Numerical Evaluation of Protein Global Vibrations at Terahertz - - PowerPoint PPT Presentation

numerical evaluation of protein global vibrations at
SMART_READER_LITE
LIVE PREVIEW

Numerical Evaluation of Protein Global Vibrations at Terahertz - - PowerPoint PPT Presentation

1st International Electronic Conference on Applied Sciences 1030 November 2020 Numerical Evaluation of Protein Global Vibrations at Terahertz Frequencies by means of Elastic Lattice Models D. Scaramozzino, G. Lacidogna, G. Piana, A.


slide-1
SLIDE 1

Numerical Evaluation of Protein Global Vibrations at Terahertz Frequencies by means of Elastic Lattice Models

1

  • D. Scaramozzino, G. Lacidogna, G. Piana, A. Carpinteri

Department of Structural, Geotechnical and Building Engineering Politecnico di Torino, Italy 1st International Electronic Conference on Applied Sciences

10–30 November 2020

slide-2
SLIDE 2

Outline of the Work

2

  • 1. Introduction
  • 2. Elastic Lattice Models (ELMs) for Protein Vibrations
  • 3. Validation of the Numerical Models: B-factors
  • 4. Protein Normal Modes and Biological Mechanism
  • 5. Conclusions and Future Developments

1st International Electronic Conference on Applied Sciences

10–30 November 2020

1/22

slide-3
SLIDE 3

3

Protein: Sequence of several different amino acids, with a complex three-dimensional shape and function

1st International Electronic Conference on Applied Sciences

10–30 November 2020

Three structural levels Primary structure Secondary structures Tertiary structure Protein folding

  • 1. Introduction

2/22

slide-4
SLIDE 4

4

1st International Electronic Conference on Applied Sciences

10–30 November 2020

Sequence Structure Dynamics Function

  • 1. Introduction

3/22

The fundamental paradigm of protein action

slide-5
SLIDE 5

5

1st International Electronic Conference on Applied Sciences

10–30 November 2020

How to study the Structure – Dynamics relationship?

Type of analysis Protein representation Form of potentials Type of

  • utput

Molecular Dynamics (MD) All atoms Complex semi- empirical Trajectories Normal mode analysis (NMA) All atoms Multi-parameter harmonic Normal modes All-atom Elastic Lattice Model (aaELM) All atoms Single- parameter harmonic Normal modes Coarse-grained Elastic Lattice Model (cgELM) Only one node per amino acid Single- parameter harmonic Normal modes

Increasing computational efficiency Increasing model complexity

  • 1. Introduction

4/22

slide-6
SLIDE 6

6

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 2. Elastic Lattice Models (ELMs) for Protein Vibrations

5/22

Elastic Lattice Model (ELM) From the single bar element… … to the spatial ELM i (xi, yi, zi) Ei,j, Ai,j, Li,j j (xj, yj, zj) 𝑙",$ = 𝐹",$𝐵",$ 𝑀",$

slide-7
SLIDE 7

7

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 2. Elastic Lattice Models (ELMs) for Protein Vibrations

6/22

Elastic Lattice Model (ELM) – Finite Element (FE) approach

𝐥𝐣,𝐤

∗ = 𝐹",$𝐵",$

𝑀",$ 1 −1 −1 1 𝐎𝐣,𝐤 = 𝑦$ − 𝑦" 𝑀",$ 𝑧$ − 𝑧" 𝑀",$ 𝑨

$ − 𝑨"

𝑀",$ 𝑦$ − 𝑦" 𝑀",$ 𝑧$ − 𝑧" 𝑀",$ 𝑨

$ − 𝑨"

𝑀",$ 2x2 stiffness matrix of the elastic bar in the local system 2x6 rotation matrix of the elastic bar, between the local and global systems 𝐥𝐣,𝐤 = 𝐎𝐣,𝐤

𝑼𝐥𝐣,𝐤 ∗𝐎𝐣,𝐤

6x6 stiffness matrix of the elastic bar in the global system

slide-8
SLIDE 8

8

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 2. Elastic Lattice Models (ELMs) for Protein Vibrations

7/22

Elastic Lattice Model (ELM) – Finite Element (FE) approach

𝐥𝐣,𝐤 = 𝐎𝐣,𝐤

𝑼𝐥𝐣,𝐤 ∗𝐎𝐣,𝐤 =

𝛃𝐣,𝐤 −𝛃𝐣,𝐤 −𝛃𝐣,𝐤 𝛃𝐣,𝐤 𝛃𝐣,𝐤 = 𝐹",$𝐵",$ 𝑀",$ 𝑦$ − 𝑦"

6

𝑀",$

6

𝑦$ − 𝑦" 𝑧$ − 𝑧" 𝑀",$

6

𝑦$ − 𝑦" 𝑨

$ − 𝑨"

𝑀",$

6

𝑦$ − 𝑦" 𝑧$ − 𝑧" 𝑀",$

6

𝑧$ − 𝑧"

6

𝑀",$

6

𝑧$ − 𝑧" 𝑨

$ − 𝑨"

𝑀",$

6

𝑦$ − 𝑦" 𝑨

$ − 𝑨"

𝑀",$

6

𝑧$ − 𝑧" 𝑨

$ − 𝑨"

𝑀",$

6

𝑨

$ − 𝑨" 6

𝑀",$

6

6x6 stiffness matrix of the elastic bar in the global system

slide-9
SLIDE 9

9

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 2. Elastic Lattice Models (ELMs) for Protein Vibrations

8/22

Elastic Lattice Model (ELM) – Finite Element (FE) approach

𝐋 = 8

𝒋,𝒌|𝑴𝒋,𝒌=𝒔𝒅

𝐃𝐣,𝐤

𝑼𝐎𝐣,𝐤 𝑼𝐥𝐣,𝐤 ∗𝐎𝐣,𝐤𝐃𝐣,𝐤

3Nx3N stiffness matrix of the ELM 𝐃𝐣,𝐤 6x3N expansion matrix of the elastic bar to reach the dimension of the structural problem 3Nx3N mass matrix of the ELM 𝐍 = 𝐍𝟐 𝟏 … 𝟏 𝟏 𝐍𝟑 … 𝟏 … … … 𝟏 𝟏 𝟏 𝟏 𝐍𝐎 𝐍𝐣 = 𝑛" 𝑛" 𝑛" 3x3 mass matrix of the ith node

slide-10
SLIDE 10

10

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 2. Elastic Lattice Models (ELMs) for Protein Vibrations

9/22

Elastic Lattice Model (ELM) – Anisotropic Network Model (ANM)

3Nx3N Hessian matrix of the ANM It can be easily demonstrated that there exists complete consistency between the FE- based ELM stiffness matrix K and the ANM Hessian matrix H 𝐈 = 𝐈𝟐,𝟐 𝐈𝟐,𝟑 … 𝐈𝟐,𝐎 𝐈𝟑,𝟐 𝐈𝟑,𝟑 … 𝐈𝟑,𝐎 … … … … 𝐈𝐎,𝟐 𝐈𝐎,𝟑 … 𝐈𝐎,𝐎 𝐈𝐣,𝐤 = 𝜖6𝑊

",$

𝜖𝑦"𝜖𝑦$ 𝜖6𝑊

",$

𝜖𝑦"𝜖𝑧$ 𝜖6𝑊

",$

𝜖𝑦"𝜖𝑨

$

𝜖6𝑊

",$

𝜖𝑧"𝜖𝑦$ 𝜖6𝑊

",$

𝜖𝑧"𝜖𝑧$ 𝜖6𝑊

",$

𝜖𝑧"𝜖𝑨

$

𝜖6𝑊

",$

𝜖𝑨"𝜖𝑦$ 𝜖6𝑊

",$

𝜖𝑨"𝜖𝑧$ 𝜖6𝑊

",$

𝜖𝑨"𝜖𝑨

$

𝐈𝐣,𝐣 = − 8

𝒌J𝟐,𝒌K𝒋 𝑶

𝐈𝐣,𝐤 𝑊

",$ = 𝛿

2 𝑠

",$ − 𝑠 ",$P 6

𝛿 ∝ 1 𝑠

",$R

slide-11
SLIDE 11

11

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 2. Elastic Lattice Models (ELMs) for Protein Vibrations

10/22

Elastic Lattice Model (ELM) – Modal Analysis

𝐋𝐯 + 𝐍 ̈ 𝐯 = 𝟏 𝐯 = 𝛆 sin 𝜕𝑢 𝐋 − 𝜕6𝐍 𝛆 = 𝟏 |𝐋 − 𝜕6𝐍| = 𝟏 Non-trivial solution Set of 3N-6 non-zero eigenvalues 𝜕\ = 2𝜌𝑔

\

Set of 3N-6 non-rigid eigenvectors (mode shapes) 𝛆𝐨

slide-12
SLIDE 12

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 2. Elastic Lattice Models (ELMs) for Protein Vibrations

11/22

Effect of the selected cutoff value on the generated ELM

rc = 8Å rc = 10Å rc = 12Å rc = 15Å rc = 20Å Lysozyme (PDB: 4YM8) – LUSAS FE software used for the construction of the model

*The Cα atoms of the protein are the representative points for each amino acid!

slide-13
SLIDE 13

13

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 3. Validation of the Numerical Models: B-factors

12/22

How to set up the values of the axial rigidity EA? With the B-factors!

𝐶" = 8 3 𝜌6𝑙c𝑈 8

\Je fg 𝜀",\ 6

𝜕\6 B-factors are a measure of the protein flexibility and can be found in the PDB file, as obtained from the X-ray crystallographic experiment B-factors can also be associated to the normal modes

20 40 60 80 100 120

Residue (amino acid)

25 30 35 40 45 50 55 60 65

Experimental B-factor [Å2]

Average value Lysozyme

slide-14
SLIDE 14

14

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 3. Validation of the Numerical Models: B-factors

13/22

How to set up the values of the axial rigidity EA? With the B-factors!

Imposing that the average value of the computed B-factors matches the average value of the experimental ones allows to define the rigidity of the ELM elastic bars Model Cutoff (Å) Mean length of the elastic bar (Å) EA (pN) Stiffness of the mean connection (N/m)

A 8 5.71 831 1.455 B 10 7.21 235 0.326 C 12 8.61 124 0.144 D 15 10.59 71 0.067 E 20 13.46 45 0.033

slide-15
SLIDE 15

15

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 3. Validation of the Numerical Models: B-factors

14/22

How to validate the models? With the B-factors!

Model Correlation

A 0.57 B 0.67 C 0.66 D 0.69 E 0.72

Correlation coefficients from 57% to 72%! These are very high values if you think how much the model is simplified and how much the physics of the problem is complex!

slide-16
SLIDE 16

16

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 4. Protein Normal Modes and Biological Mechanism

15/22

Looking at the 1st vibration modes…

… we find a clear representation

  • f

the cleft

  • pening-closing

motion, which is known to be the actual biological mechanism

  • f the lysozyme
slide-17
SLIDE 17

17

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 4. Protein Normal Modes and Biological Mechanism

16/22

Looking at the 2nd vibration modes…

… we find an overall torsional twisting

  • f

the lysozyme, still with a significant flexibility in the cleft region

slide-18
SLIDE 18

18

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 4. Protein Normal Modes and Biological Mechanism

17/22

Does the cutoff parameter affect the mode shapes?

MAC matrix 1st vibration mode Absolute displacements

slide-19
SLIDE 19

19

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 4. Protein Normal Modes and Biological Mechanism

18/22

Does the cutoff parameter affect the mode shapes?

MAC matrix 2nd vibration mode Absolute displacements

slide-20
SLIDE 20

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 4. Protein Normal Modes and Biological Mechanism

19/22

What about the vibrational frequencies? … we are in the (sub-)THz frequency range!

50 100 150 200 250 300 350

Mode number

0.2 0.4 0.6 0.8 1 1.2 1.4

Frequency [THz]

Model A Model B Model C Model D Model E

slide-21
SLIDE 21

21

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 4. Conclusions and Future Developments

20/22

Conclusions

  • We have shown that simplified mechanical models, such as ELMs, can

be efficiently used to extract the vibrational states of proteins;

  • The computed B-factors from the normal modes have a good

correlation with the experimental values, although the cutoff parameter has a certain influence;

  • The resulting mode shapes are well correlated with the biological

mechanism of the protein;

  • The corresponding vibrational frequencies lie in the (sub-)THz

frequency range;

  • Might

resonances at these frequencies play a role in the conformational changes and biological processes?

slide-22
SLIDE 22

22

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 4. Conclusions and Future Developments

21/22

Future Developments

What happens if we also apply (dynamic) forces to the protein ELM?

𝐋𝐯 + 𝐍 ̈ 𝐯 = 𝐆 𝑡𝑗𝑜 𝜕m𝑢 𝐯(𝑢) = 8

\Je fg

𝛆𝐨𝑞\(𝑢) 𝑞\(𝑢) = 𝛆𝐨

𝑼𝐆

𝜕\6 −𝜕m6 sin 𝜕m𝑢 − 𝜕m 𝜕\ sin 𝜕\𝑢 𝜕m6 → 𝜕\6

Resonance according to the nth mode MDOF forced modal analysis

slide-23
SLIDE 23

23

1st International Electronic Conference on Applied Sciences

10–30 November 2020

  • 4. Conclusions and Future Developments

22/22

Future Developments

Toy model with random force field applied at various frequencies

2 4 6 8 10 12 14 16 18

Mode number

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45

Frequency [Hz]

𝑔

m = 0.4 Hz

𝑔

m = 0.2 Hz

𝑔

m = 0.13 Hz

𝑔

m = 0.1 Hz

Dmax = 5.9 Dmax = 23.2 Close to 2nd resonance Dmax = 2.4 Dmax = 56.8 Close to 1st resonance

slide-24
SLIDE 24

24

1st International Electronic Conference on Applied Sciences

10–30 November 2020

Thank you for your attention!

  • D. Scaramozzino, G. Lacidogna, G. Piana, A. Carpinteri (2019) A finite-element-based

coarse-grained model for global protein vibration. Meccanica. 54, 1927-1940.