KINARI-Mutagen Case Study: 2LZM

Lysozyme from bacteriophage T4

Filip Jagodzinski

25 April 2012

We evaluated whether rigidity analysis can identify destabilizing mutations. From the literature[1,2, 3, 4] we retrieved stability data for 163 different point mutations in lysozyme from bacteriophageT4 (PDB ID 2lzm for Wild-type, Figure 1). The experimentally derived value ∆∆G, the free en-ergy of unfolding, measures the stability of a variant against a reference protein (nearly always thewild-type protein). The lower the ∆∆G value, the more unstable is the variant. From the available∆∆G dataset, we selected the 8 mutations that involved a substitution to a glycine. We compared∆∆G values of these mutations that had been performed in the physical protein to the rigiditycalculation predictions of KINARI-Mutagen.

Figure 1: Rigidity analysis performed on Lysozyme frombacteriophage T4 (PDB ID 2zlm). The color bodies indicateclusters of atoms that are rigid, as identified by an efficientpebble game algorithm.

Table 1 lists for each lysozymemutant several rigidity measures,that we evaluated as predictors ofprotein stability. For the amino acidwhich was mutated at a particularsequence location, we list its Sol-vent Accessible Surface Area, the vol-ume of the wild-type amino acid, thechange of the volume of the residuewhen mutated to glycine, as wellas the stability data from the lit-erature (∆∆G), which we considerthe “ground truth” stability measure-ment. The loss in number of hydro-gen bonds and hydrophobic interac-tions that were caused by the muta-tion to glycine are listed, as well asthe change of the largest rigid bodyrelative to the wild-type.

When KINARI-Mutagen was used to mutate residues 96, 105, 157, and 124, the size of thelargest rigid body decreased in size in parallel to a decrease in the ∆∆G value. For example,residue 96, an arginine, when mutated to a glycine, caused the largest rigid body of the mutant todecrease by 130 atoms relative to the largest rigid body in the wild-type protein. When residue 96was mutated to a glycine in the physical lysozyme, the stability of the protein decreased significantly,as indicated by the low ∆∆G value. Similarly, for residues 105, 157, and 124, the size of the largestrigid body decreased in size relative to the wild-type protein. The ∆∆G values for mutations atresidues 105, 157, and 124, indicate that their destabilizing effect is not as great as when a mutationis performed on residue 96. In these cases KINARI-Mutagen was able to predict a change in the

1

Table 1: Rigidity data of 8 lysozyme mutants. For each mutant, the experimental ∆∆G value,the change in volume of the amino acid when mutated to glycine, the loss of hydrogen bondsand hydrophobic interactions, the change of the largest rigid body, and the Cluster ConfigurationEntropy (CCE) values are listed. For the wild-type of the protein, the CCE value is 0.43, and thelargest rigid body contains 830 atoms. Mutants are ordered by ∆∆G values, and rows shaded grayindicate amino acid mutations that KINARI-Mutagen correctly identified as destabilizing.

Seq

uen

ceN

um

ber

WT

Am

ino

Aci

d,

Volu

me

(A3)

SA

SA

(A2)

of

WT

Am

ino

Aci

d

AA

Volu

me

Ch

ange

for

mu

ta-

tion

toG

LY

∆∆

Gfr

omli

tera

ture

Hyd

roge

nB

ond

sL

ost

inm

uta

nt

Hyd

rop

hob

ics

Lost

inm

uta

nt

Ch

ange

toL

arg

est

Rig

idB

od

yin

mu

-ta

nt

CC

E[5

]

99 LEU, 166 0 -106 -6.3 0 0 0 0.4396 ARG, 173 73.04 -113 -2.6 2 0 130 0.613 ILE, 166 43.72 -106 -2.1 0 1 0 0.4359 THR, 116 128.75 -56 -1.6 1 0 0 0.39105 GLN, 143 64.8 -83 -1.5 2 0 11 0.44157 THR, 116 100.97 -56 -1.1 2 1 63 0.4955 ASN, 114 113.45 -54 -0.6 0 0 0 0.43124 LYS, 168 103.19 -108 -0.1 1 2 15 0.44

protein’s stability.KINARI-Mutagen was not able in all instances to predict the effect of a mutation on the

protein’s stability. For residues 99, 3, 59, and 55, the lack of loss of hydrogen bonds and/orhydrophobic interactions when these residues were mutated to a glycine explains why KINARI-Mutagen’s could not be used as a discerning measure of protein stability. For these mutations, theloss of hydrogen bonds and hydrophobic interactions were not as great as when mutations wereperformed on residues 96, 105, 157, and 124. The predictive ability of KINARI-Mutagen relies onthe change in the molecular model due to a loss of these interactions. For these mutation instances,the change in the protein’s stability is caused by phenomena that KINARI-Mutagen does notcurrently capture. We suspect that the change in volume of the wild-type amino acid to a glycinecauses a large-enough collapse or reorientation of the protein’s structure in the vicinity of thesubstitution, which affects the protein’s stability.

Lastly, we compared KINARI-Mutagen’s predictions to the Cluster Configuration Entropy(CCE) measurement[5]. CCE is a function of the probability that a vertex in the mechanicalmodel is part of a cluster of size s. To compute CCE, a normalized cluster number, ns is de-fined as the number of clusters of size s divided by the total number of vertices in the mechanicalmodel. The probability that a vertex belongs to an s-cluster, ws, and the CCE value of the entiremechanical model are given as the following:

ws =sns∑s sns

CCE = −∑s

ws ln ws (1)

For two conformations of a protein, the one with the higher CCE value is more disordered, andhence is more unstable. The CCE values for several variants correlated well with the largest rigidbody metrics for those mutants. For mutants that had residue substitutions at amino acids 105,

2

124, 157, and 96, the CCE values were 0.43, 0.44, 0.49, and 0.61, respectively, while the change inthe size of the largest rigid body for those variants were 11, 15, 63, and 130.

Figure 2: Distribution of Rigid Bodies, By Residue :The left axis lists mutants that were analyzed. The verticalcolor legend on the right-hand side assigns colors to the rigidbody sizes found among the mutants. The color at each x -yposition in the plot indicates the size of the largest clusterthat residue x belongs to for the mutant in row y.

To investigate why we did not pre-dict the T59G mutant to be desta-bilizing, we referred to the Distri-bution of Rigid Bodies, By Residue(DRBR) plot (Figure 2). It was usedto distinguish between mutations thathave only a local effect on the rigid-ity of a protein and mutations thatdrastically affect a protein’s stabil-ity. The row 2lzm.A.0059 indicatesthat a mutant was generated by excis-ing residue 59 of chain A of protein2lzm. For the residue 59 mutation,the change of the protein’s rigidity isnot localized to the largest rigid clus-ter, which explains why using the sizeof the largest rigid cluster was not agood predictor of protein stability.

References

[1] T. Alber, S. Dao-pin, J.A. Wozniak, S.P., and Cook B.W. Matthews. Contributions of hydrogenbonds of Thr 157 to the thermodynamic stability of phage T4 lysozyme. Nature, 330:41–46,1987.

[2] J.A. Bell, W.J. Becktel, U. Sauer, W.A. Baase, and B.W. Matthews. Dissection of helix cappingin T4 lysozyme by structural and thermodynamic analysis of six amino acid substitutions atThr 59. Biochemistry, 31(14):3590–3596, 1992.

[3] B. Mooers, W.A. Baase, J.W. Wray, and B.W. Matthews. Contributions of all 20 amino acidsat site 96 to the stability and structure of T4 lysozyme. Protein Science, 18(5):871–880, 2009.

[4] H. Nicholson, E. Soderlind, D.E. Tronrud, and B.W. Matthews. Contributions of left-handed he-lical residues to the structure and stability of bacteriophage T4 lysozyme. Journal of MolecularBiology, 210(1):181–193, 1989.

[5] H. Gohlke and S. Radestock. Exploiting the link between protein rigidity and thermostabilityfor data-driven protein engineering. Engineering Life Science, 8:507–522, 2008.

3