Difference between revisions of "Team:Fudan-CHINA/Results PR"

Line 59: Line 59:
 
<div class="fiName">Receptor Optimisation</div>
 
<div class="fiName">Receptor Optimisation</div>
 
<div class="pSentence">
 
<div class="pSentence">
"The essence of mathematics lies in its freedom."
+
"No road of flowers lead to glory."
 
</div>
 
</div>
 
</div>
 
</div>

Revision as of 00:57, 18 October 2018

Receptor Optimisation
"No road of flowers lead to glory."
Summary
In our dry lab, to improve the binding affinity, we have redesigned 3 receptor binding domains:

1.VEGF-scFv
2.KDR Ig domain 3
3.D-Dimer scFv
Also, we have redesigned the DNA sequence of the tTA promoter, to enhance downstream expression.

Figure 0. Binding energy variation after redesigning


Methods
Computational macromolecule design
Figure 1. Computational macromolecule design

Aiming at enhancing the binding affinity of the signal molecule and the receptor by sequence redesigning, we should firstly make it clear:

WHAT DO WE HAVE?

To simulate the macromolecule interaction, the structure information is necessary. At best, the crystal structure of the signal-receptor complex is already tested by experiment. That means the interface of those two molecules is known, the interaction of the two molecules can be calculated, and further protein designing will be carried out. If there is only individual structures of the signal molecule and the receptor, we can use RosettaDock to simulate the binding process of the two molecules. If there are only amino acid sequences of those two molecules, we can still predict the structure by large-scale computations. If there is a homologous structure (for example, antibodies), this operation would be easier. Otherwise, modeling only with sequence is still a tough work.
Next step is to design the sequence with the complex structure. If the receptor is an antibody, or have Immunoglobulins as its binding site, we can use RosettaAntibodyDesign, an automatic application that designs antibodies with Monte Carlo method. Another thing we should remember is that before antibody design, the input structure must be renumbered, so that the program would know which polypeptide chain, or what secondary structure, is to be designed.
For more details, go to principles.

Ni affinity chromatography
1. Pick 3 colonies from plate and grow in 30mL LB + antibiotic (1‰ Kana or Amp according to vector) at 37℃ overnight. Transfer 10mL culture at 1:30 ratio into 300mL LB + antibiotic. After new culture could be grown at 37℃ into OD600 0.6~0.8 (which normally takes ~3hrs), add IPTG to induce protein expression at 16℃ for 18~20hrs.
2. Collect culture in a 50mL tube, centrifugate consecutively at 4℃ 5,000rpm for 5 min and discard of LB supernate. Resuspend sediment in 10mL lysis buffer (20mM HEPES, 500nM KCl) and centrifugate again. Then discard supernate and resuspend sediment in 20mL lysis buffer.
3. Add protease inhibitor at 1:1000 ratio and extract suspension through ultrasonication at 3sec/3sec for 10 min twice with 10 min interval. Centrifugate lysate at 4℃ 11,000rpm for 60 min. Take samples from supernate and sediment for future examination.
4. Blend supernate with Ni resin in a 50mL tube and incubate horizontally on ice at 160rpm shake for 1.5hr, then erect the tube and place on ice for 0.5 hr to settle mixture. Take sample from solution which would be marked as flow-through. 5. Segregate protein from affinity chromatography at 20nM-500nM imidazole gradient elution.
6. Equilibrate with storage buffer (20mM HEPES, 150mM NaCl).
7. Examine protein purification through SDS-PAGE.

VEGF-scFv
Redesigning in silicon
Sequence before and after mutation:

Heavy chain WT:
E V Q L V E S G G G L V Q P G G S L R L S C A A S G F T I S D Y W I H W V R Q A P G K G L E W V A G I T P A G G Y T Y Y A D S V K G R F T I S A D T S K N T A Y L Q M N S L R A E D T A V Y Y C A R F V F F L P Y A M D Y W G Q G T L V

Heavy chain after mutation:
E V Q L V E S G G G L V Q P G G S L R L S C A A S G F T I S D Y W I H W V R Q A P G K G L E W V A G I V P A G G Y T E Y A D S V K G R F T I S A D T S K N T A Y L Q M N S L R A E D T A V Y Y C A R F V F F W P F A M D Y W G Q G T L V

This is the simplest job, since the VEGF & VEGF antibody complex structure have been texted by X-ray diffraction. That means the interface of those two proteins has been determined. What we have to do is just remolding the amino acid sequence to have a stronger binding affinity. However, this is the best chance to test RosettaDock, since we have got the correct response. After 5000 times of global docking, we finally chose a best structure, in consideration of total energy, binding energy and interface (Antibodies have stationary interface, CDRs). Then, another 1000 times of local docking was operated. (see more from principles)
Figure 2. Docked structure compared to crystal structure
(RMSD= 6.030 Å)
(A) Crystal structure
(B) Docked structure

Finally, we have got a not-so-bad docked structure. As Figure 2 shows, the docked structure has only minor difference compared to the crystal structure.
Next step is sequence design of the VEGF-scFv. With 200 rounds of sequence design of the CDR loop by RosettaAntibodyDesign, we finally mutated 4 residues on the heavy chain: (using AHo Scheme numbering)

Thr-59 → Val
Tyr-69 → Glu
Leu-113 → Trp
Tyr-134 → Phe

The Thr-59 and Tyr-69 belongs to CDR H2, while Leu-113 and Tyr-134 belongs to CDR H3. Let’s see what happens after antibody design.

1.Thr-59 → Val

Figure 3. Energy before and after mutating Thr-59 to Val
Other terms of energy have minor influence
(Terms of Rosetta Energy Function can be found in principles)

As we all know, Thr has polar side chain, while Val has hydrophobic side chain. After mutating Thr-59 to Val, the solvation energy (Figure 3 fa_sol) has dropped dramatically, while the Van der Waals interaction energy (Figure 3 fa_atr & fa_rep, while calculating total energy, fa_rep should multiply by 0.55) has decreased a little. That means this mutation is in favor of the folding of protein, benefiting total energy.

2. Tyr-69 → Glu

This mutation results in notable changes in charge distribution. With negative electrification, Glu has accumulated negative charge. (Figure 4) Since the area surrounding residue 69 has positive electrical charge, the electrostatic energy has reduced. Calculating by the Rosetta energy function, the electrostatic energy is decreasing by 1.25 kcal/mol after the mutation.

Figure 4. Vacuum electrostatics plot around residue 69
(A)Before mutation
(B)After mutation

3. Leu-113 → Trp & Tyr-134 → Phe

These two residues are on CDR H3. In fact, they are close to each other, although they seem separate in AHo Scheme residue numbering. More important, they are directly on the interface of VEGF and VEGF-scFv. The pentapeptide on the interface, Phe Phe Leu Pro Tyr, has been in placed of by Phe Phe Trp Pro Phe.

Figure 5. Interface of VEGF-scFv CDR H3 with VEGF
(A)Before mutation
(B)After mutation
Spheres shows VEGF molecule, while sticks shows the pentapeptide on the interface.
Polar contracts between VEGF and VEGF-scFv are marked with distance. (unit: Å)

This area collected high ratio of aromatic amino acids. π–π stacking plays an important role in the stabilization of conformation. After mutation, the Leu-113 is replaced by Trp, which has an indolyl, an aromatic heterocyclic substituent. This causes another π–π stacking interaction (Figure 6), which increases the Van der Waals interactions, thus stabilizing the conformation. What’s more, Trp-113 generates another H-bond between VEGF and VEGF-scFv, which is contributes directly to the binding energy.

Figure 6. π–π stacking between Trp-113 and Phe-112 after mutation

The mutation of Tyr-134, however, has no significant influence on the Gibbs free energy. It may have minor effects on the conformation stability. Based on Monte Carlo method, RosettaAntibodyDesign does not always give the best answer.
After 4 point mutations, the binding affinity has increased a bit. The interface energy has decreased from -46.405 to -50.641 REU, calculated by RosettaDock.
Also, we have tested a previous part, BBa_K1694003. This is another VEGF-scFv, differing from the one we used. First, we used PIGSPro [1][2][3][4][5][6][7], a web-based tool to build the structure of immunoglobulins. The next step is finding the interface. After 5000 times of global docking and 1000 times of local docking, finally we find the interface of the previous parts, along with VEGF. After some structure refinements, we tested the binding energy of them. From the interface score calculated by RosettaDock, we can see a big difference. Therefore, our parts improvement work is successful.
Figure 7.Comparing binding energy between three types of VEGF-scFv
The interface score is calculated by RosettaDock, which directly shows binding energy.

Figure 8. Parts (BBa_K1694003) local docking RMSD
The interface score is calculated by RosettaDock, which directly shows binding energy.

RMSD is calculated compared to the final docking result, expressing the difference between local docking and the final result. There is clear interdependency between interface score and RMSD, which shows a potential well near the final result. Within the potential well, the structure will be stable. Therefore, our docking result is reliable.
Testing in wet lab
Above, VEGF-scFv was modified in silico to enhance its binding ability to VEGF ligand in our modelling. In protein expression and interaction section, we hope to mutate wildtype VEGF-scFv according to modification provided by dry lab and monitor its binding process with VEGF through a thermodynamic method known as isothermal titration calorimetry (ITC) developed by Adrian Velazquez-Campoy and Ernesto Freire in 2006 [8].

1. Plasmid construction
VEGF-scFv and VEGF-scFv’ were cloned into pGEX expression vector. VEGF sequence was synthesized by Sangon Biotech on a pET28a expression vector.

2. Protein expression and purification
Figure 9. Gradient elution in affinity chromatography for VEGF-scFv’(above) and VEGF (below).
M: marker; L: lysate; SM: sediment; SN: supernate; FT: flow-through; numbers: eluting concentration of imidazole (in mM)

Figure 10. protein purification products equilibrated with storage buffer (see protocols above for detail).
M: marker; L: lysate; SM: sediment; SN: supernate; FT: flow-through; numbers: eluting concentration of imidazole (in mM)

3. ITC
The titration curve shown in Figure 11 is isothermal titration of VEGF and wild-type VEGF-scFv, of which concentrations are 90uM, 15uM respectively. However, there is no significant exotherm in the titration curve, since fluctuations may be related to the influx of the solution and the background noise of the instrument.

Figure 11. Isothermal titration of VEGF and VEGF-scFv

Figure 12. Isothermal titration of VEGF and VEGF-scFv’

The titration curve shown in Figure 12 is isothermal titration of VEGF and mutation VEGF-scFv (named VEGF-scFv’), of which concentrations are 90uM, 15uM respectively. It can be seen that there is a significant exothermic phenomenon when the first two drops of VEGF are instilled, indicating that VEGF is completely bound to VEGF-scFv', which is superior to wild type. However, subsequent data has no obvious meaning.
Because the ITC experiment requires a higher protein concentration and buffer composition. The preliminary analyses lead to unsatisfactory results are as follows. 1) The low concentration of ligand and receptor leads to less meaningful data. 2)VEGF is an insoluble protein, probably due to its poor solubility in buffer (20 mM HEPES + 150 mM NaCl) resulting to lower protein concentration in solution. 3) The activation of proteins in solution buffer were weaken.
We’ ll do further experiments to confirm our conclusion, by optimizing the method of protein expression and purification as well as the contains of solution buffer.

KDR Ig domain 3
Sequence before and after mutation:

WT:
S H G I E L S V G E K L V L N C T A R T E L N V G I D F N W E Y P S S K H Q H K K L V N R D L K T Q S G S E M K K F L S T L T I D G V T R S D Q G L Y T C A A S S G L M T K K N S T F V R V

After mutation:
S H G I E L S V G E K L V L N C V A E T S L N V G I D M T W E Y P S S K H Q H K K L V N R D L K T Q S G S E M K K F L S T L T I D G V T R S D Q G L Y T C T T M T L I V I Q T S E R Q Y T I W R M D V K N S T F V R V

KDR is a receptor with immunoglobulins-like domain. Not a standard antibody, KDR still has some homology with antibodies. We got the crystal structure of KDR binding to VEGF from protein data bank (3v2a.pdb). However, some key residues are missing, probably because of the accuracy of solution structure. We used SWISS-MODEL [9][10][11][12][13] to rebuild the structure. Then, RosettaAntibodyDesign was operated. We tried GraftDesign, replacing whole CDR H3 with a CDR structure from antibody database. The interface energy has decreased from -41.850 to -43.639 REU. Thanks to the GraftDesign, the CDR H3 has been replaced by a longer one, and the contact area has increased a lot.
Figure 13. Interface of KDR and VEGF
Green=VEGF; Cyan=KDR
(A) Before mutation
(B) After mutation
The interface is enlarged after mutation.

D-Dimer scFv
Sequence before and after mutation:

Heavy chain WT:
Q V Q L K Q S G P G L V Q P S Q S L S I T C T V S G F S L T T Y G V H W I R Q S P G K G L E W L G V I W S G G S T D Y N A A F I S R L S I N K D N S K S Q V F F K M N S L Q A N D T A I Y Y C A R N Y W G T S M D Y W G Q G T S V T V S S
Heavy chain after mutation:
Q V Q L K Q S G P G L V Q P S Q S L S I T C E A S E T D M K N A G F G W I R Q S P G K G L E W L G G V D G G L G K E H G T T Y N A A F I S R L S I N K D N S K S Q V F F K M N S L Q A N D T A I Y Y C T S A T G S A N W G Q G T S V T V S S

Light chain WT:
D I K M T Q S P S S M Y A S L G E R V T V T C K A S Q D I N S Y L S W I Q Q K P G K S P K T L I Y R G N R L V A G V P S R F S G S G S G Q D Y S L T I S S L E Y E D V G V Y Y C L R Y D E F P F T F G S G T K L E I K R A G Q G S S V
Light chain after mutation: D I K M T Q S P S S M Y A S L G E R V T V T C Q A S Q D I G N D L S W I Q Q K P G K S P K T L I Y K V S K R A S G V P S R F S G S G S G Q D Y S L T I S S L E Y E D V G V Y Y C Q Q G Y S S P S T F G S G T K L E I K R A G Q G S S V

Another signal molecule we are fascinating with is D-Dimer. What we have found is the crystal structure of D-Dimer (PDB ID: 2Q9I), and the amino acid sequence of D-Dimer scFv. SWISS-MODEL helps us to model the structure of D-Dimer scFv. Next is docking, finding out the interface.
Figure 14. D-Dimer scFv local docking RMSD

After rounds of docking, finally we found the interface (Figure 14). As has mentioned before, this RMSD plot shows a reliable result of docking. The final step is RosettaAntibodyDesign. Here, we made a bold attempt, execute the command -random_start. This will start graft design with a new set of CDRs from the database, not biasing the run with native CDRs. The result is successful. The binding energy has been optimized from -31.561 to -54.875 (unit:REU). Since the CDR region is totally different after mutation, we can claim to create a new antibody. Merely, the position of interface is the same.

tTA promoter
DNA Sequence before and after mutation:

WT:
T C T A T C A C T G A T A G G

After mutation:
T C T A T C A C T G C T A G G

We can also redesign DNA sequence. Here is an example, a promoter and transcription factor complex, tTA. In fact, the structure of tTA has not been tested through experiment. However, TetR, another transcription factor that has the same sequence on the protein-DNA interface, has known structure (PDB ID: 1QPI). So it is equivalent to redesign the interface of TetR.

Figure 15. Vacuum electrostatics plot of interface near base pair 11
(A)Before mutation
(B)After mutation
The transcription factor is colored to show its electrostatics distribution. It should be mentioned that the area with less electric charge is always hydrophobic.

After mutation on base pair 11 (A to C), the interface energy has decreased from-37.467 to -38.411 REU, tested by RosettaDock. The most significant improve is lk_ball_wtd, or orientation-dependent solvation energy. Because of the hydrophobic area on the transcription factor’s interface, solvation energy has major effect. (Figure 16)

Figure 16. Δ lk_ball_wtd on base pair 10-12
The energy term lk_ball_wtd means orientation-dependent solvation energy.
(Terms of Rosetta Energy Function can be found in principles)

References
[1] Marcatili, P., Rosi, A. and Tramontano, A. (2008) PIGS: automatic prediction of antibody structures. Bioinformatics, 24, 1953-1954. [2] Chothia, C. and Lesk, A.M. (1987) Canonical structures for the hypervariable regions of immunoglobulins. J Mol Biol, 196, 901-917. [3] Morea, V., Tramontano, A., Rustici, M., Chothia, C. and Lesk, A.M. (1998) Conformations of the third hypervariable region in the VH domain of immunoglobulins. J Mol Biol, 275, 269-294. [4] Tramontano, A., Chothia, C. and Lesk, A.M. (1990) Framework residue 71 is a major determinant of the position and conformation of the second hypervariable region in the VH domains of immunoglobulins. J Mol Biol, 215, 175-182. [5] Chailyan, A., Marcatili, P., Cirillo, D. and Tramontano, A. (2011) Structural repertoire of immunoglobulin lambda light chains. Proteins, 79, 1513-1524. [6] Chailyan, A., Marcatili, P. and Tramontano, A. (2011) The association of heavy and light chain variable domains in antibodies: implications for antigen specificity. FEBS J, 278, 2858-2866. [7] Messih, M.A., Lepore, R., Marcatili, P. and Tramontano, A. (2014) Improving the accuracy of the structure prediction of the third hypervariable loop of the heavy chains of antibodies. Bioinformatics, 30, 2733-2740. [8] Adrian Velazquez-Campoy and Ernesto Freire. Isothermal titration calorimetry to determine association constants for high-affinity ligands. Nature Protocols, 2006, 1(1):186-191. doi:10.1038/nprot.2006.28 [9] Waterhouse, A., Bertoni, M., Bienert, S., Studer, G., Tauriello, G., Gumienny, R., Heer, F.T., de Beer, T.A.P., Rempfer, C., Bordoli, L., Lepore, R., Schwede, T. SWISS-MODEL: homology modelling of protein structures and complexes. Nucleic Acids Res. 46(W1), W296-W303 (2018). [10] Bienert, S., Waterhouse, A., de Beer, T.A.P., Tauriello, G., Studer, G., Bordoli, L., Schwede, T. The SWISS-MODEL Repository - new features and functionality. Nucleic Acids Res. 45, D313-D319 (2017). [11] Guex, N., Peitsch, M.C., Schwede, T. Automated comparative protein structure modeling with SWISS-MODEL and Swiss-PdbViewer: A historical perspective. Electrophoresis 30, [12] Benkert, P., Biasini, M., Schwede, T. Toward the estimation of the absolute quality of individual protein structure models. Bioinformatics 27, 343-350 (2011). [13] Bertoni, M., Kiefer, F., Biasini, M., Bordoli, L., Schwede, T. Modeling protein quaternary structure of homo- and hetero-oligomers beyond binary interactions by homology. Scientific Reports 7 (2017).

  Address



G604, School of Life Sciences, Fudan University
2005 Songhu Road, Yangpu, Shanghai, China