Knowledge-based potential for partial covalent interactions (PCI-

KBP)

Steven Combs

1

Partial Covalent Interactions (PCI)

• An interaction between an e-

deficient atom to a slightly positive atom

– Cation-pi

– Pi-pi

– Salt bridge

– Hydrogen Bonds

2

Cation-π π - π

Hydrogen Bonds / Salt bridges

Placement of Orbitals on Atoms Allows for an Accurate Description of PCIs

3

Orbital Type Name

Hybridization Atom Type Example Picture

C.pi.sp2 sp2aroCCOO

CNH2

Tyr, Trp, PheAsp, GluAsn, Gln

N.pi.sp2 sp2

NtrpNargNhis

NH20

TrpArgHis

Asn, Gln

N.p.sp2 sp2 Nhis His

O.pi.sp2 sp2ONH2OOC

Asn, GlnAsp, Glu

O.p.sp2 sp2ONH2OOC

Asn, GlnAsp, Glu

O.p.sp3 sp3 OH Ser, Thr, Tyr

S.p.sp3 sp3 S Cys, Met

θ H

X’

H

Orbital Placement Improves Sharpness and Depth of Energy Wells

Cation-π

θ Hpol

X’ 4

Rosetta Orbital Type Name

Rosetta Orbital Type

Hybridization Rosetta

Atom Type Picture

C.pi.sp2 pi sp2 aroC COO CNH2

-1.50

-1.20

-0.90

-0.60

-0.30

0.00

0.05

0.55

1.05

1.55

2.05

2.55

3.05

3.55

180:

120:

90:

An

gleθ

Distance δH-Y’(Å)

Ener

gy (

no

t sc

aled

)

0:

-1.50

-1.20

-0.90

-0.60

-0.30

0.00

0.0

5

0.5

5

1.0

5

1.55

2.0

5

2.5

5

3.0

5

3.5

5

Orbital Placement Improves Sharpness and Depth of Energy Wells

π - π

180:

120:

90:

An

gleθ

Distance δH-Y’(Å)

θ Haro

X’

5

Rosetta Orbital Type Name

Rosetta Orbital Type

Hybridization Rosetta

Atom Type Picture

C.pi.sp2 pi sp2 aroC COO CNH2

Ener

gy (

no

t sc

aled

)

0:

Orbital Placement Improves Sharpness and Depth of Energy Wells

6

-3.00

-2.70

-2.40

-2.10

-1.80

-1.50

-1.20

-0.90

-0.60

-0.30

0.00

0.05

0.55

1.05

1.55

2.05

2.55

3.05

3.55

θ Hpol

X’

180:

120:

90:

An

gleθ

Distance δH-Y’(Å)

Ener

gy (

no

t sc

aled

)

0:

Rosetta Orbital Type Name

Rosetta Orbital Type

Hybridization Atom Type Picture

O.p.sp2 p sp2 ONH2 OOC

The PCI-KBP Is Divided into Two Scoring Terms

Orbital Type Name

Hybridization Atom Type Example Picture

C.pi.sp2 sp2aroCCOO

CNH2

Tyr, Trp, PheAsp, GluAsn, Gln

N.pi.sp2 sp2

NtrpNargNhis

NH20

TrpArgHis

Asn, Gln

N.p.sp2 sp2 Nhis His

O.pi.sp2 sp2ONH2OOC

Asn, GlnAsp, Glu

O.p.sp2 sp2ONH2OOC

Asn, GlnAsp, Glu

O.p.sp3 sp3 OH Ser, Thr, Tyr

S.p.sp3 sp3 S Cys, Met

orbitals_hpol

orbitals_haro

7

Weights Used During Optimization

Free Removed

8

Score Term Description

Fa_atr Attractive portion of LJ

Fa_rep Repulsive portion of LJ

Fa_sol Implicit solvation potential

Fa_intra_rep Intra-residue repulsion

Pro_close Close the proline ring

Hbond Hydrogen bond potential

Dslf Disulphide potential

Rama Phi-psi angles in ramachandran plot

Fa_dun Dunbrack rotamer library

P_aa_pp Probability of an amino acid given phi-psi angle

hpol, haro, sol pair, hbond_sc

hbond_sc, pair, sol

hpol, haro, bb_hpol, sol

pair, hbond_sc, hbond_bb_sc

1.

2.

3.

8

Score12

Benchmark Set-up

• Dataset from et al Grishin 2001

• Better than 1.8A

• 100-500 Residues

• Monomeric

• No ligands

• Complete Side chains

9

Introduction of Partial Covalent Interactions within Rosetta Improves

Protein Metrics

Location % SR

Buried 63

Surface 31

Overall 46

Location % SR

Buried 65

Surface 35

Overall 49

Orbitals

Location % SR

Buried 69

Surface 36

Overall 52

Score 12 Optimized Score12

Location % PSSM

Buried 72

Boundary 73

Surface 73

Overall 72

Location % PSSM

Buried 77

Boundary 74

Surface 76

Overall 76

Location % PSSM

Buried 78

Boundary 78

Surface 77

Overall 78

Packstats 0.68 Packstats 0.63 Packstats 0.60

Identifying non-Classical PCIs Sulfur-Aromatic

11

Orbital Type Name

Hybridization Atom Type Example Picture

C.pi.sp2 sp2aroCCOO

CNH2

Tyr, Trp, PheAsp, GluAsn, Gln

N.pi.sp2 sp2

NtrpNargNhis

NH20

TrpArgHis

Asn, Gln

N.p.sp2 sp2 Nhis His

O.pi.sp2 sp2ONH2OOC

Asn, GlnAsp, Glu

O.p.sp2 sp2ONH2OOC

Asn, GlnAsp, Glu

O.p.sp3 sp3 OH Ser, Thr, Tyr

S.p.sp3 sp3 S Cys, Met

180:

120:

90:

An

gleθ

Distance δH-Y’(Å)

0:

0 2 4

Identifying non-Classical PCIs Anion-Pi

12

Orbital Type Name

Hybridization Atom Type Example Picture

C.pi.sp2 sp2aroCCOO

CNH2

Tyr, Trp, PheAsp, GluAsn, Gln

N.pi.sp2 sp2

NtrpNargNhis

NH20

TrpArgHis

Asn, Gln

N.p.sp2 sp2 Nhis His

O.pi.sp2 sp2ONH2OOC

Asn, GlnAsp, Glu

O.p.sp2 sp2ONH2OOC

Asn, GlnAsp, Glu

O.p.sp3 sp3 OH Ser, Thr, Tyr

S.p.sp3 sp3 S Cys, Met

180:

120:

90:

An

gleθ

Distance δH-Y’(Å)

0:

0 2 4

Conclusions

• Orbital placement allows for accurate description of residue interactions

• PCI-KBP can perform as well as score12

• New interactions are captured by PCI-KBP

13

Future Work

• Extend PCI-KBP to ligands/DNA/RNA

• Improve geometry of interactions

• Fix little bugs/memory leaks/speed

14

Acknowledgements Meiler Lab Sam Deluca Gordon Lemmon Nils Woetzel Kuhlman Lab Andrew Leaver-Fay Matt O’Mera

Committee Dr. Brian Bachmann Dr. Richard Armstrong Dr. Jens Meiler Dr. Martin Egli