Title
Monomeriz atio n of Far-Red Fluo rescen t Proteins
Authors
Timothy M. Wan nier
1,2,*
, Sar ah Gill espie
1
, Nicholas Hu tchins
1
, R. Sco tt Mc Isaac
3
, K evin S. Brown
4
,
Steph en L. Mayo
1,*
Affiliations
1.
Division
of
Biology
and
Biol ogical
Engineering,
Mail
Code
114-96,
Californ ia
Insti tut e
o f
Technology, Pasadena, Californi a, Unit ed States of Ame rica
2. Present addr ess: Harvard Me dical School, Departm ent of Gen etics. 77 Avenue Louis Pasteur,
New Resea rch Building 238, Bos ton, MA 02115
3.
Division
of
Chemistry
and
Chemical
Engineering,
Mail
Code
210-41,
Califor nia
Inst itu te
of
Technology, Pasadena, Californi a, Unit ed States of Ame rica
4. Depar tments of Biomedical Engineeri ng, Physics, Chemical and Biomedical E ngineering, an d
Marine Scie nces, Universi ty of Connectic ut, Sto rrs, Connec ticut, Uni ted S ta tes of America
* Correspondi ng autho rs:
timo thy_wanni er@hms.harvar d.ed u
, steve@mayo.cal te ch.edu
Abstract
Anthozoa
class red fluorescen t p rot eins (RFPs) are frequ ently us ed as biologica l markers, with
far-red
emit ting
variants
(λ
em
~
600
–
900
nm)
sought
for
whole
animal
imaging
because
biological tissues are p erme able to light in this rang e. A barri er to t he use of nat urally occurrin g
RFP variants as molecular markers is tha t all are te tram eric, which is not ideal for cell biological
applications .
Efforts
t o
engin eer
mo no meric
RFPs
have
usually
pr oduced
dim mer
and
blu e-
peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not
.
http://dx.doi.org/10.1101/162842
doi:
bioRxiv preprint first posted online Jul. 20, 2017;
shifted
variants ,
as
the
chr omophor e
i s
sensitive
to
small
struc tural
p er turb ations.
In
fact,
despite much effor t, only four na tive RFP s have been successfully monomeriz ed, l eaving the vast
majority
of
RFP
biodiversity
untappe d
in
biomarker
development.
He re
we
report
th e
genera tion of monomeric variants of HcRed and mCardinal, both far-red dime rs, and describe a
comprehensive
me thodol ogy
for
the
r apid
monomeriz ation
of
novel
r ed-shifted
oligome ric
RFPs. Among the resultan t variants , is mKelly1 (emission maximum: λ
em
= 656 nm), which along
with the r ecently r epor ted mG arne t2 (1), forms a new class of bright, monomeric, far-red FPs.
Introduction
The
developmen t
of
red
fluor escen t
pr oteins
(RFPs)
as
tags
for
molecular
im aging
has
long
focused on monomerizati on, incre ased b rightness, and pushing exci tati on and e mission to ever-
longer wavelengths. These tr aits ar e desirable for live animal imaging, as far-red to near infrare d
light pen etr at es tissu e with minimal abs orption in wha t is known as th e n ear i n frared window
(~625–1300
nm)
(2,
3).
Monomericity
is
important
becaus e
oligomeriza tion
of
an
FP
tag
can
artificially aggregat e its linked pro tein ta rget, al teri ng diffusion rates and int erfer ing with targe t
transpo rt, t rafficking, and activity (4, 5). Recently a new class of infrared fluorescent pro teins
(iRFPs)
was
developed
from
t he
bac te rial
phytochr ome
(6),
but
th ese
r equir e
the
coval en t
linkage of a small molecule chromopho re, biliverdin , limiting thei r use to cells and organisms
that make this molecul e in sufficien t qua ntity.
Anthozoa
class RFPs (such as mChe rry and mKa te)
have the advantag e tha t th e chromoph o re is crea ted via a s elf-processing re actio n, necessit ating
only molecular O
2
for chromopho re form ation .
To our knowledge, ~50 native RFPs and ~40 chromoprot eins (CPs) with peak absorbance in th e
red or far-red (absorbanc e maximum: λ
abs
>550 nm) have been described to da te, but most have
not
been
e xte nsively
characteri zed
bec ause
they
are
as
a
class
tetrame ric,
an d
thus
are
less
peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not
.
http://dx.doi.org/10.1101/162842
doi:
bioRxiv preprint first posted online Jul. 20, 2017;
useful
as
biological
markers
(7,
8).
An
underlying
biological
reason
for
the
obligat e
tet rameri zati on
of
native
RFPs
has
been
hint ed
at ,
but
is
not
well
unde rstood
(9-12).
Oligomeriza tion
do es
seem
t o
play
an
importa nt
st ructur al
role ,
howeve r,
as
bre aking
tet rameri zati on
without
abr ogating
fluorescence
has
proved
difficult,
a nd
successful
monomeriza tion h as always led to ei the r a hypsochromic shift to λ
em
or a decr eas e in brigh tness
(13-16).
Previous
efforts
to
monome rize
n ative
R FP
te tram ers
have
reli e d
on
le ngthy
engineer ing
traject ories,
an d
have
bee n
successful
in
only
four
cases
(Table
1).
Generally ,
mutations
a re
first
int roduced
in to
tigh t
interfac es
to
weaken
oligome riza tion,
an
inefficient
process tha t impairs fluor escence, and t hen random mu tagenesis a nd screeni ng isolate p arti ally
recovere d varian ts. Aft er many such cycles monomeric varian ts have be en fou nd, but pr ot ein
core an d chromop hore-p roximal muta ti ons ar e invaria bly intr oduced, making it difficult to exe rt
any significant degr ee of contr ol ove r th e fluoresc ent p rop erti es of t he result ant monomer . I t i s
thus difficult t o know whe ther th e poo r spectrosco pic charac teris tics of engin ee red monom ers
are an un avoidable cons equenc e of monomeriza tion or only the manifes tati on of a suboptimal
evolutiona ry path .
Here we pres ent a compr ehensive engi neering str at egy for the monomeriz atio n of novel RFPs
that
differe ntia tes
its elf
by
treating
s epara tely
the
p roblems
of
prot ein
sta bilization ,
core
optimiza tion,
and
surface
d esign.
We
sample
mutati onal
space
bo th
stochas tically,
throug h
erro r-prone
mut agenesis,
and
ra tiona lly,
by
analysis
of
multiple
sequence
align ments
(MSAs)
and
comput ation al
pro tein
design
(CPD).
Two
far-red
oligomeric
pro teins
we re
t arge ted
fo r
monomeriza tion: HcR ed (λ
em
= 633 nm), a dimer/te trame r (17), and mCardinal (λ
em
= 658 nm), a
repor ted mon omer that w e have confir med to in fact b e dimeric . The monom er ic RFPs report ed
here includ e two monomeric HcRed va riants: mGi nger1 (λ
em
= 637 nm) and mGinger2 (λ
em
=
peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not
.
http://dx.doi.org/10.1101/162842
doi:
bioRxiv preprint first posted online Jul. 20, 2017;
631), and two monomeric mCardi nal va riants: mKelly1 (λ
em
= 648 nm) and mKe lly2 (λ
em
= 643
nm), which are among the bright est far-r ed monomeric FPs to have be en re por te d.
Results
Step-wise
monomerization
of
HcRed.
We
first
chose
HcRed,
a
far-red
FP
that
has
bee n
engineer ed bu t neve r successfully mono merized (17, 18). As w e have pr eviously demonstr at ed
that oligomerici ty and b rightness ca n b e tr ea ted as sepa rat e pro tein design p r oblems (19), we
devised
a
workflow
that
se para tely
t a rgets
th e
chromopho re
e nvironmen t— to
engine er
a
prote in core tha t maint ains st ructur al i ntegri ty absent stabilizi ng oligomeric in t eracti ons—and
the
prot ein
surface— to
drive
monomeriza tion.
Anthozoa
class
RFPs
have
two
oligomeric
interfaces , name d AB and AC (20), with the AC inte rface b eing th e mor e st able of the two an d
burying
a
large
hydrophobic
surface
(21).
Early
engineering
to
HcRed
part ially
disrupted
oligomeriza tion a t th e AB in terfac e, but all mutati ons to the AC interface w ere f ound to viti at e
fluorescence . To t est t he in tegrity of the AC interface , we made succ essive dele ti ons to HcRe d’s
C-terminal t ail (residu es 219-227), which plays an in tegr al rol e in the AC in ter acti on (Figure 1A) .
HcRed lost significant bright ness with th e dele tion of just one C-terminal residue , and was non-
fluorescent
after
a ny
further
d ele tion,
demonstr ating
th at
op timizati on
would
be
necessary
prior to mon omeriz ation .
So, we end eavored to e nginee r a mor e stable co re, id entifying two muta tion al h otspots from a n
alignment of fa r-red R FPs: (A) a group o f residues tha t surr ounds al te rnative co nformations of
the
chromoph ore ’s
phenola te
ring
and
(B)
a
region
above
the
plane
of
th e
chromophor e ,
betwee n th e cent ral α-helix and the un broken AC oligome ric int erface (Figure S 1). Gene rally i n
RFPs,
the
cis
chromophore —th e
phen olate
ring
sits
cis
to
the
pro ximal
nit rogen
on
the
imidazolinone ring ra the r than
trans
to i t—is the fluor escent speci es (22). In engineering HcRe d
peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not
.
http://dx.doi.org/10.1101/162842
doi:
bioRxiv preprint first posted online Jul. 20, 2017;
from
its
chromopr otei n
par ent
HcCP,
the
cis
chromophor e
was
sta bilized
over
th e
non-
fluorescent
trans
chr omophor e by way of a cysteine to seri ne muta tion a t posit ion 143, which
provides a hydr ogen bon d to th e
cis
ph enolat e o xygen (Figure 1C). W e re ason ed tha t fur the r
stabiliza tion
of
the
cis
chr omophor e
w ould
incre ase
b rightness ,
and
so
desig ned
a
firs t
cor e
library (cLibA) to target hotspo t A, mutating
trans
-sta bilizing amino acids, placing bulkier side
chains
into
the
trans
pocket ,
and
allo wing
varied
hydrogen
bonding
geomet ries
to
th e
cis
chromophor e. A s econd cor e libra ry (cLibB) targe ted ho tspo t B along with two chromophor e-
backing positions (Gly28 and Met41 are implicated in maturation and color) (21, 23, 24). Two
key
features
of
this
hotspo t
ar e
a
channel
popula ted
by
struc tural
wat er
molecules
tha t
stre tches t o t he p rot ein surface , and Arg 67, a key ca talytic r esidue . Mu ta tions to this regio n may
serve to occlude access to the chromop h ore by bulk solvent upon monomerizati o n, and to allow
room for chr omophor e pr ocessing. Small librari es of < 1 ,000 pro tein varian ts wer e guided by th e
far-red RFP alignment (Table S1), and after screening each library to > 95% coverage on large LB
agar
plates
supplem ent ed
with
IPTG,
we
fully
characteri zed
16
cLibA
variant s
and
21
cLibB
variants. Th e varian ts showed brigh tness increases of up to t en-fold and displ aye d an incredibl e
range
of
emission
profiles,
with
λ
em
between
606
and
647
nm.
To
determin e
which
if
any
variants
would
be
amenable
to
mono merizati on,
we
tested
a
five-residue
tai l
deletion .
Eight
variants showed de tec table fluo rescenc e after the t ail dele tion, wit h a double mutant (HcRed7 :
R67K/I196Y)
being
the
most
red-shifted
(λ
em
=
642
nm).
The
core
mutati ons
in
HcRed7
bathochr omically
shift
its
emission
by
9
nm,
improve
its
quantum
yield
(Φ)
by
60%,
and
thermos tabiliz e
the
pro tein
by
6
°C.
HcRed7,
however,
loses
significant
brigh tness
with
the
deleti on of a six th tail residu e – HcR ed7 ∆6) and becomes 16°C less the rmostabl e , indicating th at
the pro tein is no t wholly optimized for m onomeriza tion (Table 2).
peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not
.
http://dx.doi.org/10.1101/162842
doi:
bioRxiv preprint first posted online Jul. 20, 2017;
To
further
op timize
HcR ed7∆6
for
mo nomeriza tion,
we
took
aim
a t
improvi ng
the
th ermo-
stability of the pro tein Thermos tabili ty has been shown to increas e a prot ein’s e volvability (25)
and
consensus
design
is
one
of
the
best
tools
for
improving
thermost a bility
(26).
We
construct ed
a
large
MSA
that
consists
of
every
Aequorea victoria
class
FP;
a
total
of
741
sequences
(se e
supplem ental
Me thod s),
and
th en
buil t
a
lib rary
to
sampl e
all
105
n on-
consensus positions in HcRed with the c onsensus amino acid, and compared t his to a strat egy of
erro r-prone mu tagen esis. W e scre ened the conse nsus (~1.2 muta tions p er vari ant) and e rror-
prone (~1 .8 muta tions pe r varian t) librar ies at 675 nm t o allow maxim al differen t iation b etwe en
far-red variants whose λ
em
was between 630-640 nm and a large population of ne ar-red varian ts
whose emission peaked betw een 605-620 nm, but which were often brighter . The consensus
library was screene d to 40x cover age (~4,300 clones) and ~8,600 clones were scr e ened from th e
erro r-prone
libra ry.
Consensus
libra ry
variants
significantly
outp erformed
er r or-prone
libra ry
variants (Figure S2), and so w e combin ed seven of the top c onsensus vari ants togeth er in to a
chimeric prot ein, HcRe d77, which recov ered all of HcRed7∆6’s los t bright ness a nd much of its
thermos tabili ty.
Finally, to monomerize HcRe d77 we tar geted th e AC interface with a CPD procedure th at we
describe
in
previous
work
(19).
We
focused
on
a
set
of
five
hydrophobic
re sidues
(Val146,
Val159,
Ile170,
Phe191 ,
and
Phe193)
a t
th e
hea rt
of
the
AC
inte rface
t hat
make
ex tensiv e
intermol ecular c ontac ts (Figure S3), a nd built a small combinat orial li brary guid ed by the d esign.
We
isolate d
a
first-generati on
monomer:
HcRedm1
and
verified
it
to
be
monomeric
by
fast
prote in liquid chromat ography (FPLC) and analytical ultr acent rifugation (AUC) (Figure 2), but the
prote in
was
dim
and
e xpress ed
poo rly.
We
a tt ribut ed
th ese
po or
a ttri butes
to
incomple t e
thermo-st abiliza tion of HcRed77, and so we screened mut ations from th e err or-p rone libra ry via
DNA shuffling and then increased the t empera tur e of screening from 30°C to 37°C for a final
peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not
.
http://dx.doi.org/10.1101/162842
doi:
bioRxiv preprint first posted online Jul. 20, 2017;