in the Analysis of Protein and Nucleic Acid Secondary ...

2
Figure 1. The backbone path of a protein (PDB code 1ag6) is simplified by ignoring R-group appendages and by defining the orientation of the residue based on generic backbone features. In this case, each peptide bond is idealized as a planar shape whose orientation is based on the orientation of local backbone atoms. Novel Use of Novel Use of Quaternions Quaternions in the Analysis of in the Analysis of Protein and Nucleic Acid Secondary Structure Protein and Nucleic Acid Secondary Structure Daniel D. Kohler and Robert M. Hanson St. Olaf College, Northfield, Minnesota 237 th National Meeting of the American Chemical Society, Salt Lake City, Utah, March 24, 2009 Quaternionshave been known for some time to be expedient with problems involving rigid body rotations, but despite heavy usage of these highly visual objects, meaningful visualization within applications such as molecular modeling has been largely devoid. To address this disparity, this study applies quaternionsto parameterize protein backbone structure, focusing on the identification of helix parameters through visual and metric means. Quaternion maps are introduced as a tool for analysis of backbone character, anda measure of secondary structure backbone volatility, which we define as quaternion straightness, is introduced. We find that this quaternion analysis provides a simple way to identify structural motifs in secondary structure, as well as a metric means to define helix parameters. Straightness is further applied as a statistical tool for gross evaluation of peptide structural integrity. A direct relation between relative quaternion differences and Ramachandrantorsion angles is found. A unit quaternion is a normalized 4- dimensional vector. A quaternion is frequently separated into a scalar part and a vector part. Rotational operators in R 3 are homomorphic to the unit quaternions. Quaternions can be parameterized by Rodriguez parameters: for a rotation of θ radians about normal vector n, Abstract Abstract Rigid Body Idealization Rigid Body Idealization Let a peptide's amino acid residues be indexed following the order of peptide bonding, from N- terminus to C-terminus. We wish to characterize the structural difference from residue i to residue i+1. As described by Quine, 1 one can think of a protein's secondary structure as a “discrete curve” in R 3 , where, as shown in Figure 1, backbone atoms of a given residue are defined on the curve as points in space. It is asserted that structural parameters can be simplified by ignoring the minor fluctuations in bond angles and lengths ofthe backbone atoms, so that all backbone moieties are equivalent. The coordinates of residue atoms can then be simplified analytically by reducing the expression of points to a common geometric feature. Not only does this reduce the number of independent structures in the modelingprocedure (individual points in the moiety can now be considered a single collection of dependently linked points), but it also invites a characterization of the backbone structure analogous to motion of a rigid body. Any positional change of a rigid body can be described through a single rotation and a single translation. 2 Quaternions Quaternions and Quaternion Maps and Quaternion Maps Quaternion Differences Quaternion Differences Helices Helices Quaternion Straightness Quaternion Straightness Structural Integrity Structural Integrity Conclusions Conclusions Acknowledgements Acknowledgements Works Cited Works Cited Peptide Frames Peptide Frames To describe the orientation of peptide residues we define special bases, or frames, for each residue. A frame is a triplet of orthonormal vectors associatedwith a point in space. Frames can be defined according to the generic bonds of a residue. Consequently, frames can fail to represent the residue accurately when the residue does not obey generic bond angles sufficiently. Frames should be chosen based on the more consistent bonding characteristics of the backbone. Figure 2. C-, P-, and N-based frames. The P and N frames are examples of plane-based frames, where the plane is colored green. The C frame is based on the tetrahedral geometry of the alpha carbon. A reference frame is given adjacent to each model. To view the multitude of quaternionsmeasured, we use quaternion maps, 3 which plot the vector component of quaternions(Figure 3). Figure 3. The protein crambin(PDB 1crn) with frames and a quaternion map associated with those frames. Coloring scheme is from N-terminus (blue) to C-terminus (red). Jmolis used for visualization. 4 Quaternionsprovide a novel way to express both differences and similarities between individual pairs of residues as well among entire strands. The mathematicalbasis of structural integrity and straightness correlates well with simple measures using Ramachandranangles and has the advantage of easily being extended to non-protein systems. While this study focuses primarily on sequential residues in peptides, another study in our group (poster BIOL 149) examined a broader use of quaternions to look at the flexibility of residues over time in a moleculardynamics calculation. Statistical approaches might be used to correlate straightness values and more qualitative visual depictions of secondary structure. Figure 7. Relative distributions of straightness values for various molecules. Models with a relatively higher proportion of straightness near 1.00 have relatively more order in their secondary structure. Models 1ppk and 1a6g have very different straightness distributions, which can be seen visually by inspection and measured quantitatively using quaternions. Quaternion differences can be measured two different ways depending upon the intended application. One option is to measure the difference between two quaternions with respect to the reference frame, RF (the x, y, and z coordinates of a pdbfile). Additionally, one can consider the quaternion that rotates one peptide frame, PF i , to the next, PF i+1 with respect to PF i (Figure 4). Figure 4. Rotations are a non-commutative algebra, and pathways to a given basis can yield entirely different results with different information. The frame pf i is a pseudo-frame, not physically present in the molecule. 1ppk 1a6g Absolute difference (dq). The quaternion that rotates PF i to PF i+1 . This preserves information of global orientation. Relative difference (dq*). The quaternion that rotates PF i to PF i+1 as if PF i were the reference frame. This relativizesthe domains of each transformation. Figure 5. Structural measurements of an ideal helix. The structure was generated by iterating a rigid body transformation of an idealized alanineresidue 16 times using Ramachandranangles (phi, psi, omega) = (-65,-40,0), and obtaining quaternion maps of the C frame. Figures (a),(b), and (c) have the same reference frame perspective, as shown by the global axes assignments. In (b), the observed point is between the viewer and the center when viewed down the axis of the helix. Quaternion map (d) displays (c) from a different perspective. Figure 6. The M2 membrane H+ channel of influenza (1mp6) and its absolute quaternion difference map. The cluster of points is between the viewer and the center when viewed down the axis of the helix. (a) (b) (c) (d) • Prof. Andrew J. Hanson, University of Indiana-Bloomington • St. Olaf College • HHMI [1] Quine, J. Journal of Molecular Structure 1999, 460, 53-66. [2] Kneller, G. R.; Calligari, P. ActaCryst. 2006, 62, 302. [3] Hanson, A. “Visualizing Quaternions;”Morgan Kaufmann, 2006. [4] Jmol, http://jmol.sourceforge.net/ Shown on the right are two depictions of structure 1a6g. The model on the left is colored by structure, as defined in the PDB file. On the right, coloring is based on “straightness”as defined by the four-dimensional dot product of consecutive quaternion differences, which are displayed below the structures. Several regions of high straightness where no PDB structure is defined indicate short motifs not found by the DSSP algorithm. Shown on the left is a correlation between straightness, as defined above, and a function of Ramachandranangles defined as: f(i) = phi i+1 + psi i –phi i –psi i-1 Cis-prolinesare anomalous, as expected. This Ramachandranfunction could be used as an alternative measure of straightness if access to quaternion information is limited.

Transcript of in the Analysis of Protein and Nucleic Acid Secondary ...

Page 1: in the Analysis of Protein and Nucleic Acid Secondary ...

Fig

ure

1.

Th

e b

ack

bo

ne

pat

h o

f a

pro

tein

(P

DB

co

de

1ag

6) i

s si

mp

lifi

ed b

y i

gn

ori

ng

R-g

rou

p a

pp

end

ag

es a

nd

by

d

efin

ing

th

e o

rien

tati

on

of

the

resi

du

e b

ased

on

gen

eric

bac

kb

on

e fe

atu

res.

In

th

is c

ase,

eac

h p

epti

de

bo

nd

is

idea

lize

d a

s a

pla

nar

sh

ape

wh

ose

ori

enta

tio

n i

s b

ased

on

th

e o

rien

tati

on

of

loca

l b

ack

bo

ne

ato

ms.

No

vel

Use

of

No

vel

Use

of

Qu

ater

nio

ns

Qu

ater

nio

ns

in t

he

An

aly

sis

of

in t

he

An

aly

sis

of

Pro

tein

an

d N

ucl

eic

Aci

d S

eco

nd

ary

Str

uct

ure

Pro

tein

an

d N

ucl

eic

Aci

d S

eco

nd

ary

Str

uct

ure

Da

nie

l D

. K

oh

ler

an

d R

ob

ert

M. H

an

son

St.

Ola

f C

oll

ege,

No

rth

fiel

d,

Min

nes

ota

23

7th

Na

tio

na

l M

eeti

ng

of

the

Am

eric

an

Ch

emic

al

So

ciet

y, S

alt

La

ke

Cit

y, U

tah

, Ma

rch

24

, 200

9

Qu

ater

nio

ns

hav

e b

een

kn

ow

n f

or

som

e ti

me

to b

e ex

ped

ien

t w

ith

pro

ble

ms

inv

olv

ing

rig

id

bo

dy

ro

tati

on

s, b

ut

des

pit

e h

eav

y u

sag

e o

f th

ese

hig

hly

vis

ual

ob

ject

s, m

ean

ing

ful

vis

ual

izat

ion

w

ith

in a

pp

lica

tio

ns

such

as

mo

lecu

lar

mo

del

ing

has

bee

n l

arg

ely

dev

oid

. T

o a

dd

ress

th

is

dis

par

ity

, th

is s

tud

y a

pp

lies

qu

ater

nio

ns

to p

aram

eter

ize

pro

tein

bac

kb

on

e st

ruct

ure

, fo

cusi

ng

o

n t

he

iden

tifi

cati

on

of

hel

ix p

aram

eter

s th

rou

gh

vis

ual

an

d m

etri

c m

ean

s.

Qu

ater

nio

n m

aps

are

intr

od

uce

d a

s a

too

l fo

r an

aly

sis

of

bac

kb

on

e ch

arac

ter,

an

da

mea

sure

of

seco

nd

ary

st

ruct

ure

bac

kb

on

e v

ola

tili

ty, w

hic

h w

e d

efin

e as

qu

ater

nio

n s

trai

gh

tnes

s, i

s in

tro

du

ced

. W

e fi

nd

th

at t

his

qu

ater

nio

n a

nal

ysi

s p

rov

ides

a s

imp

le w

ay t

o i

den

tify

str

uct

ura

l m

oti

fs i

n

seco

nd

ary

str

uct

ure

, as

wel

l as

a m

etri

c m

ean

s to

def

ine

hel

ix p

aram

eter

s.

Str

aig

htn

ess

is

furt

her

ap

pli

ed a

s a

stat

isti

cal

too

l fo

r g

ross

ev

alu

atio

n o

f p

epti

de

stru

ctu

ral

inte

gri

ty. A

dir

ect

rela

tio

n b

etw

een

rel

ativ

e q

uat

ern

ion

dif

fere

nce

s an

d R

amac

han

dra

nto

rsio

n a

ng

les

is f

ou

nd

.

•A

un

it q

uat

ern

ion

is

a n

orm

aliz

ed 4

-d

imen

sio

nal

vec

tor.

A q

uat

ern

ion

is

freq

uen

tly

sep

arat

ed i

nto

a s

cala

r p

art

and

a

vec

tor

par

t.

•R

ota

tio

nal

op

erat

ors

in

R3

are

ho

mo

mo

rph

icto

th

e u

nit

qu

ater

nio

ns.

Qu

ater

nio

ns

can

be

par

amet

eriz

ed b

y

Ro

dri

gu

ez p

aram

eter

s: f

or

a ro

tati

on

of

θra

dia

ns

abo

ut

no

rmal

vec

tor

n,

Ab

stra

ctA

bst

ract

Rig

id B

od

y I

deali

zati

on

Rig

id B

od

y I

deali

zati

on

Let

a p

epti

de'

s am

ino

aci

d r

esid

ues

be

ind

exed

fo

llo

win

g t

he

ord

er o

f p

epti

de

bo

nd

ing

, fro

m N

-te

rmin

us

to C

-ter

min

us.

W

e w

ish

to

ch

arac

teri

ze t

he

stru

ctu

ral

dif

fere

nce

fro

m r

esid

ue

ito

re

sid

ue

i+1.

As

des

crib

ed b

y Q

uin

e,1

on

e ca

n t

hin

k o

f a

pro

tein

's s

eco

nd

ary

str

uct

ure

as

a “d

iscr

ete

curv

e”in

R3, w

her

e, a

s sh

ow

n i

n F

igu

re 1

, b

ack

bo

ne

ato

ms

of

a g

iven

res

idu

e ar

e d

efin

ed o

n t

he

curv

e as

po

ints

in

sp

ace.

It

is a

sser

ted

th

at s

tru

ctu

ral

par

amet

ers

can

be

sim

pli

fied

b

y i

gn

ori

ng

th

e m

ino

r fl

uct

uat

ion

s in

bo

nd

an

gle

s an

d l

eng

ths

of

the

bac

kb

on

e at

om

s, s

o t

hat

all

b

ack

bo

ne

mo

ieti

es a

re e

qu

ival

ent.

Th

e co

ord

inat

es o

f re

sid

ue

ato

ms

can

th

en b

e si

mp

lifi

ed

anal

yti

call

y b

y r

edu

cin

g t

he

exp

ress

ion

of

po

ints

to

a c

om

mo

n g

eom

etri

c fe

atu

re. N

ot

on

ly d

oes

th

is r

edu

ce t

he

nu

mb

er o

f in

dep

end

ent

stru

ctu

res

in t

he

mo

del

ing

pro

ced

ure

(in

div

idu

al p

oin

ts

in t

he

mo

iety

can

no

w b

e co

nsi

der

ed a

sin

gle

co

llec

tio

n o

f d

epen

den

tly

lin

ked

po

ints

), b

ut

it a

lso

in

vit

es a

ch

arac

teri

zati

on

of

the

bac

kb

on

e st

ruct

ure

an

alo

go

us

to m

oti

on

of

a ri

gid

bo

dy

. A

ny

p

osi

tio

nal

ch

ang

e o

f a

rig

id b

od

y c

an b

e d

escr

ibed

th

rou

gh

a s

ing

le r

ota

tio

n a

nd

a s

ing

le

tran

slat

ion

.2

Qu

ate

rnio

ns

Qu

ate

rnio

ns

an

d Q

uate

rnio

n M

ap

san

d Q

uate

rnio

n M

ap

s

Qu

ate

rnio

n D

iffe

ren

ces

Qu

ate

rnio

n D

iffe

ren

ces

Heli

ces

Heli

ces

Qu

ate

rnio

n S

traig

htn

ess

Qu

ate

rnio

n S

traig

htn

ess

Str

uct

ura

l In

teg

rity

Str

uct

ura

l In

teg

rity

Co

ncl

usi

on

sC

on

clu

sio

ns

Ack

no

wle

dg

em

en

tsA

ckn

ow

led

ge

me

nts

Wo

rks

Cit

ed

Wo

rks

Cit

ed

Pep

tid

e F

ram

es

Pep

tid

e F

ram

es

To

des

crib

e th

e o

rien

tati

on

of

pep

tid

e re

sid

ues

we

def

ine

spec

ial

bas

es, o

r fr

ames

, fo

r ea

ch

resi

du

e. A

fra

me

is a

tri

ple

t o

f o

rth

on

orm

al v

ecto

rs a

sso

ciat

edw

ith

a p

oin

t in

sp

ace.

Fra

mes

can

b

e d

efin

ed a

cco

rdin

g t

o t

he

gen

eric

bo

nd

s o

f a

resi

du

e. C

on

seq

uen

tly

, fra

mes

can

fai

l to

re

pre

sen

t th

e re

sid

ue

accu

rate

ly w

hen

th

e re

sid

ue

do

es n

ot

ob

ey g

ener

ic b

on

d a

ng

les

suff

icie

ntl

y.

Fra

mes

sh

ou

ld b

e ch

ose

n b

ased

on

th

e m

ore

co

nsi

sten

t b

on

din

g c

har

acte

rist

ics

of

the

bac

kb

on

e.

Fig

ure

2.

C-,

P-,

an

d N

-bas

ed f

ram

es.

Th

e P

an

d N

fram

es a

re e

xam

ple

s o

f p

lan

e-b

ased

fra

mes

, w

her

e th

e p

lan

e is

co

lore

d g

reen

. T

he

C f

ram

e is

bas

ed o

n t

he

tetr

ah

edra

l g

eom

etry

of

the

alp

ha

carb

on

. A

ref

eren

ce f

ram

e is

giv

en

adja

cen

t to

eac

h m

od

el.

To

vie

w t

he

mu

ltit

ud

e o

f q

uat

ern

ion

sm

easu

red

, w

e u

se q

uat

ern

ion

map

s,3

wh

ich

plo

t th

e v

ecto

r co

mp

on

ent

of

qu

ater

nio

ns

(Fig

ure

3).

Fig

ure

3.

Th

e p

rote

in c

ram

bin

(PD

B 1

crn

) w

ith

fr

ames

an

d a

qu

ater

nio

n m

ap a

sso

ciat

ed w

ith

th

ose

fr

ames

. C

olo

rin

g s

chem

e is

fro

m N

-ter

min

us

(blu

e)

to C

-ter

min

us

(red

).

Jmo

lis

use

d f

or

vis

ual

izat

ion

.4

Qu

ater

nio

ns

pro

vid

e a

no

vel

way

to

exp

ress

bo

th d

iffe

ren

ces

and

sim

ilar

itie

s b

etw

een

in

div

idu

al

pai

rs o

f re

sid

ues

as

wel

l am

on

g e

nti

re s

tran

ds.

Th

e m

ath

emat

ical

bas

is o

f st

ruct

ura

l in

teg

rity

an

d

stra

igh

tnes

s co

rrel

ates

wel

l w

ith

sim

ple

mea

sure

s u

sin

g R

amac

han

dra

nan

gle

s an

d h

as t

he

adv

anta

ge

of

easi

ly b

ein

g e

xte

nd

ed t

o n

on

-pro

tein

sy

stem

s. W

hil

e th

is s

tud

y f

ocu

ses

pri

mar

ily

on

se

qu

enti

al r

esid

ues

in

pep

tid

es, a

no

ther

stu

dy

in

ou

r g

rou

p (

po

ster

BIO

L 1

49)

exam

ined

a b

road

er

use

of

qu

ater

nio

ns

to l

oo

k a

t th

e fl

exib

ilit

y o

f re

sid

ues

ov

er t

ime

in a

mo

lecu

lar

dy

nam

ics

calc

ula

tio

n.

Sta

tist

ical

ap

pro

ach

es m

igh

t b

e u

sed

to

co

rrel

ate

stra

igh

tnes

s v

alu

es a

nd

mo

re

qu

alit

ativ

e v

isu

al d

epic

tio

ns

of

seco

nd

ary

str

uct

ure

.

Fig

ure

7.

Rel

ativ

e d

istr

ibu

tio

ns

of

stra

igh

tnes

s v

alu

es f

or

va

rio

us

mo

lecu

les.

Mo

del

s w

ith

a r

elat

ivel

y h

igh

er

pro

po

rtio

n o

f st

raig

htn

ess

nea

r 1.

00 h

av

e re

lati

vel

y m

ore

ord

er i

n t

hei

r se

con

dar

y s

tru

ctu

re.

Mo

del

s 1p

pk

an

d 1

a6g

h

av

e v

ery

dif

fere

nt

stra

igh

tnes

s d

istr

ibu

tio

ns,

wh

ich

ca

n b

e se

en v

isu

all

y b

y i

nsp

ecti

on

an

d m

easu

red

qu

an

tita

tiv

ely

u

sin

g q

uat

ern

ion

s.

Qu

ater

nio

n d

iffe

ren

ces

can

be

mea

sure

d t

wo

d

iffe

ren

t w

ays

dep

end

ing

up

on

th

e in

ten

ded

ap

pli

cati

on

. O

ne

op

tio

n i

s to

mea

sure

th

e d

iffe

ren

ce b

etw

een

tw

o q

uat

ern

ion

sw

ith

res

pec

t to

th

e re

fere

nce

fra

me,

RF

(th

e x,

y,

and

zco

ord

inat

es o

f a

pd

bfi

le).

Ad

dit

ion

ally

, on

e ca

n

con

sid

er t

he

qu

ater

nio

n t

hat

ro

tate

s o

ne

pep

tid

e fr

ame,

PF

i, to

th

e n

ext,

PF

i+1

wit

h r

esp

ect

to P

Fi

(Fig

ure

4).

Fig

ure

4.

Ro

tati

on

s ar

e a

no

n-c

om

mu

tati

ve

alg

ebra

, a

nd

p

ath

way

s to

a g

iven

bas

is c

an y

ield

en

tire

ly d

iffe

ren

t re

sult

s w

ith

dif

fere

nt

info

rmat

ion

. T

he

fra

me

pfiis

a

pse

ud

o-f

ram

e, n

ot

ph

ysi

call

y p

rese

nt

in t

he

mo

lecu

le.

1ppk

1a6g

•A

bso

lute

dif

fere

nce

(d

q).

T

he

qu

ater

nio

n t

hat

ro

tate

s P

Fito

PF

i+1.

Th

is p

rese

rves

in

form

atio

n

of

glo

bal

ori

enta

tio

n.

•R

ela

tiv

e d

iffe

ren

ce (

dq

*).

Th

e q

uat

ern

ion

th

at

rota

tes

PF

ito

PF

i+1

as i

fP

Fiw

ere

the

refe

ren

ce

fram

e.

Th

is r

elat

iviz

esth

e d

om

ain

s o

f ea

ch

tran

sfo

rmat

ion

.

Fig

ure

5.

Str

uct

ura

l m

easu

rem

ents

of

an

id

eal

hel

ix.

Th

e st

ruct

ure

was

gen

erat

ed b

y i

tera

tin

g a

rig

id b

od

y

tran

sfo

rmat

ion

of

an

id

eali

zed

ala

nin

ere

sid

ue

16 t

imes

usi

ng

Ram

ach

an

dra

na

ng

les

(ph

i, p

si, o

meg

a) =

(-6

5,-4

0,0)

, an

d

ob

tain

ing

qu

ater

nio

n m

aps

of

the

Cfr

ame.

F

igu

res

(a),

(b),

an

d (

c) h

ave

the

sam

e re

fere

nce

fra

me

per

spec

tiv

e, a

s sh

ow

n

by

th

e g

lob

al a

xes

ass

ign

men

ts.

In

(b

), t

he

ob

serv

ed p

oin

t is

bet

wee

n t

he

vie

wer

an

d t

he

cen

ter

wh

en v

iew

ed d

ow

n t

he

axis

of

the

hel

ix.

Qu

ater

nio

n m

ap (

d)

dis

pla

ys

(c)

fro

m a

dif

fere

nt

per

spec

tiv

e.

Fig

ure

6.

Th

e M

2 m

emb

ran

e H

+ ch

an

nel

of

infl

uen

za

(1m

p6)

an

d i

ts a

bso

lute

qu

ater

nio

n d

iffe

ren

ce m

ap.

T

he

clu

ster

of

po

ints

is

bet

wee

n t

he

vie

wer

an

d t

he

cen

ter

wh

en v

iew

ed d

ow

n t

he

axis

of

the

hel

ix.

(a)

(b)

(c)

(d)

•P

rof.

An

dre

w J

. H

anso

n,

Un

iver

sity

of

Ind

ian

a-B

loo

min

gto

n

•S

t. O

laf

Co

lleg

e

•H

HM

I

[1]

Qu

ine,

J.

Jou

rnal

of

Mol

ecu

lar

Str

uct

ure

1999,

460,

53-

66.

[2]

Kn

elle

r, G

. R

.; C

alli

gar

i, P

. A

cta

Cry

st. 2

006,

62,

302.

[3]

Han

son

, A

. “V

isu

aliz

ing

Qu

ater

nio

ns;

”M

org

an

Ka

ufm

an

n,

2006

.[4

] Jm

ol,

htt

p:/

/jm

ol.

sou

rcef

org

e.n

et/

Sh

ow

n o

n t

he

rig

ht

are

two

d

epic

tio

ns

of

stru

ctu

re 1

a6g

. T

he

mo

del

on

th

e le

ft i

s co

lore

d b

y

stru

ctu

re, a

s d

efin

ed i

n t

he

PD

B f

ile.

O

n t

he

rig

ht,

co

lori

ng

is

bas

ed o

n

“str

aig

htn

ess”

as d

efin

ed b

y t

he

fou

r-d

imen

sio

nal

do

t p

rod

uct

of

con

secu

tiv

e q

uat

ern

ion

dif

fere

nce

s,

wh

ich

are

dis

pla

yed

bel

ow

th

e st

ruct

ure

s. S

ever

al r

egio

ns

of

hig

h

stra

igh

tnes

s w

her

e n

o P

DB

str

uct

ure

is

def

ined

in

dic

ate

sho

rt m

oti

fs n

ot

fou

nd

by

th

e D

SS

P a

lgo

rith

m.

Sh

ow

n o

n t

he

left

is

a co

rrel

atio

n b

etw

een

st

raig

htn

ess,

as

def

ined

ab

ov

e, a

nd

a f

un

ctio

n

of

Ram

ach

and

ran

ang

les

def

ined

as:

f(i)

= p

hi i+

1+

psi

i–

ph

i i–

psi

i-1

Cis

-pro

lin

esar

e an

om

alo

us,

as

exp

ecte

d. T

his

R

amac

han

dra

nfu

nct

ion

co

uld

be

use

d a

s an

al

tern

ativ

e m

easu

re o

f st

raig

htn

ess

if a

cces

s to

q

uat

ern

ion

in

form

atio

n i

s li

mit

ed.

Page 2: in the Analysis of Protein and Nucleic Acid Secondary ...

Usi

ng

Qu

ate

rnio

ns

and

Mo

lecu

lar

Dy

na

mic

s S

imu

lati

on

s in

Dru

g D

esU

sin

g Q

uat

ern

ion

s an

d M

ole

cula

r D

yn

am

ics

Sim

ula

tio

ns

in D

rug

Des

ign

ign

Sea

n B

. Jo

hn

sto

n a

nd

Ro

ber

t M

. Ha

nso

nS

t. O

laf

Co

lleg

e, N

ort

hfi

eld

, M

inn

eso

ta23

7thN

ati

on

al

Mee

tin

g o

f th

e A

mer

ica

n C

hem

ica

l S

oci

ety

, Sa

lt L

ak

e C

ity

, Uta

h, M

arc

h 2

4, 2

009

•Q

uat

ern

ion

s ar

e an

ex

ten

sio

n o

f th

e co

mp

lex

nu

mb

ers

(wh

ere

i2= j

2= k

2= i

jk=

-1).

2

•U

nit

qu

ater

nio

ns

(qu

ater

nio

ns

of

len

gth

on

e) h

ave

bee

n i

nv

esti

gat

ed f

or

thei

r u

niq

ue

app

lica

tio

n t

o t

he

mea

sure

men

t an

d v

isu

aliz

atio

n o

f o

rien

tati

on

s.3

•A

ny

fra

me

of

refe

ren

ce (

incl

ud

ing

th

ose

rep

rese

nti

ng

mo

lecu

lar

ori

enta

tio

ns)

can

be

rep

rese

nte

d b

y a

un

it q

uat

ern

ion

.

•U

nit

qu

ater

nio

n d

iffe

ren

ces

(or

qu

ater

nio

n d

eriv

ativ

es)

rep

rese

nt

rota

tio

ns,

an

d c

an b

e ex

pre

ssed

in

ter

ms

of

an a

bso

lute

fra

me

of

refe

ren

ce (

q2q

1-1

) o

r a

rela

tiv

e fr

ame

of

the

rota

tin

g o

bje

ct (

q1

-1q

2)

.

Mo

lecu

lar

dy

nam

ics

sim

ula

tio

ns

are

rou

tin

ely

car

ried

ou

t w

ith

bio

log

ical

mac

rom

ole

cule

s to

m

od

el t

he

way

th

ey i

nte

ract

wit

h s

olv

ent.

1F

rom

su

ch s

imu

lati

on

s, a

pp

rox

imat

e b

ind

ing

fre

e en

erg

y o

f a

pro

tein

to

a l

igan

dca

n b

e ca

lcu

late

d.

Mu

ch o

f m

edic

inal

ch

emis

try

is

con

cern

ed

wit

h i

nd

ivid

ual

in

tera

ctio

ns

bet

wee

n s

pec

ific

am

ino

aci

ds

and

a d

rug

, h

ow

ever

. T

hro

ug

h a

m

ole

cula

r d

yn

amic

s tr

ajec

tory

, an

am

ino

aci

d p

arti

cip

atin

g i

n b

ind

ing

a l

igan

dm

ay h

ave

dec

reas

ed f

lex

ibil

ity

. U

nit

qu

ater

nio

ns

off

er a

un

iqu

e w

ay t

o q

uan

tify

th

is f

lexi

bil

ity

. W

e es

tab

lish

qu

ater

nio

ns

for

resi

du

es i

n H

IV-1

pro

teas

e, a

nd

ex

amin

e th

e sa

me

resi

du

es w

hen

th

e en

zym

e is

bo

un

d t

o t

he

inh

ibit

or

lop

inav

ir(L

PV

).

Qu

ater

nio

ns

are

des

ign

ated

by

a c

oo

rdin

ate

fram

e u

sin

g t

hre

e at

om

s.

•T

he

firs

t at

om

is

the

cen

ter

(in

th

is e

xam

ple

, th

e al

ph

a-ca

rbo

n).

•T

he

seco

nd

ato

m (

carb

oxy

l ca

rbo

n)

giv

es t

he

dir

ecti

on

of

the

x-a

xis

(re

d

arro

w).

•T

he

y-a

xis

(g

reen

arr

ow

) is

per

pen

dic

ula

r to

th

e x

-ax

is a

nd

in

th

e p

lan

e fo

rmed

by

th

e al

ph

a ca

rbo

n, c

arb

on

yl

carb

on

, an

d b

ack

bo

ne

nit

rog

en.

•T

he

z-ax

is i

s p

erp

end

icu

lar

to b

oth

x a

nd

y a

xes

, an

d f

ou

nd

usi

ng

th

e ri

gh

t-h

and

-ru

le.

Ab

stra

ctA

bst

ract

Qu

ater

nio

ns

in 3

DQ

uat

ern

ion

s in

3D

Qu

ater

nio

n M

ap

sQ

uat

ern

ion

Ma

ps

•U

nit

qu

ater

nio

ns

can

be

dis

pla

yed

in

a g

rap

hic

al m

ann

er.3

(Fig

. 1)

•Q

uat

ern

ion

der

ivat

ives

(re

pre

sen

tin

g t

he

rota

tio

n r

equ

ired

to

pro

ceed

fro

m o

ne

ori

enta

tio

n t

o

ano

ther

) ca

n b

e d

isp

lay

ed a

s w

ell.

•B

oth

ver

sio

ns

giv

e in

form

atio

n a

bo

ut

the

stru

ctu

re.

•H

IV-1

pro

teas

e is

th

e su

bje

ct o

f m

uch

res

earc

h a

nd

sim

ula

tio

n.4

-6

•V

ario

us

inh

ibit

ors

hav

e b

een

stu

die

d a

nd

sim

ula

ted

.

•T

he

enzy

me

affo

rds

an o

pp

ort

un

ity

to

see

if

add

itio

n o

f lo

pin

avir

affe

cts

the

flex

ibil

ity

of

amin

o a

cid

s.

•C

on

tin

uin

g r

esea

rch

co

uld

co

mp

are

inh

ibit

ors

fo

r w

hic

h b

ind

ing

fre

e en

erg

y d

ata,

bo

th

com

pu

tati

on

al a

nd

ex

per

imen

tal,

is

kn

ow

n.

HIV

HIV

-- 1 P

rote

ase

1 P

rote

ase

Qu

ater

nio

n d

esig

nat

ion

sQ

uat

ern

ion

des

ign

atio

ns

Qu

ater

nio

n d

esig

nat

ion

s ar

e se

nsi

tiv

e to

glo

bal

pro

tein

ro

tati

on

or

"tu

mb

lin

g"

du

rin

g t

he

mo

lecu

lar

dy

nam

ics

calc

ula

tio

n.

Tu

mb

lin

g i

s o

bse

rvab

le i

n t

he

qu

ater

nio

n m

ap, w

her

e a

syst

emat

ic s

pre

adin

g

of

qu

ater

nio

ns

wit

hin

th

e m

ap i

nd

icat

es t

um

bli

ng

of

the

pro

tein

. T

his

is

seen

in

th

e fi

rst

two

q

uat

ern

ion

map

s in

Fig

ure

2.

Th

e re

lati

ve

qu

ater

nio

n d

iffe

ren

ce o

f tw

o q

uat

ern

ion

s (F

ig. 2

c) r

emo

ves

th

is a

rtif

act

of

the

calc

ula

tio

n.

Sim

ilar

ly, t

he

rela

tiv

e d

iffe

ren

ce b

etw

een

tw

o r

efer

ence

fra

mes

lo

cate

d

on

a s

ing

le r

esid

ue

can

rem

ov

e th

e ef

fect

of

tum

bli

ng

.

Th

e T

um

bli

ng

Pro

ble

mT

he

Tu

mb

lin

g P

rob

lem

Sam

ple

qu

ater

nio

n m

aps

Sam

ple

qu

ater

nio

n m

aps

In p

ure

wat

er,

thre

e si

mu

lati

on

s u

sin

g t

he

AM

BE

R m

ole

cula

r d

yn

amic

s su

ite7

wer

e ca

rrie

d o

ut,

o

ne

wit

h t

he

A c

on

form

atio

n o

f lo

pin

avir

, on

e w

ith

th

e B

co

nfo

rmat

ion

, an

d o

ne

un

bo

un

d:

•ea

ch s

olu

te w

as s

olv

ated

exp

lici

tly

usi

ng

TIP

3P w

ater

s

•th

e i

nit

ial

stru

ctu

re’s

en

erg

y w

as m

inim

ized

, th

e sy

stem

was

hea

ted

fro

m 0

K t

o 3

00K

fo

r 50

p

s, t

he

syst

em e

qu

ilib

rate

d a

t co

nst

ant

pre

ssu

re f

or

50 p

s, a

nd

a 5

00 p

seq

uil

ibra

tio

n w

as r

un

to

ch

eck

fo

r tr

end

s in

tem

per

atu

re, d

ensi

ty,

and

to

tal

ener

gy

to

ensu

re s

yst

em s

tab

ilit

y a

nd

b

ack

bo

ne

roo

t-m

ean

-sq

uar

e d

evia

tio

n t

o c

hec

k s

tru

ctu

re c

han

ges

Th

e sy

stem

s w

ere

then

sim

ula

ted

fo

r ~1

.5 n

s, o

utp

utt

ing

str

uct

ure

dat

a ev

ery

10

ps.

Usi

ng

Jm

ol,

8q

uat

ern

ion

s an

d q

uat

ern

ion

dif

fere

nce

s w

ere

calc

ula

ted

fo

r ea

ch s

nap

sho

t in

th

e tr

ajec

tory

. A

spar

tate

resi

du

es 2

5, 2

9, a

nd

60

on

bo

th c

hai

ns

wer

e st

ud

ied

pri

mar

ily

.

•“a

sp”

fram

es a

re d

esig

nat

ed i

n t

he

man

ner

des

crib

ed i

n p

ost

er B

IOL

110

(D

an K

oh

ler)

.

•“a

spO

”fr

ames

are

bas

ed o

n t

he

pla

ne

of

the

carb

ox

yl

gro

up

.

•“b

ase”

qu

ater

nio

ns

serv

e as

a r

efer

ence

po

int

for

com

par

ing

bo

un

d a

nd

un

bo

un

d p

rote

ase.

•A

sp25

is

the

cata

lyti

c re

sid

ue,

asp

29 i

s an

im

po

rtan

t b

ind

ing

res

idu

e, a

nd

asp

60 i

s o

n t

he

ou

tsid

e o

f th

e en

zym

e, s

erv

ing

as

a co

ntr

ol.

Met

ho

d d

eta

ils

Met

ho

d d

eta

ils

Tes

tin

g b

ou

nd

/un

bo

un

d h

yp

oth

esi

sT

esti

ng

bo

un

d/u

nb

ou

nd

hy

po

the

sis

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

124-base

124O-base

25-base

25O-base

60-base

60O-base

29-base

29O-base

128-base

128O-base

quaternion difference label

unbound

A conform

er

B conform

er

Exa

min

ing

E

xam

inin

g l

igan

dli

gan

db

ind

ing

bin

din

g

Th

e m

easu

rem

ent

and

vis

ual

izat

ion

of

qu

ater

nio

n d

iffe

ren

ces

len

ds

insi

gh

t in

to t

he

ran

ge

of

mo

tio

n a

nd

fle

xib

ilit

y o

f p

rote

in r

esid

ues

an

d b

ou

nd

lig

and

sd

uri

ng

a m

ole

cula

r d

yn

amic

s ca

lcu

lati

on

. Jm

ol

11.7

in

clu

des

a v

arie

ty o

f q

uat

ern

ion

-bas

ed f

un

ctio

nal

ity

and

all

ow

s a

bro

ad

ran

ge

of

inv

esti

gat

ion

s u

sin

g q

uat

ern

ion

s in

way

s th

at h

ave

no

t b

een

av

aila

ble

pre

vio

usl

y.

Co

ncl

usi

on

sC

on

clu

sio

ns

I w

ou

ld l

ike

to a

ckn

ow

led

ge

the

Un

iver

sity

of

Min

nes

ota

Su

per

com

pu

tin

g I

nst

itu

te f

or

the

use

of

the

AM

BE

R s

oft

wa

re s

uit

e o

n C

alh

ou

n.

Th

an

ks

to

Dr.

Jef

f S

chw

inef

us

for

intr

od

uci

ng

me

to s

uch

an

in

tere

stin

g a

rea

. Th

is w

ork

wa

s su

pp

ort

ed b

y a

St.

O

laf

Co

lleg

e Magnus the Good c

oll

ab

ora

tiv

e re

sea

rch

g

ran

t.

Ack

no

wle

dg

em

en

tsA

ckn

ow

led

ge

me

nts

1.

Pat

rick

, G

. L

. A

n i

ntr

odu

ctio

n t

o m

edic

ina

l ch

emis

try

, 3

rd e

d.;

Ox

ford

Un

iver

sity

Pre

ss:

Ox

ford

; N

ew Y

ork

, 2

005

.2

. N

ich

ols

on

, W

. K

. In

trod

uct

ion

to

abs

trac

t a

lgeb

ra,

3rd

ed

.; W

iley

-In

ters

cien

ce:

Ho

bo

ke

n,

N.J

., 2

007

.3

. H

anso

n,

A.

Vis

ua

lizi

ng

qu

ate

rnio

ns;

Mo

rgan

Ka

ufm

ann

: S

an

Fra

nci

sco

, C

A,

20

06

.4

. K

oy

ano

, K

.; N

akan

o,

T. J

ourn

al

of S

yn

chro

tron

Rad

iati

on20

08, 1

5, 2

39-2

42.

5.

Wit

tay

anar

aku

l, K

.; H

ann

on

gb

ua,

S.;

Fei

g,

M.

Jou

rna

l of

Com

pu

tati

ona

l Ch

emis

try

2008

, 29

, 673

-68

5.6

. P

uro

hit

, R

.; R

ajas

ekar

an,

R.;

Su

dan

dir

ad

oss

, C

.; D

oss

, C

. G

. P

.; R

aman

ath

an,

K.;

Rao

, S

. In

tern

ati

ona

l Jo

urn

al

of B

iolo

gic

al M

acr

omol

ecu

les

20

08,

42, 3

86-3

91.

7.

Cas

e, D

. A

.; D

ard

en,

T.

A.;

T.E

. C

he

ath

am,

I.;

Sim

mer

lin

g,

C.

L.;

Wan

g,

J.;

Du

ke,

R.

E.;

Lu

o,

R.;

Mer

z,

K.

M.;

Pea

rlm

an,

D.

A.;

Cro

wle

y,

M.;

Wal

ke

r, R

. C

.; Z

han

g,

W.;

Wa

ng

, B

.; H

ayik

, S

.; R

oit

ber

g,

A.;

Sea

bra

, G

.; W

on

g,

K.

F.;

Pae

san

i, F

.; W

u,

X.;

Bro

zel

l, S

.; T

sui,

V.;

G

oh

lke,

H.;

Yan

g,

L.;

Tan

, C

.; M

on

gan

, J.

; H

orn

ak

, V

.; C

ui,

G.;

Ber

oz

a, P

.; M

ath

ews,

D.

H.;

Sch

afm

eist

er,

C.;

Ro

ss,

W.

S.;

P.A

. K

oll

ma

n(2

00

6)

AM

BE

R 9

, U

niv

ersi

ty o

f C

ali

forn

ia,

San

Fra

nci

sco

.8

. J

mo

l. h

ttp

://J

mo

l.so

urc

efo

rge.

net

Wo

rks

Cit

edW

ork

s C

ited

Fig

ure

1.

a. A

n i

dea

lize

d h

elix

. b

. R

ota

tin

g t

o l

oo

k d

ow

n t

he

hel

ix.

c.

Sw

itch

ing

to

a q

uat

ern

ion

map

dir

ectl

y f

rom

b.

(no

tice

th

e ax

es a

re t

he

sam

e),

we

see

that

th

e q

uat

ern

ion

s fo

rma

per

fect

cir

cle.

d.

In t

he

qu

ater

nio

n d

eriv

ativ

e m

ap,

all

po

ints

map

to

th

e ti

p o

f th

e ar

row

bec

au

se e

ach

dif

fere

nce

fro

m o

ne

fram

e to

th

e n

ext

is t

he

sam

e.

Vis

ual

izat

ion

s w

ere

ma

de

usi

ng

th

e Jm

ol

sig

ned

ap

ple

t.8

a.b

.c.

d.

Fig

ure

2.

a. Q

uat

ern

ion

map

fo

r th

e p

lan

e o

f th

e ca

rbo

ny

l o

f as

p12

4. b

. T

he

qu

ater

nio

n m

ap f

or

the

hy

dro

xy

gro

up

of

lop

inav

ir.

c. T

he

rela

tiv

e q

uat

ern

ion

dif

fere

nce

bet

wee

n t

he

qu

ater

nio

ns.

d.

Th

e fr

ames

rep

rese

nti

ng

th

e q

uat

ern

ion

s st

ud

ied

in

th

is s

ecti

on

. V

isu

aliz

atio

ns

wer

e m

ade

usi

ng

th

e Jm

ol

sig

ned

ap

ple

t.8

a.b

.c.

d.

Fig

ure

3.

a. Q

uat

ern

ion

dif

fere

nce

map

fo

r as

p12

8O a

nd

asp

128

bas

e. b

Q

uat

ern

ion

dif

fere

nce

map

fo

r as

p29

O a

nd

an

un

rela

ted

qu

ater

nio

n o

n l

op

inav

ir.

c. Q

uat

ern

ion

dif

fere

nce

map

fo

r as

p60

on

th

e o

uts

ide

of

the

pro

tein

an

d t

he

hy

dro

xy

gro

up

of

lop

inav

ir.

Vis

ual

izat

ion

s w

ere

ma

de

usi

ng

th

e Jm

ol

sig

ned

ap

ple

t.8

Fig

ure

4.

Rel

ativ

e q

uat

ern

ion

dif

fere

nce

s w

ith

in s

pec

ific

res

idu

es g

ive

an

ind

icat

ion

of

the

ran

ge

of

flex

ibil

ity

of

the

resi

du

e si

de

chai

n d

uri

ng

th

e si

mu

lati

on

. N

ote

th

at b

ind

ing

of

lop

ina

vir

is n

ot

nec

essa

rily

a p

red

icto

r o

f re

sid

ue

flex

ibil

ity

, ev

en i

n t

he

acti

ve

site

. T

he

larg

e d

evia

tio

ns

in b

oth

ou

r co

ntr

ol

(asp

60)

and

th

eb

ind

ing

sit

e re

sid

ue

asp

29

are

evid

ence

th

at t

hes

e tw

o c

arb

ox

yla

teg

rou

ps

flip

ped

ori

enta

tio

n d

uri

ng

th

e si

mu

lati

on

.

Fig

ure

5.

Co

mp

aris

on

s o

f q

uat

ern

ion

dif

fere

nce

s fo

r a

var

iety

of

inte

ract

ing

an

d n

on

-in

tera

ctin

g l

op

inav

ir/p

rote

in

inte

ract

ion

s. O

ver

all,

it

seem

s th

at t

he

B c

on

form

atio

n i

s sl

igh

tly

mo

re “

sett

led

.”A

s in

Fig

ure

4,

thes

e re

sult

s in

dic

ate

that

in

th

e A

co

nfo

rmat

ion

sim

ula

tio

n a

sp60

an

d a

sp29

fli

pp

ed.

(No

te t

hat

on

ly q

uat

ern

ion

s b

ased

on

ca

rbo

ny

l o

xy

gen

ssh

ow

th

is e

ffec

t.)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

LPV1-asp25

LPV1-asp25O

LPV1-asp124

LPV1-asp124O

LPV1-asp60

LPV1-asp60O

LPV2-asp25

LPV2-asp25O

LPV2-asp124

LPV2-asp124O

LPV2-asp60

LPV2-asp60O

LPV3-asp29

LPV3-asp29O

LPV4-asp128

LPV4-asp128O

average deviation

A conform

ation

B conform

ation