of 20
1
Supporting Information
4S
-Hydroxylation of insulin at ProB28 accelerates hex
amer dissociation and delays fibrillation
Seth A. Lieblich
†,‡
, Katharine Y. Fang
†,‡
, Jackson K. B. Cahn
, Jeffrey Rawson
§,||
, Jeanne
LeBon
§
, H. Teresa Ku
§,||,¶
, David A. Tirrell
†,*
†Division of Chemistry and Chemical Engineering, Ca
lifornia Institute of Technology, Pasadena, CA 9112
5, USA.
§
Department of Translational Research and Cellular T
herapeutics, Diabetes and Metabolism Research Insti
tute, City
of Hope, Duarte, CA 91010, USA.
||
Beckman Research Institute of City of Hope, Duarte,
CA 91010, USA.
Irell & Manella Graduate School of Biological Scien
ces, City of Hope, Duarte, CA 91010, USA.
*Corresponding author. Email:
tirrell@caltech.edu
Table of Contents
Materials and Methods: ............................
...................................................
..................................................
2
Figure S1. Insulin expression and incorporation of
hydroxyprolines. ..................................
........................ 8
Figure S2. Immunoblot detection of insulin receptor
activation. ......................................
............................ 9
Figure S3. Example fits from sedimentation analysis
. .................................................
............................... 10
Figure S4. Example fits for analysis of dissociatio
n kinetics. .......................................
............................. 11
Figure S5. Alignment at position B28. .............
...................................................
....................................... 12
Figure S6. Alignment of R6-AspI and T2-HzpI at posi
tion B28. .........................................
...................... 13
Figure S7. Independent measurement of fibrillation
lag time. .........................................
........................... 14
Figure S8 Transmission electron micrographs of insu
lin fibrils. ......................................
.......................... 15
Figure S9. Biophysical characterization of ProI sam
ples prepared in-house and obtained from a
commercial source. ................................
...................................................
..................................................
16
Table S1. Expression Yields and Incorporation Level
s of Hydroxyinsulins. .............................
................ 17
Table S2. Data Tables and Refinement Values. ......
...................................................
................................. 18
References: .......................................
...................................................
...................................................
..... 20
2
Materials and Methods:
Materials
. All canonical amino acids and (4
R
)hydroxyLproline (Hyp) were purchased
from Sigma.
(
4
S)
hydroxyLproline (Hzp) was purchased from Bachem
Americas. All
solutions and buffers were made using doubledistil
led water (ddH
2
O).
Strains and plasmids.
The proinsulin (PI) gene with an
N
terminal hexahistidine tag
(6xHIS), and flanked by
Eco
R1
and
Bam
H1 cut sites was ordered as a gBlock
(Integrated DNA Technologies). Both the gBlock and
vector pQE80L for IPTGinducible
expression were digested with
Eco
RI
and
Bam
HI. Linearized vector pQE80L was
dephosphorylated by alkaline phosphatase (NEB). Li
gation of the digested PI gene and
linearized vector yielded plasmid pQE80PI (to produ
ce ProI). To make plasmid
pQE80PIproS (to produce HzpI and HypI): Genomic DN
A was extracted from
E. coli
strain DH10β using DNeasy Blood and Tissue Kit (Qia
gen). Primers (Integrated DNA
Technologies) were designed to amplify the
E. coli proS
gene, encoding prolyltRNA
synthetase, under constitutive control of its endog
enous promoter, from purified
genomic DNA, and to append
Nhe
I and
Nco
I sites. The digested
proS
gene was then
inserted into pQE80PI between transcription termina
tion sites by ligation at
Nhe
I and
Nco
I restriction sites. Prolineauxotrophic
E. coli
strain CAG18515 was obtained from
the Coli Genetic Stock Center at Yale University.
Prototrophic
E. coli
strain BL21 was
used for rich media expression of canonical insulin
s (ProI, AspI). Sitedirected
mutagenesis of pQE80PI at B28 was performed to make
plasmid pQE80PIasp, which
differs from pQE80PI by three nucleotides that spec
ify a single amino acid mutation to
aspartic acid. All genes and plasmids were confirme
d by DNA sequencing.
Protein expression.
Plasmids pQE80PI and pQE80PIasp were transformed i
nto BL21
cells and grown on ampicillinselective agar plates
. A single colony was used to
inoculate 5 mL of LuriaBertani (LB) medium and gro
wn overnight; the resulting
saturated culture was used to inoculate another 1 L
of LB medium. All expression
experiments were conducted at 37°C, 200 RPM in shak
e flasks (Fernbach 2.8 L flasks,
Pyrex®). Each culture was induced with 1 mM IPTG at
midexponential phase (OD
600
~0.8). For incorporation of Hyp and Hzp, pQE80PIp
roS was transformed into
CAG18515 cells, which were grown on ampicillinsele
ctive agar plates. To facilitate
growth, a single colony was used to inoculate 25 mL
of LB medium and the culture was
grown overnight prior to dilution into 1 L of 1X M9
, 20 amino acids (8.5 mM NaCl, 18.7
mM NH
4
Cl, 22 mM KH
2
PO
4
, 47.8 mM Na
2
HPO
4
, 0.1 mM CaCl
2
, 1 mM MgSO
4
, 3 mg/L
FeSO
4
, 1 Gg/L of trace metals (Cu
2+
, Mn
2+
, Zn
2+
, MoO
4
2
), 35 mg/L thiamine
hydrochloride, 10 mg/L biotin, 20 mM Dglucose, 200
mg/L ampicillin with 50 mg/L of L
amino acids, each). At an appropriate cell density
(OD
600
~0.8), the culture was
subjected to a medium shift; briefly, cells were ce
ntrifuged and washed with saline prior
to resuspension into 0.8 L of 1.25X M9, 19 aa (M9,
20 aa medium without Lproline).
After cells were further incubated for 30 min to de
plete intracellular proline, 200 mL of
5X additives (1.5 M NaCl, 2.5 mM Hyp or Hzp) was ad
ded to the culture. After another
3
15 min of incubation at 37°C to allow amino acid up
take prior to induction, IPTG was
added to a final concentration of 1 mM. At the end
of 2 h, cells were harvested by
centrifugation and stored at 80°C until further us
e.
Cell lysis and refolding from inclusion bodies.
Cells were thawed on the benchtop
for 15 min prior to resuspension in lysis buffer (B
PER®, 0.5 mg/mL lysozyme, 50 U/mL
benzonase nuclease). Cells were gently agitated at
RT for 1 h prior to centrifugation
(10 000 g, 10 min, RT); supernatant was discarded a
nd the pellet was washed thrice:
once with wash buffer (2 M urea, 20 mM Tris, 1% Tri
ton X100, pH 8.0) and twice with
sterile ddH
2
O; centrifugation followed each wash and the supern
atant was discarded.
The final washed pellet containing inclusion bodies
(IBs, ~50% PI) was resuspended in
NiNTA binding buffer (8 M urea, 300 mM NaCl, 50 mM
NaH
2
PO
4
, pH 8.0) overnight at
4°C or at RT for 2 h, both with gentle agitation.
The suspension was centrifuged to
remove insoluble debris; the remaining pellet was d
iscarded and the supernatant was
mixed with preequilibrated NiNTA resin (Qiagen) a
t RT for 1 h in order to purify PI from
the IB fraction. Unbound proteins in the IB fracti
on were collected in the flowthrough
(FT), and the resin was washed with NiNTA wash buf
fer (8 M urea, 20 mM Tris base, 5
mM imidazole, pH 8.0) and NiNTA rinse buffer (8 M
urea, 20 mM Tris base, pH 8.0)
prior to stripping PI from the resin with NiNTA el
ution buffer (8 M urea, 20 mM Tris
base, pH 3.0). Fractions (IBs, FT, W, elution) wer
e collected and run under reducing
conditions on SDSPAGE (Bis/Tris gels, Novex®); elu
tion fractions containing PI were
pooled and solution pH was adjusted to 9.6 with 6 N
NaOH in preparation for oxidative
sulfitolysis. Oxidative sulfitolysis was performed
at RT for 4 h, with the addition of
sodium sulfite and sodium tetrathionate (0.2 M Na
2
SO
3
, 0.02 M Na
2
S
4
O
6
); the reaction
was quenched by 10fold dilution with ddH
2
O. To isolate PI from the quenched solution,
the pH was adjusted to between 3.5 and 4.5 by addin
g 6 N HCl dropwise; the solution
became cloudy. The solution was centrifuged (10 00
0 g, 10 min, RT) and supernatant
discarded. The PI pellet was then resuspended in
refolding buffer (0.3 M urea, 50 mM
glycine, pH 10.6) and protein concentration was est
imated by the bicinchoninic acid
assay (BCA assay, Pierce®). The concentration of P
I was adjusted to 0.5 mg/mL.
Refolding was initiated by addition of βmercaptoet
hanol to a final concentration of 0.5
mM and allowed to proceed at 12°C overnight with ge
ntle agitation (New Brunswick®
shaker, 100 RPM). Postrefolding, soluble PI was h
arvested by adjusting the pH of the
solution to 45 by dropwise addition of 6 N HCl and
by high speed centrifugation to
remove insoluble proteins. The supernatant was adj
usted to pH 88.5 by dropwise
addition of 6 N NaOH and dialyzed against fresh PI
dialysis buffer (7.5 mM sodium
phosphate buffer, pH 8.0) at 4°C with five buffer c
hanges to remove urea. The retentate
(PI in dialysis buffer) was then lyophilized and su
bsequently stored at 80°C until further
processing. Typical yields were 2550 mg PI per L
of culture (2530 mg/L for non
canonical PI, 4050 mg/L for canonical PI expressio
n in rich media)
Proteolysis and chromatographic (HPLC) purification
.
The dry PI powder was re
dissolved in water to a final concentration of 5 mg
/mL PI (final concentration of sodium
phosphate buffer is 100 mM, pH 8.0). Trypsin (Sigma
Aldrich) and carboxypeptidaseB
(Worthington Biochemical) were added to final conce
ntrations of 20 U/mL and 10 U/mL,
respectively to initiate proteolytic cleavage. The
PI/protease solution was incubated at
4
37°C for 2.5 h; proteolysis was quenched by additio
n of 0.1% trifluoroacetic acid (TFA)
and dilute HCl to adjust the pH to 4. Matured insul
in was purified by reversed phase
highperformance liquid chromatography (HPLC) on a
C
18
column using a gradient
mobile phase of 0.1% TFA in water (solvent, A) and
0.1% TFA in acetonitrile (ACN;
solvent, B). Elution was carried from 0% B to 39% B
with a gradient of 0.25% B per
minute during peak elution. Fractions were collecte
d and lyophilized, and the dry
powder was resuspended into 10 mM sodium phosphate
, pH 8.0. Insulincontaining
fractions were verified by matrixassisted laser de
sorption/ionizationmass spectrometry
(MALDIMS; Voyager MALDITOF, Applied Biosystems) a
nd SDSPAGE to ensure
identify and purity. Typical yields were 510 mg in
sulin per 100 mg PI. Fractions were
stored at 80°C in 10 mM phosphate buffer, pH 8.0 u
ntil further use.
Verification of Hyp and Hzp incorporation levels an
d maturation
. A 30 GL aliquot of
PI solution (8 M urea, 20 mM Tris, pH 8) was subjec
ted to cysteine reduction and
alkylation (5 mM DTT, 55°C, 20 min; 15 mM iodoaceta
mide, RT, 15 min, dark) prior to
10fold dilution into 100 mM NH
4
HCO
3
, pH 8.0 (100 GL final volume). Peptide digestion
was initiated with 0.6 GL of gluC
stock solution (reconstituted at 0.5 Gg/GL with ddH
2
O,
Promega) at 37°C for 2.5 h. The reaction was quenc
hed by adding 10 GL of 5% TFA
and immediately subjected to C
18
ZipTip (Millipore) peptide purification and desalt
ing
according to the manufacturer’s protocol. Peptides
were eluted in 50% ACN, 0.1%
TFA; the eluent was then diluted threefold into ma
trix solution (saturated α
cyanohydroxycinnamic acid in 50% ACN, 0.1% TFA) and
analyzed by mass
spectrometry (Voyager MALDITOF, Applied Biosystems
). Hyp and Hzp incorporation
levels were analyzed prior to and after refolding;
incorporation percentage was
calculated by comparing total AUC (area under the c
urve, arbitrary units) of the non
canonical peak (m/z = 1573 Da for the proinsulin pe
ptide containing B28Hzp or
B28Hyp) with total AUC of its wildtype counterpart
(m/z = 1557 Da). Incorporation
levels stated in Table S1 were obtained from mass s
pectra of peptides acquired from at
least four different expressions of HzpPI and Hyp
PI. Maturation of HypI and HzpI was
analyzed after HPLC purification. TFA (1.6 GL, 5%)
was added to 15 GL mature insulin
solution (10 mM phosphate buffer pH 8.0) and subjec
ted to C
18
ZipTip (Millipore)
peptide purification and desalting according to the
manufacturer’s protocol. MALDIMS
conditions described above were used to confirm ins
ulin maturation.
Insulin receptor (IR) phosphorylation immunoblot.
In vitro
analysis of insulin
receptor (IR) phosphorylation was performed using H
EK293 cells according to a
previous report
1
. Briefly, HEK293 cells were maintained in a 37°C,
5% CO
2
humidified
incubator chamber using Dulbecco’s modified Eagle’s
medium with 4.5 g/L glucose, 2
mM Lglutamine and phenol red (DMEM, Life Technolog
ies) supplemented with 10%
fetal bovine serum (FBS, Life Technologies), 5% pen
icillin/streptomycin (P/S, Life
Technologies). Every 3 days, at approximately 80%
confluency, cells were subcultured
and seeded in a 6well plate at a cell density of 8
x10
3
cells / cm
2
(or 8x10
4
cells per
well) for 24 h prior to insulin addition. Insulins
or vehicle were added directly to the
medium at 200 nM (10 GL of a 50 GM solution in vehi
cle PBS) and incubated for 10 min
prior to PBS washes to remove excess medium. HEK29
3 cells were lysed onplate
using IP Lysis Buffer (ThermoFisher, Pierce) with 5
0 U/mL benzonase nuclease
5
(SigmaAldrich) for 20 min at 4°C; lysates were pre
cipitated using ice cold acetone and
resuspended in 8 M urea, 20 mM Tris, pH 10.0. The
protein concentration in the lysate
was quantified by the BCA assay (ThermoFisher, Pier
ce) according to the
manufacturer’s protocol and normalized for even pro
tein loading across lanes. Lysates
were separated by SDSPAGE (412% Novex Bis/Tris SD
SPAGE gels, Life
Technologies) in duplicate and transferred to a nit
rocellulose membrane (Hybond ECL,
GE Healthcare) using a wet transfer system. The me
mbrane was blocked at RT in 5%
nonfat milk in Trisbuffer saline with 0.1% Tween 2
0 (TBS/Tween) and washed with
TBS/Tween prior to blotting with antibodies. Primar
y antibodies for insulin receptor,
phosphorylated insulin receptor (from Cell Signalin
g Technologies) and βactin (as
loading control, from Invitrogen) were added at 1:1
000 dilution in TBS/Tween with
gentle agitation either at RT for 4 h or overnight
at 4°C. Blots were washed and
secondary antibodies (Invitrogen) were added at 1:2
000 dilution in TBS/Tween. Blots
were washed again prior to fluorescence imaging on
a Typhoon Trio (GE Healthcare).
Reduction of blood glucose in diabetic animals.
NODscid (NOD.CB17
Prkdc
scid
/J)
mice were obtained from Jax Mice (Bar Harbor, Maine
). Mice were maintained under
specific pathogenfree conditions, and experiments
were conducted according to
procedures approved by the Institutional Animal Car
e and Use Committee at the City of
Hope. Adult (812 week old) male NODscid mice were
injected intraperitoneally (50
mg/kg/day for 3 consecutive days) with freshly prep
ared streptozotocin (STZ) in 0.05 M
citrate buffer, pH 4.5 to induce diabetes. Diabetes
was confirmed 3 weeks after the last
dose of STZ by detection of high glucose levels (de
fined as >200 mg/dL), measured by
using a glucomonitor (FreeStyle; Abbott Diabetes Ca
re, Alameda, CA) in blood (10 GL)
sampled from the lateral tail vein. Insulin analogs
concentrations were determined from
A
280
measurements using a molar extinction coefficient
of 6080 M
1
cm
1
and diluted to
100 μg/mL into a formulation buffer according to a
previous report
2
. Insulin analogs in
solution were injected subcutaneously at the scruff
and blood glucose was measured at
the indicated time points.
Hexamer dissociation assay.
Insulins were quantified by both UV absorbance
(NanoDrop Lite, ThermoFisher) and BCA assay, and no
rmalized to 125 GM insulin prior
to dialysis against 50 mM Tris/perchlorate, 25 GM z
inc sulfate, pH 8.0 overnight at 4°C
using a Dtube dialyzer (Millipore Corp.) with MWCO
of 3.5 kDa. Aliquots of dialyzed
insulin solution were mixed with phenol to yield sa
mples of the following composition:
100 GM insulin, 20 GM zinc sulfate, 100 mM phenol.
Dissociation was initiated by
addition of terpyridine (SigmaAldrich) to a final
concentration of 0.3 mM from a 0.75
mM stock solution. A Varioskan multimode plate read
er (Thermo Scientific) was used to
monitor absorbance at 334 nm. Kinetic runs were don
e at least in triplicate, and the data
were fit to a monoexponential function using Origi
n software. Post assay insulin
samples were pooled and sample quality was determin
ed by SDSPAGE.
Fibrillation Assay.
Insulin samples (60 GM in 10 mM phosphate, pH 8.0)
were
centrifuged at 22 000 g for 1 h immediately after a
ddition of thioflavin T (ThT) (EMD
Millipore) to a final concentration of 1 GM. Sampl
es were continuously shaken at 960
rpm on a Varioskan multimode plate reader at 37°C,
and fluorescence readings were
6
recorded every 15 min for 48 h (excitation 444 nm,
emission 485 nm). Assays were run
in quadruplicate, in volumes of 200 GL in sealed (P
erkinElmer), black, clearbottom 96
well plates (Grenier BioOne).
Circular dichroism.
Spectra were collected in a 1 cm quartz cuvette at
an insulin
concentration of 60 GM in 50 mM sodium phosphate bu
ffer pH 8.0. Data were collected
from 185 nm to 250 nm, with step size of 0.25 nm an
d averaging time of 1 s on a Model
410 Aviv Circular Dichroism Spectrophotometer; spec
tra were averaged over 3 repeat
scans. A reference buffer spectrum was subtracted f
rom the sample spectra for
conversion to mean residue ellipticity.
Analytical ultracentrifugation.
Sedimentation velocity (SV) and sedimentation
equilibrium (SE) experiments were carried out on an
XL1 AUC (BeckmanCoulter). SV
experiments were conducted with insulin samples dia
lyzed against 50 mM Tris, 0.1 mM
EDTA, pH 8.0, which also served as the reference bu
ffer. Two sector cells with sapphire
windows were filled with sample and reference buffe
r. These cells were centrifuged at
50,000 rpm with absorbance data collected at 280 nm
, or for concentrations above 1
mg/mL, 281 nm or 287 nm. SV data were analyzed in S
EDFIT with the c(s) algorithm for
a continuous distribution
3
. Buffer density and viscosity were calculated from
SEDNTERP; the partial specific insulin volume used
was 0.735
4
. SE experiments were
conducted with insulin samples dialyzed against 50
mM Tris, 0.1 mM EDTA, pH 8.0,
which also served as the reference buffer. Two sect
or cells with sapphire windows were
filled with sample and reference buffer and centrif
uged at 15,000, 24,000, 36,000 and
50,000 rpm with absorbance data collected at 280 nm
. Equilibrium was ascertained by
analysis in SEDFIT and nonequilibrated scan speeds
were excluded from data
analysis. SE and SV data from multiple concentratio
ns were fitted to a monomerdimer
hexamer reversible selfassociation model in SEDPHA
T with best model chosen by
inspection of residuals as well as critical
χ
value deviation
5
. Radial dependent baselines
were computationally determined using TI noise. Fig
ures were generated using
GUSSI
6
.
Crystallographic studies.
Insulin crystals were obtained from sitting drop tr
ays set
using a Mosquito robot (TTP Labtech). Drops were se
t by mixing 0.4 GL insulin solution
with 0.4 GL well solution. Well solution conditions
were as follows: 462.5 mM sodium
citrate, 100 mM HEPES, pH 8.25 for 5HQI; 300 mM Tri
s, 0.5 mM zinc acetate, 8.5%
acetone, 0.5 M sodium citrate pH 8.0 for 5HPR; 300
mM Tris, 17 mM zinc acetate, 1%
phenol, 7.5% acetone, 2.675 M sodium citrate pH 8.0
for 5HRQ; 300 mM Tris, 17 mM
zinc acetate, 1% phenol, 7.5% acetone, 1.95 M sodiu
m citrate pH 8.0 for 5HPU. Cells
were cryoprotected in a mother liquor containing 30
% glycerol prior to looping and flash
freezing in liquid nitrogen. Data were collected at
SSRL beamline BL122 using a
DECTRIS PILATUS 6M pixel detector. Initial indexing
and scaling was performed with
XDS; for some structures, data were rescaled in al
ternative space groups using
Aimless
7
. Initial phases were generated by molecular replac
ement in PHASER with
3T2A (5HQI and 5HPR) or 1EV3 (5HRQ and 5HPU)
8
. Structure refinement was carried
out in Coot and Refmac5
910
. Data were deposited in the PDB with the following
codes:
5HQI (T
2
HzpI), 5HPR (T
2
HypI), 5HRQ (R
6
HzpI), 5HPU (R
6
HypI).
7
Transmission electron microscopy.
Insulin samples (60 GM in 10 mM phosphate, pH
8.0) were continuously agitated in microfuge tubes
at 42°C, 960 RPM for 48 h in an
Eppendorf ThermoMixer to obtain fibrils. Samples we
re stained on 200mesh copper
grids (formar/carbon coated, plasma cleaned) with 1
% uranyl acetate. Imaging was
done by Dr. Alasdair McDowell at the Beckman Instit
ute Center for Transmission
Electron Microscopy on a Tecnai T12 LaB6 120 eV tra
nsmission electron microscope.
8
Figure S1. Insulin expression and incorporation of
hydroxyprolines.
(
A, B
)
SDS
PAGE of cell lysates with lanes labeled for preind
uction (PRE) and post
induction in minimal media supplemented with either
nothing (19aa), Hyp (
A
),
Hzp (
B
), or Pro at 0.5 mM. (
C-E
)
:
MALDIMS traces of isolated proinsulin
peptide fragment
46
RGFFYT
P
KTRRE
57
obtained by gluC digestion. Peptide
fragment masses correspond to either wild type mass
(1558 Da) (
C
) or shifted
mass (1574 Da) if Hyp (
D
) or Hzp (
E
) is incorporated. Inset is whole protein
MALDIMS. All MALDIMS spectra contain ion counts >
10
3
.
B
A
PRE 19aa +Hzp +Pro
PRE 19aa
+Hyp
+Pro
C
% Intensity
D
E
mass (m/z)
9
Figure S2. Immunoblot detection of insulin receptor
activation.
HEK293 cells
treated with insulin (200 nM in PBS, pH 7.4) or veh
icle. Whole cell lysates were
then run on an SDSPAGE gel and transferred to nitr
ocellulose membrane to
detect insulin receptor (IR) and IR phosphorylation
. βactin immunoblot shown as
loading control. Lane 1: Vehicle (PBS); Lane 2: 10
% ProI serving as a second
negative control due to presence of 10% wt in HzpI
and HypI preparations; Lane
3: HzpI; Lane 4: HypI; Lane 5: ProI. Quantification
of PIR and IR bands was
done using ImageJ software.
10
Figure S3. Example fits from sedimentation analysis
.
Insulin samples in 50 mM Tris
pH 8.0. (
A
) c(s) curves overlaid for ProI, AspI, HzpI and Hyp
I at 60 GM.
(
B, C
)
c(s) curves for HzpI (
B
) and HypI (
C
) at indicated concentrations.
(
D-F
) Example
fits for 60 GM ProI (
D
), 34 GM HzpI (
E
), and
60 GM HypI (
F
), overlaid on top of
noise corrected velocity (
D, F
) or equilibrium data (
E
). The SEDPHAT monomer
dimerhexamer model was utilized over a range of ve
locity and equilibrium
experiments. Global multimethod analysis residuals
for the dataset are
displayed below each plot.
HypI
Overall residual of 0.007377 (GMMA).
HzpI
0.35 mg/mL
0.75 mg/mL
1.50 mg/mL
A
B
C
D
E
F
Sedimentation Velocity of ProI at 60 μM.
ProI
HzpI
HypI
AspI
Overall residual of 0.03431 (GMMA)
Sedimentation Velocity at 60 μM.
Sedimentation Velocity of HzpI.
Sedimentation Velocity of HypI.
ProI
HzpI
HypI
AspI
Sedimentation Velocity of HypI at 60 μM.
Sedimentation Equilibrium of HzpI at 34 μM.
Overall residual of 0.005362 (GMMA).
0.10 mg/mL
0.20 mg/mL
0.35 mg/mL
1.21 mg/mL
11
Figure S4. Example fits for analysis of dissociatio
n kinetics.
(
A
) Representative
dissociation kinetic traces for Zn
2+
sequestration. Raw data shown and used to fit
to a monoexponential using Origin Software (yy
0
= A
e
(t/τ)
), where fitted value τ
is the characteristic dissociation time constant. (
B, C
)
Fitted monoexponential
decay traces for dissociation kinetics, correspondi
ng to (
A
) shown in (
B
). Fitted
values for y
0
, A used to convert raw data (
A
) to monoexponential decay
representation shown in (
C
). (
D
) Overlay of (
B
) and (
C
) show fitted and raw data
to demonstrate quality of fits. *Denotes fitted cu
rves
12
Figure S5. Alignment at position B28.
(
A, D
) Alignment of T
2
ProI (tan, PDB:3T2A),
and T
2
HzpI (grey) or T
2
HypI (blue) centered on position B28. (
B, E
) Alignment
of R
6
ProI (tan) and R
6
HzpI (grey) or R
6
HypI (blue) highlighting the overlap of
the backbone at the
C
terminus. B29 not shown in (
E
)
due to lack of electron
density. (
C, F
) Alignment of R
6
insulins (ProI, and HzpI or HypI), and AspI
(orange, PDB: 1ZEG) centered on position B28 illust
rates the similarity of the
polypeptide backbones of ProI, HzpI and HypI, and t
he distinct backbone
trajectory of AspI. B29 (
C, F
) and B30 (
A-F
) amino acids not shown for clarify.
Arrows denote the NtoC terminal direction of the
backbone originating from
carbonyl carbon.
13
Figure S6. Alignment of R6-AspI and T2-HzpI at posi
tion B28.
(
A
)
R
6
AspI (dark
orange; PDB: 1ZEH) does not maintain the backbone t
rajectory of ProI at
position B28. The
C
terminus of the AspI Bchain
is shifted, and a
m
cresol
ligand (light orange) fills the site occupied by B2
8Pro in ProI. The hydroxyl group
of
m
cresol
forms hydrogen bonds with the backbone carbonyl of
Glu21′ and a
nearby water molecule. (
B
)
The same representation of R
6
AspI overlaid with R
6

HzpI (dark grey). Interatomic distances were determ
ined using Chimera. Amino
acid B30 is not shown.
14
Figure S7. Independent measurement of fibrillation
lag time.
Representative
fibrillation curves for 60 μM insulins (37°C, 960 R
PM; n as indicated). Samples
were prepared through separate growth, refolding, p
urification (on a different C18
HPLC column) steps as compared to samples shown in
Figure 2C. Insulin fibrils
were detected by the rise in Thioflavin T (ThT) flu
orescence that accompanies
binding to fibrillar aggregates. HzpI did not fibr
illate prior to termination of this
experiment after 18 h.
15
Figure S8. Transmission electron micrographs of ins
ulin fibrils.
ProI (
A
), HzpI (
B
),
and HypI (
C
). Scale bar 100 nm; staining with 1% uranyl acetat
e on 200mesh
copper grids.
16
Figure S9. Biophysical characterization of ProI sam
ples prepared in-house and
obtained from a commercial source.
ProI (Sigma) was purchased from Sigma
Aldrich and purified by HPLC. (
A
) Hexamer dissociation kinetics (
B
)
Sedimentation velocity at 60 μM.