungarian Academy of Sciences,
`PO Box 7, Budapest
`
`H-1518, Hungary
`
`
`
`Received 18 July 1996; revised version received 25 September 1996
`
`
`
`
`
`
`
`
`
`tertiary structure. Our analysis complements that of Overing­
`
`
`
`
`
`
`
`ton et al. [8] who considered only pairwise substitution fre­
`
`
`
`quencies for amino acids in structurally aligned families.
`
`2. Materials and methods
`
`Abstract Analysis of the relationship between surface accessi­
`bility and amino acid conservation in multiple sequence
`alignments of homologous proteins confirms expected trends
`for hydrophobic amino acids, but reveals an unexpected
`difference between the conservation of Asp, Glu and Gin. Even
`when not in an active site, Asp is more highly conserved than Glu.
`There is a clear preference for conserved and buried Asp to be
`present in coil, but there is no tendency for Asp to conserve q,/'lf in
`the ++ region of the Ramachandran map. Glu does not show any
`preference to be conserved in a particular secondary structure.
`Analysis of recently derived substitution matrices (e.g. BLO­
`SUM) confirms that Glu tends to substitute mote frequently with
`other amino acids than does Asp. Analysis of relative accessi­
`bility versus relative conservation for individual amino acid
`positions in alignments shows a negative correlation for all amino
`acid types. With the exception of Arg, Lys, Gly, Glu, Asp and
`Tyr, a relative conservation of > 2 suggests the amino acid will
`have a relative accessibility of < 50%. Observation of conserved
`Cys, Gly or Asp in a reliable multiple alignment suggests a
`position important for the structure of the protein. Furthermore,
`the Asp is likely to be involved in polar interactions through its
`side chain oxygen atoms. In contrast, Gin is the least conserved
`amino acid overall.
`
`sequence Key words: Conservation analysis; Multiple
`
`
`
`
`
`
`alignment; Protein structure prediction
`
`2.1. Data base
`A non-redundant set of 81 proteins was generated from the April
`
`
`
`
`
`1993 release of the Brookhaven Protein Data Bank (PDB) [9]. The set
`
`
`
`
`
`was chosen in a two-step procedure. First, all pairs of chains (over 50
`
`
`
`residues and resolution better than 2.5 A) in the data bank were
`
`
`
`
`compared by calculating correlation coefficients between the dipeptide
`
`
`
`
`frequencies in each protein. A set of IOI protein chains was selected
`
`such that all pairs had a correlation of < 0.4. All pairs in this set were
`
`
`
`
`then compared by a rigorous sequence comparison method [10,11]
`
`
`
`
`
`followed by cluster analysis. This reduced the set to 81 protein chains
`
`
`
`that show no obvious sequence similarity (PDB code and chain iden­
`
`tifiers: 155C !ACX IALC IBBP_A !CC5 IECA IFKF IFNR IGCR
`IGPLA IHDSB !HIP !HOE ILRD_ 4 l PAZ IPCY !PHH IPRCC
`IRBP IRHD l RNH ISN3 l TGS ITPK_A IWSY_B 256B_A 2ALP
`2AZA_A 2CAB 2CD4 2CDV 2CPP 2FXB 2GN5 2LH7 2LIV
`2LTN_A 2ORl_L 2PAB_A 2RNT 2RSP_A 2SECI 2SNL_E 2SNS
`2SOD_B 2SSI 2STV 2TSI 2UTG_A 3ADK 3B5C 3CLA 3FXC
`3GAP _B 3LZM 3SGB_I 451C 4BP2 4FDI 4FXN 4HHB_A 4PEP
`4PFK 4PTP 4TNC 5CTS 5CYT_R 5EBX 5RUB_A 5RXN 6LDH
`6TMN_E 7PTI 8ADH 8ATC_B 8CAT_A 8DFR 9PAP 9RSA_A
`9WGA_A).
`Each protein in the set was compared by the Smith-Waterman
`
`
`
`
`
`
`algorithm [11,12] to the NBRF-PIR sequence data bank (Release
`of < 10-6 by a
`
`
`38) and all sequences that gave probability values
`
`
`
`length-dependent scoring scheme (program SCANPS ftp://geoff.bio­
`
`
`
`p.ox.ac.uk/programs/scanps) were multiply aligned with the query se­
`
`
`
`
`
`quence by the algorithm of Barton and Sternberg [13]. This gave 81
`
`
`
`
`alignments with between 3 and 499 sequences in each (median of 28
`Knowledge of protein sequences is growing much faster
`
`
`
`sequences).
`
`
`
`
`than knowledge of either three-dimensional structure or func­
`
`
`
`tion. Accordingly, the interpretation of sequence data to iden­
`2.2. Calculation of conservation and accessibility
`
`
`
`
`
`
`tify structurally or functionally important residues is essential
`
`
`
`
`Conservation scores based upon the physico-chemical properties of
`
`
`the amino acids were calculated for each position in each alignment
`
`
`
`if the data are to be effective in furthering understanding of
`
`
`
`
`according to Livingstone and Barton [2]. Conservation scores range
`
`
`biological systems. Multiple sequence alignments of families
`
`
`from O to 10 and represent the number of the properties: Hydropho­
`
`
`
`
`of protein sequences are now used routinely to indicate resi­
`
`
`
`
`
`
`bic, Positive, Negative, Polar, Charged, Small, Tiny, Aliphatic, Aro­
`
`
`
`dues of key importance to the function of the protein. A
`
`
`
`
`
`
`matic, Proline and their negations ( e.g. not positive) that are shared at
`
`
`
`
`a position. The program AMAS [2], which calculates conservation
`
`
`
`
`position in an alignment that has identical residues in all
`
`
`
`values from a multiple alignment, may be run over the World Wide
`
`
`members of a protein family may have a key catalytic role.
`Web (http://geoff.biop.ox.ac.uk/servers/amas-server.html).
`
`
`
`
`A position where similar physico-chemical properties (e.g. hy­
`
`
`
`
`
`Although conservation scores are absolute, the relative importance
`
`
`
`drophobicity) are shared may suggest importance in stabilis­
`
`
`
`
`of a conservation score is dependent on the overall similarity between
`
`
`
`
`the sequences in the multiple alignment. For example, in an alignment
`
`
`
`
`ing the native conformation of the protein [1,2]. Identification
`
`
`
`of 20 sequences that all share > 90% pairwise identity conservation
`
`
`
`
`of such conserved features in multiple alignments has been
`
`
`
`
`scores above 8 may be interesting. In contrast, if the pairwise identity
`
`
`
`used to good effect to improve the accuracy of prediction of
`
`
`is below 30% then lower conservation scores will be informative.
`
`
`secondary structure and buried residues (a-helix and �-strand)
`
`
`
`
`Accordingly, in this study we normalised conservation scores by the
`(e.g. [3-7]).
`
`
`
`
`average conservation for each alignment to give relative conservation
`if C > I.
`
`
`scores C. We refer to a position as 'conserved'
`
`
`Here we report a systematic study of residue conservation
`
`
`
`
`Accessible surface areas were calculated by the program DSSP [14]
`
`
`
`in multiple alignments where at least one protein is of known
`
`
`
`
`and converted to relative accessibilities by dividing by the accessibility
`
`
`
`
`
`of the residue in a Gly-X-Gly tripeptide [15]. Two relative accessibility
`
`
`
`classes were considered 0 ::s A ,s 0.25 (buried) and 0.25 < A ,s max(A)
`(exposed).
`
`1. Introduction
`
`*Corresponding author. Fax: (44) (1865) 510454.
`
`
`
`
`E-mail: gjb@bioch.ox.ac.uk
`
`
`
`0014-5793/96/$12.00 © 1996 Federation of European Biochemical Societies. All rights reserved.
`
`
`
`
`
`
`
`
`PIIS 0 0l4 - 5793(96 )01181-7
`
` 18733468, 1996, 2-3, Downloaded from https://febs onlinelibrary wiley com/doi/10 1016/S0014-5793(96)01181-7, Wiley Online Library on [15/11/2022] See the Terms and Conditions (https://onlinelibrary wiley com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
`
`Bausch Health Ireland Exhibit 2036, Page 1 of 5
`Mylan v. Bausch Health Ireland - IPR2022-00722
`
`

`

`226
`
`2.3. Exclusion of active site residues
`
`
`It was anticipated that residues involved in active sites will be more
`highly conserved than residues in the bulk of the protein and that this
`might bias any analysis. Accordingly, active site and binding residues
`were identified in the data set and the data were examined with and
`without these residues. From the 81 proteins, 21 have a site record in
`their PDB file. For the remainder, the original literature on the struc­
`tures was consulted. This added a further 35 proteins with sites. Un­
`fortunately, the description of active sites varies from author to
`author. A precise definition is often difficult because either the active
`site pocket does not make covalent connections with the substrate,
`e.g. bilin binding protein (16], or it does not take part in the enzymatic
`action, e.g. rhodanese (17]. We considered those residues active site
`residues that either are attached to a prosthetic group ( e.g. haem, Fe-S
`cluster) or take part in the enzymatic reaction, or if they were con­
`sidered crucial by the authors of the structures even if they make only
`second-order interactions with the substrate (van der Waals interac­
`tions or hydrogen bonds). Among the 81 proteins 56 have active site
`residues giving a total of 331 residues. The most frequent active site
`residues are Cys(52/341), His(39/327), Tyr(21/520), Trp(7/184),
`Met( l 1/294) and Asp(28/818). The most conserved are His (mean
`C,=1.6), Cys (1.5), Asp (1.4) and Gly (1.4).
`
`3. Results and discussion
`
`
`
`A. Fiser et al./FEBS Letters 397 (1996) 225-229
`
`
`
`Distribution of Amino Acids
`
`Surface(%)
`
`Fig. I. Distribution of amino acids in the data base analysed by rel­
`ative accessibility. Buried: percentage of amino acids < 25% ex­
`posed to solvent. Surface: percentage of amino acids 2:25% exposed
`to solvent. The line indicates where buried= exposed. The axes refer
`to percentages of the total number of amino acids in the sample.
`
`3.1. Distribution of amino acids
`
`Fig. 1 shows the distribution of residues in buried and ex­
`posed positions. There are no surprises in this distribution
`with the amino acids that are predominantly hydrophobic
`(W, M, F, I, V, L, A) seen to be more frequently buried
`than exposed, and polar amino acids (T, S, N, Q, R, D, E,
`K) seen to be more frequently exposed than buried. Glycine
`and histidine are seen equally exposed and buried while pro­
`line is predominantly on the surface, presumably due to its
`frequent location in turns [18]. Cys is the most highly con­
`served residue in this data set and the rarest on the surface
`probably because it has the most reactive side chain [19]. The
`distribution of half-cystines and cysteines among the buried
`and exposed residues is approximately equal (79% and 80%
`inside, respectively). The average relative conservation score
`between the two covalent forms of Cys is so wide, that even
`the standard deviations (o) are comparable with the differ­
`ence: 1.56 (o=0.53) and 1.13 (o=0.44) for cystines and cy­
`steines, respectively. The data set excluding the active site
`residues shows no appreciable differences (data not shown).
`
`show the data set with active site residues removed. As ex­
`pected, accessibility is negatively correlated with conservation.
`For example, hydrophobic residues which are often conserved
`are usually buried (M, V, I, L, F) while hydrophilic residues
`are less conserved and usually exposed (E, K, N). However,
`there are five interesting outliers. The three outliers trypto­
`phan, cysteine and glycine show high conservation for their
`mean accessibility values. The simple explanation for this is
`that tryptophan is nearly always buried and mutation of the
`large residue to any other amino acid is likely to disrupt the
`protein core. Similarly, Cys when participating in a disulphide
`bridge will not favour mutation to another residue as this
`would leave a single free sulphydryl group. The unique prop­
`erties of glycine, which can adopt q>/\Jf angles unfavoured by
`other residues, allowing tight packing of the polypeptide
`chain, lead to its conservation.
`The most surprising observations are the positions of Asp
`and Gin. Asp has a slightly smaller relative accessibility than
`Glu, but is significantly more conserved. Gin is significantly
`less conserved than Glu. The differing interactions and envi­
`3.2. Relationship between accessibility and conservation
`
`
`
`ronments of Asp and Glu are examined in more detail in the
`Fig. 2 illustrates average relative accessibility versus average
`following sections. Exclusion of the active site residues from
`relative conservation of each amino acid. Uppercase letters
`the data set has the greatest effect on His and Cys, which
`show data for all amino acids in the set, lowercase letters
`
`Table I
`Comparison of mean conservation for Asp and Glu in different secondary structures
`Asp (exposed)
`N C,
`x (a)
`0.92 (0.162)
`0.84 (0.163)
`0.95 (0.175)
`0.90 (0.163)
`
`ss
`
`Helix (HJ
`Strand (S)
`Coil (C)
`H+S
`
`Asp (buried)
`N C,
`.\' (a)
`1.11 (0.249)
`1.13 (0.293)
`1.25 (0.216)
`1.13 (0.277)
`
`58
`49
`121
`107
`
`199
`51
`340
`250
`
`Glu (buried)
`
`N Cr
`x (a)
`1.12 (0.224)
`0.96 (0.287)
`1.01 (0.186)
`1.05 (0.188)
`
`76
`44
`51
`120
`
`228
`
`1.19
`
`590
`
`0.93
`
`171
`
`1.04
`
`Total
`(H+S+C)
`N = number of residues in sample, x = mean relative conservation, a= standard deviation. Secondary structure was defined by DSSP (14], then
`reduced to 3 states as follows: helix (H) = a, 310 and re helix (DSSP H, G and I). Strand (S) =�sheet and bridge (DSSP E and B). Coil (CJ= DSSP
`bend (SJ and turn (T).
`
`Glu (exposed)
`
`N Cr
`x (a)
`0.80 (0.141)
`0.81 (0.156)
`0.81 (0.162)
`0.80 (0.144)
`
`303
`84
`275
`387
`
`662
`
`0.81
`
` 18733468, 1996, 2-3, Downloaded from https://febs onlinelibrary wiley com/doi/10 1016/S0014-5793(96)01181-7, Wiley Online Library on [15/11/2022] See the Terms and Conditions (https://onlinelibrary wiley com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
`
`Bausch Health Ireland Exhibit 2036, Page 2 of 5
`Mylan v. Bausch Health Ireland - IPR2022-00722
`
`

`

`A. Fiser et al./FEBS Letters 397 (1996) 225-229
`
`227
`
`
`
`
`
`Frequencies of Interacting Partners for Glu and Asp
`
`
`
`appear less conserved, but it does not affect the relative posi­
`
`
`tions of the amino acids.
`A Normalised
`
`
`
`
`
`Analysis of relative accessibility versus relative conservation
`
`
`
`for individual amino acid positions shows a negative correla­
`0 �
`
`tion for all amino acid types (data not shown). With the ex­
`
`
`ception of Arg, Lys, Gly, Glu, Asp and Tyr a relative con­
`
`
`servation of > 2 suggests the amino acid will have a relative
`
`accessibility of < 50% (data not shown).
`Asp
`
`
`
`might be unusually conserved due to backbone conformation
`
`
`
`
`
`preferences, secondary structure preferences, or specific side­
`
`chain interactions. To decide which is responsible, we first
`
`
`
`
`examined the proportion of conserved Asp in the ++ q>/\j/
`
`
`
`
`conformation. 20/411 (4.55%) conserved Asp residues are in
`
`
`
`the ++ conformation, while 40/818 (4.86%) of all Asp are in
`
`
`
`this conformation. These data do not suggest a preference to
`
`
`
`conserve Asp due to maintenance of unusual backbone infor­
`mation.
`
`
`
`The secondary structure distribution for Asp and Glu is
`
`
`summarised in Table I. The highest mean Cr is seen for buried
`
`
`
`Asp in coil (1.25). This is significantly higher than the mean
`
`
`of Cr for Asp in strand and helix (t-test gives probability
`
`
`95.5% for difference). In contrast, Glu shows no such prefer­
`
`
`
`ence for coil in either buried or exposed states. Thus, there
`
`
`
`
`appears to be a preference for buried Asp to be conserved in
`coil.
`
`
`
`Since our data do not suggest a significant preference for
`
`
`
`++ q>/\j/, the preferred conservation of Asp is likely to be due
`
`
`
`
`to differing side-chain interactions. The most obvious hypoth­
`
`
`esis is that since Glu has a higher proportion of non-polar
`
`atoms than Asp it can make more non-specific interactions
`
`
`and so there are fewer constraints on its environment. In
`
`
`
`order to test this idea, we examined the residue types that
`
`interact with Asp and Glu.
`
`
`
`Residue pairs were considered to be interacting if the dis­
`
`
`tance between any of their heavy atoms was smaller than the
`
`sum of the van der Waals radii plus I A. The occurrence of
`
`Asp and Glu in our set of proteins was almost equal (818 and
`greater relative accessibility of Glu when compared to Asp
`
`
`
`
`
`
`
`833, respectively). Despite its smaller size, Asp has the same
`(Fig. 2).
`
`
`number of interactions as Gluon average (10276/818= 12.56
`
`
`
`
`We calculate the normalised frequency of interaction for
`
`and 10314/833=12.38). This may be due to the observed
`Asp, P Asp, with each of the 20 amino acids as follows:
`
`0
`
`0
`
`I§
`0
`
`
`
`
`
`3.2.1. Why do Asp and Glu show different conservation?
`3
`a
`
`8
`0
`
`0
`
`0
`
`0.0
`
`0.02
`
`0 04
`
`0.06
`
`0.08
`
`0 10
`
`B
`
`AIIGlu
`
`G
`
`I
`
`�
`
`00
`
`0 02
`
`0.04
`
`0.06
`
`0 08
`
`0 10
`
`
`
`Conserved Glu
`
`Fig. 3. Normalised frequencies of interacting partners for Glu and
`Asp. A: All alignment positions. B: Only conserved positions.
`
`EB
`
`Kl<
`
`do
`
`.,
`<il
`
`0
`
`N;
`
`N
`0
`
`1
`
`"'
`
`H
`
`yy �
`
`v,
`Li
`Ii
`Fl
`
`w,,
`
`C
`
`08
`
`09
`
`1.0
`
`1.1
`
`1.2
`
`1.3
`
`
`
`
`
`Avera�e Relative Conservation
`
`Fig. 2. Average relative accessibility versus average relative conser­
`vation for amino acids including active site residues (uppercase let­
`ters) and without active site residues (lowercase letters). The outliers
`are Gin (Q, q), Asp (D, d), Gly (G, g), Trp (W, w) and Cys (C, c);
`see text.
`
`where N Asp, is the number of interactions between an Asp
`
`
`
`
`residue and. amino acid type A;. Similar frequencies were cal­
`
`culated for Glu.
`
`
`
`
`If we consider interactions between Asp/Glu and other res­
`
`
`idues at least 5 amino acids distant in the chain, then some
`
`
`interesting trends emerge (Fig. 3A,B). A cutoff of 5 amino
`
`
`
`acids was chosen to exclude local secondary structure interac­
`
`
`tions. Fig. 3A shows data for all Asp and Glu residues, while
`
`
`Fig. 3B shows data only for Asp/Glu that are conserved. If
`
`
`there was no difference in the interactions of Asp and Glu, all
`
`
`
`points would lie on the line in Fig. 3A,B. In Fig. 3A, most
`
`
`
`amino acids cluster close to the line indicating equivalent in­
`
`teractions for Asp and Glu, but Gly and Asn appear to favour
`
`
`Asp. The most common interacting residues are Lys and Leu,
`
`
`both with a slight preference for Glu.
`
`
`
`When only conserved Asp/Glu are considered as shown in
`
`
`
`Fig. 3B, greater scatter from the line is observed. Arg moves
`
` 18733468, 1996, 2-3, Downloaded from https://febs onlinelibrary wiley com/doi/10 1016/S0014-5793(96)01181-7, Wiley Online Library on [15/11/2022] See the Terms and Conditions (https://onlinelibrary wiley com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
`
`Bausch Health Ireland Exhibit 2036, Page 3 of 5
`Mylan v. Bausch Health Ireland - IPR2022-00722
`
`

`

`228
`
`
`
`A. Fiser et al./FEBS Letters 397 (1996) 225- 229
`
`from a position close to the line, to become equally favoured
`with Lys. Val and Phe move from being equally favoured to
`being preferred by Glu, while the preference of Asp for Gly is
`accentuated. These differences may be due to conserved Asp
`tending to occur in coil, where Gly is common. The longer
`aliphatic side chain of Glu can participate in more hydropho­
`bic interactions than Asp and so conserved Glu residues tend
`to interact more frequently with hydrophobic amino acids
`than do conserved Asp residues. Singh and Thornton [20]
`reported frequencies for interacting pairs .for all amino acid
`combinations. Their data show similar trends to ours for all
`data, but they did not gather statistics for conserved versus
`unconserved positions.
`
`3.3. Analysis of substitution matrices
`
`
`In this study, we consider conservation of amino acid resi­
`dues across complete families. This shows that Asp is signifi­
`cantly more conserved than Glu. However, one might expect
`that this trend would also be seen in substitution matrices
`derived from pairwise comparisons of aligned sequences. Ac­
`cordingly, we examined a number of commonly used substitu­
`tion matrices to see if a preference for Asp-Asp substitutions
`when compared to Glu-Glu was observed.
`We considered the more recently published matrices in the
`following articles: [8,21-34].
`In order to assess the relative mutability of Asp and Glu
`when compared to each other a cumulative index was calcu­
`lated for each mutation matrix as follows:
`
`where Asp(A;) and Glu(A;) are the mutation scores between
`Asp and Glu, and all 20 amino acids. This cumulative index
`results a positive score if the overall mutability of Asp is
`greater than that of Glu, zero if they mutate equally and
`negative if Glu mutates more frequently. In Table 2 we list
`the analysed mutational matrices with the calculated Ix. The
`
`Table 2
`Amino acid pair substitution matrices examined for preference to
`conserve Asp over Glu
`
`Mutation matrix
`
`Risler et al. [3 1]
`Henikoff et al. (24]
`Pongor [28]
`Gonnet et al. [22]
`Miyata et al. [27]
`Johnson et al. (29]
`Henikoff et al. [23]
`Tusnady et al. [34]
`Rao [30]
`Altschul [26]
`Dayhoff et al. [21 ]
`Overington e t al. [8]
`Tudos et al. [33]
`Levin et al. [32]
`Tusnady et al. [34]
`Jones et al. [25]
`
`Ix (Asp, Glu)
`
`-5.21
`-2.77
`-2.66
`-2.30
`-2. 1 2
`- 1 .76
`- 1 .7 1
`- 1 .44
`-0.84
`-0.42
`-0.29
`0.00
`0.55
`0.67
`0.83
`0.95
`
`Henikoff et al. [24] refers to the BLOSUM62 matrix. Dayhoff et al.
`[2 1] refers to the PAM250 matrix, Altschul [26] refers to PAM l 20,
`Henikoff et al. [23] refers to a matrix derived from the structural
`alignments of Overington et al. [8].
`
`matrices were converted into all positive values b efore calcu­
`lating the index according to Johnson and Overington [29].
`Although the maj ority of the matrices show negative values
`of Ix there is no consistent explanation for the values. For
`example, the Risler et al. matrix [31] is derived from structural
`= -5.2 1 while the BLOSUM62 matrix
`alignments and shows Ix
`[24] Ux = -2. 77) is derived purely from sequence alignment. It
`is difficult to make direct comparisons between all the ma­
`trices shown in Table 2 since they are calculated for align­
`ments of differing similarity. For example, BLOSUM62 repre­
`sents sequences at a shorter evolutionary distance than
`PAM250 or PET92 [25]. The families we have analysed in
`the present study only include sequences that are readily align­
`able by sequence comparison methods. Accordingly, our re­
`sults are more likely to be consistent with a matrix such as
`BLOSUM62 than one at a greater evolutionary distance, e.g.
`PET92.
`
`4. Conclusions
`
`In this study we have analysed multiple alignments for 81
`non-homologous protein families each of which has at least
`one member of known three-dimensional structure. We have
`examined the relationship between the conservation of physi­
`co-chemical properties at a position and the relative accessi­
`bility. The principal new observations are that Asp is more
`highly conserved for its accessibility than Glu (Fig. 2), and
`that conserved Asp is most often found in coil (Table I ). The
`differences in interacting partners for Asp and Glu show Glu
`to favour non-polar partners more than Asp (Fig. 3). This
`may be explained simply by the higher proportion of non­
`polar atoms in the Glu side chain. Although carboxylate-ami­
`no interactions in proteins have been studied in some detail
`[35,20, 36], these studies did not discriminate between con­
`served and variable positions and so do not help explain
`our current observations.
`Why, then, is Asp most highly conserved when buried in
`coil? The short Asp side chain is restricted in mobility yet able
`to make strong polar interactions. It is possible that Asp may
`form a ' pin' that stabilises non-regular structures in loops.
`Further work will be required to dissect the precise role of
`conserved Asp in specific coil structures.
`This study has elucidated the structural reasons for the
`greater conservation of Asp over Glu, but it is intriguing to
`speculate why this situation may have arisen during evolution.
`Indeed, why is Gin found in this study to be the least con­
`served of all amino acids? The differences we see here may be
`due to the relative !ability of Asp/Asn and Glu/Gln. Asp may
`cyclise into a succinimidyl ring, then hydrolyse back to Asp in
`both o and L isomers, causing the death of the protein. For
`Asn the half-life is 1 .4 days for cyclisation, for Asp 53 days,
`but Gin will only cyclise at the N-terminus [ 1 9]. Thus, Asp
`and Asn residues could be regarded as time bombs in proteins
`whereas Gin is a useful and safe 'filler'. These chemical pres­
`sures may contribute to the observed greater conservation of
`Asp over Glu and Gin.
`
`Acknowledgements: We thank Professor L.N. Johnson for providing a
`stimulating working environment, Drs. R.B. Russell and C.D. Living­
`stone for advice and guidance. This work was carried out during 1993
`when A.F. was supported by a 9 month Soros Fellowship. G.J.B.
`thanks the British Council and Royal Society for support.
`
` 18733468, 1996, 2-3, Downloaded from https://febs onlinelibrary wiley com/doi/10 1016/S0014-5793(96)01181-7, Wiley Online Library on [15/11/2022] See the Terms and Conditions (https://onlinelibrary wiley com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
`
`Bausch Health Ireland Exhibit 2036, Page 4 of 5
`Mylan v. Bausch Health Ireland - IPR2022-00722
`
`

`

`A. Fiser et al.lFEBS Letters 397 (1996) 225-229
`
`229
`
`References
`
`[19] Creighton, T.E. ( 1 993) Proteins: Structures and Molecular Prop­
`
`[5] Benner, S.A. and Gerloff, D. (1990) Adv. Enzyme Reg. 3 1 , 1 2 1-
`
`
`
`
`
`
`erties, 2nd Edn. W.H. Freeman,
`New York.
`.
`
`
`[20] Singh, J. and Thornton, J.M. ( 1990) J. Mo!. B10l. 211 , 595-615.
`[I) Casari, G., Sander, C. and Valencia, A. ( 1 995) Nature Struct.
`
`
`
`
`
`
`[21] Dayhoff, M.O., Schwartz, R.M. and Orcutt, B.C. (1978) m:
`Biol. 2, 1 71-178.
`. .
`
`
`
`Atlas of Protein Sequence and Structure (Dayhoff, M.O., Ed.),
`[2] Livingstone, C.D. and Barton, G.J. ( 1 993) Comp. Appl. B1osc1.
`
`
`
`
`
`
`Vol. 5, pp. 345-358. National Biomedical Research Foundat10n,
`9, 745-756.
`
`Washington, DC.
`[3) Barton, G.J. ( 1 995) Curr. Opin. Struct. Biol. 5, 372-376.
`
`
`
`
`
`
`
`
`[4] Livingstone, C.D. and Barton, G.J. (1996) Methods Enzymol.
`256, 1443-1444.
`266, 497-51 2.
`[23] Henikoff, S. and Henikoff, J.G. (1993) Proteins 17, 49---01 .
`
`
`
`
`
`
`[24] Henikoff, S. and Henikoff, J.G. (1992) Proc. Natl. Acad. Sc1.
`1 8 1.
`USA 89, 1 09 1 5-109 1 9.
`[6] Rost, B. and Sander, C. (1993) J. Mo!. Biol. 232, 584-599.
`
`
`
`
`
`
`
`
`[7] Russell, R.B. and Sternberg, M.J.E. (1995) Curr. Biol. 5, 488-
`
`Appl. Biosci. 8, 275-282.
`490.
`
`[26) Altschul, S.F. (199 1 ) J. Mo!. Biol. 2 19, 555-565.
`M.S., Sali, A. and Blun­
`
`[8] Overington, J., Donnelly, D., Johnson,
`
`
`
`
`
`[27] Miyata, T., Miyazawa, S. and Yasunaga, T. (1979) J. Mo!. Evol.
`
`
`dell, T.L. ( 1 992) Protein Sci. 1 , 2 16-226.
`12, 2 1 9 -236.
`G.J.B., Meyer !r.,
`
`
`[9] Bernstein, F.C., Koetzle, T.F., WilJiams,
`E.F.,
`[28) Pongor, S. (1987) Methods Enzymol. 1 54, 450-473.
`
`
`
`
`
`
`
`Brice, M.D., Rodgers, J.R., Kennard, 0., Sh1manouch1, T. and
`
`
`[29] Johnson, M.S. and Overington, J.P. (I 993) J. Mo!. Biol. 233,
`
`Tasumi, M. (1977) J. Mo!. Biol. I 1 2, 535-542.
`7 1 6-738.
`
`
`[10] Barton, G.J. ( 1 993) Protein Eng. 6, 37-40.
`.
`[30] Rao, J.K.M. ( 1987) Int. J. Peptide Protein Res. 29, 276-281.
`
`
`
`
`T.F. and Waterman, M.S. (1981) J. Mo!. B10J. 147, 195-
`[11] Smith,
`
`
`
`
`
`[31] Risler, J.L., Delorme, M.O., Delacroix, H. and Henaut, A. (1988)
`1 97.
`
`J. Mo!. Biol. 204, 1019-1029.
`[12) Barton, G.J. (1993) Comp. Appl. Biosci. 9, 729-734.
`
`
`
`
`
`
`
`
`[ 1 3) Barton, G.J. and Sternberg, M.J.E. ( 1987) J. Mo!. Biol. 198, 327-
`303-308.
`337.
`
`
`
`[14) Kabsch, W. and Sander, C. ( 1983) Biopolymers 22, 2577-2637.
`
`
`
`tein Res. 36, 236-239.
`
`
`
`[15) Rose, G., Geselowitz, A., Lesser, G., Lee, R. and Zehfus, M .
`
`
`
`
`( 1985) Science 229, 834-838.
`417-423.
`[16] Huber, R., Schneider, M., Mayr, I., Mueller, R., Deutzmann, R.,
`
`
`
`
`
`
`
`
`
`
`
`Suter, F., Zuber, H., Falk, H. and Kayser, H. ( 1 987) J. Mo!.
`FEBS Lett. 224, 161-1 7 1 .
`Biol. 1 98, 499-51 3.
`
`
`
`
`
`
`
`
`[17) Ploegman, J.H., Drent, G., Kalk, K.H. and Ho!, W.G.J. (1979)
`J. Mo!. Biol. 226, 251-262.
`J. Mo!. Biol. 1 27, 149-162.
`
`
`[ 1 8] Wilmot, A.C.M. and Thornton, J.M. (1988) J. Mo!. Biol. 203,
`221-232.
`
`[22] Gonnet, G.H., Cohen, M.A. and Benner, S.A. ( I 992) Science
`
`[25] Jones, D.T., Taylor, W.R. and Thornton, J.M. ( 1 992) Comp.
`
`[32) Levin, J.M., Robson, B. and Garnier, J. (1986) FEBS Lett. 205,
`
`
`
`[33] Tudos, E., Cserzo, M. and Simon, I. (1990) Int. J. Peptide Pro­
`
`
`
`[34) Tusnady, G.E., Tusnady, G. and Simon, I. (1995) Protein Eng. 8,
`
`[35] Singh, J., Thornton, J.M., Snarey, M. and Campbell, S.F. (1990)
`
`
`
`[36] Mitchell, J.B.O., Thornton, J.M., Singh, J . and Price, S.L. ( 1 992)
`
` 18733468, 1996, 2-3, Downloaded from https://febs onlinelibrary wiley com/doi/10 1016/S0014-5793(96)01181-7, Wiley Online Library on [15/11/2022] See the Terms and Conditions (https://onlinelibrary wiley com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
`
`Bausch Health Ireland Exhibit 2036, Page 5 of 5
`Mylan v. Bausch Health Ireland - IPR2022-00722
`
`