Proinsulin maturation, misfolding, and proteotoxicity
Ming Liu*, Israel Hodish*, Christopher J. Rhodes†, and Peter Arvan*‡
*Division of Metabolism, Endocrinology, and Diabetes, University of Michigan Medical Center, Ann Arbor, MI 48109; and†Section of Endocrinology,
Diabetes, and Metabolism, University of Chicago, Chicago, IL 60637
Edited by Donald F. Steiner, University of Chicago, Chicago, IL, and approved August 9, 2007 (received for review March 26, 2007)
As a tool to explore proinsulin (PI) trafficking, a human PI cDNA has
been constructed with GFP fused within the C peptide. In regulated
secretory cells containing appropriate prohormone convertases, the
hProCpepGFP construct undergoes endoproteolytic processing to
CpepGFP and native human insulin, which are specifically detected
and cosecreted in parallel with endogenous insulin. Expression of
C(A7)Y mutant PI results in autosomal dominant diabetes in Akita
mice. We directly identify the misfolded PI in Akita islets and also
chimera or not, engages directly in protein complexes with nonmu-
tant PI, impairing the trafficking and recovery of nonmutant PI. This
trapping mechanism decreases insulin production in ? cells. Thereaf-
ter we observe a loss of ? cell viability. The data imply that PI
misfolding leading to impaired endoplasmic reticulum exit of non-
mutant PI may be a key early step in a chain reaction of ? cell
diabetes mellitus ? endoplasmic reticulum storage disease ?
insulin secretion ? proinsulin disulfide isomers
cell insults are implicated in the onset of ? cell stress, dysfunction,
and death. Of these, a case can be made for proinsulin (PI)
misfolding with resultant endoplasmic reticulum (ER) stress as a
proximal event in the molecular pathogenesis of DM. In the PI
superfamily, a biophysical basis for misfolding both in vitro (1) and
in vivo (2) involves predisposition to disulfide mispairing. This may
account for susceptibility to ER stress in pancreatic ? cells (3)
compared with other cell types. Even when the ER stress response
is fully activated (4) and ER-associated protein degradation is
up-regulated to clear misfolded PI (5), ? cells still remain highly
susceptible to proteotoxicity (6).
The Akita mouse expresses wild-type PI from three alleles (two
Ins1 and one Ins2), which is more than sufficient to avoid DM (7).
deficiency (8, 9) with loss of ? cell mass within weeks postnatally
(10). The Akita gene encodes a mouse PI-C(A7)Y mutant, where
A7 refers to the seventh residue of the A chain that normally
engages in a crucial disulfide bond (11). One recent study reported
no difference in extent of PI misfolding in Akita versus normal
mutant PI in the presence of large quantities of wild-type PI. We
recently reported that nonreducing Tris?tricine?urea? SDS/PAGE
can identify abnormal disulfide pairing when recombinant C(A7)Y
mutant PI was expressed in 293 cells (2). The ability to ‘‘see’’
misfolded PI in live ? cells would be an important advance (13), yet
this loss of function would not necessarily address the decrease of
insulin production derived from coexpressed nonmutant PI.
To help clarify this, we have developed a strategy that incorpo-
rates the C(A7)Y mutation into a PI fusion protein bearing the
GFP. Such a strategy requires that the nonmutant version of
the chimera serves as a suitable model for the endogenous protein.
The concept of a GFP reporter fused within the C peptide GFP
allows for unique immunoreactivity with anti-GFP while allowing
favorable efficiency of intracellular transport of green PI to secre-
tory granules (14). From this concept, we have developed a GFP
chimera of human PI designed to serve as a template for introduc-
ing the C(A7)Y mutation. We establish physical interactions of
ancreatic ? cell failure is increasingly recognized as central to
progression of diabetes mellitus (DM). Different potential ?
the mutant PI with coexisting wild-type PI, retaining the latter in
Chimeras with GFP as Models for PI Folding and Trafficking. We
wished to design a GFP chimera to serve as a faithful PI reporter,
the PI B chain, the chimera is retained within the ER (15). We
therefore compared a chimeric construct containing GFP at the C
terminus of human PI (16) to that with GFP fused within the C
peptide [supporting information (SI) Data Set 1]. In transfected
293T cells (a robust protein expression system), a portion of the
human PI–GFP [i.e., C-terminal GFP (16)] was unexpectedly
cleaved to liberate a GFP-immunoprecipitable band comigrating
with authentic cytosolic GFP (SI Fig. 7, lanes 2 and 3) as well as an
insulin-immunoprecipitable band comigrating with authentic PI
(lanes 16 and 17). No established proprotein processing site exists
at the junction of PI and the C-terminal GFP, rendering this
construct unsuitable as a reporter of PI within the ER. By contrast,
in 293T cells hProCpepGFP expression released neither free GFP
was well recognized both by anti-insulin (lane 8) and anti-GFP
We next checked whether transport to secretory granules could
endoproteolytically liberate CpepGFP and insulin. An initial ex-
periment was performed in transfected AtT20/PC2 cells (regulated
secretory cells containing prohormone convertases 1 and 2), se-
lected because they lack any endogenous (pro)insulin expression.
Upon expression of hProCpepGFP and metabolic labeling, immu-
noprecipitation with anti-GFP revealed an initial translation prod-
uct corresponding to hProCpepGFP (SI Fig. 8). At 2 h after
synthesis, ongoing processing could be detected as smaller GFP-
could be secreted. Likewise, upon immunoprecipitation with anti-
insulin, a band precisely comigrating with authentic insulin, indi-
cating native disulfide bonds (17), was specifically produced within
2 h of chase (SI Fig. 8), despite that only chimera and not PI was
expressed. The data indicate production of insulin derived from
To test hProCpepGFP processing in a ? cell context, a
human insulin-specific RIA is available that does not recog-
nize uncleaved human PI or fully processed insulin of rodent
species. Neither uninfected 293T cells nor those expressing
hProCpepGFP or human PI exhibited human insulin-specific
research; M.L. and I.H. performed research; M.L., I.H., and C.J.R. contributed new reagents/
analytic tools; and M.L., I.H., and P.A. analyzed data and wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Abbreviations: ER, endoplasmic reticulum; PI, proinsulin; DM, diabetes mellitus; SEAP,
secreted alkaline phosphatase.
and Diabetes, University of Michigan Medical School, 5560 Medical Science Research
Building II, 1150 West Medical Center Drive, Ann Arbor, MI 48109-0678. E-mail:
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2007 by The National Academy of Sciences of the USA
October 2, 2007 ?
vol. 104 ?
no. 40 ?
immunoreactivity (Fig. 1, fourth, fifth, and sixth sets of bars).
Furthermore, neither INS-1 (rat) cells nor those expressing
cytosolic GFP (second and third sets of bars) showed signif-
icant human insulin-specific immunoreactivity in 2 h of un-
stimulated secretion, 2 h of secretagogue-stimulated secretion,
or final cell lysates (Fig. 1). However, INS-1 cells expressing
hProCpepGFP (first set of bars) showed significant human
insulin secreted during an initial two unstimulated hours (gray
bars), a 7-fold increase after secretagogue stimulation (black
bars), and a large remaining intracellular fraction (white bars).
This behavior is virtually identical to that of endogenous rat
insulin released from INS-1 cells (see below).
molecular masses of expected GFP-containing processing interme-
diates of hProCpepGFP. The unique mobilities of B chain–GFP,
CpepGFP, and GFP–A chain are shown in Fig. 2 Top Right.
Although untransfected INS-1 cells showed several nonspecific
bands (Fig. 2 Top Left), the initial hProCpepGFP translation
a major form comigrating with CpepGFP accumulated at 4 h of
release was amplified over that released from unstimulated cells.
Only small amounts of bands comigrating with B chain–GFP and
GFP–A chain were recovered. An experiment that shows similar
stimulus-dependent secretion of CpepGFP was conducted with
insulin immunoprecipitation performed in parallel (Fig. 2 Bottom).
Similar to CpepGFP, stimulated insulin release to the medium
decreased the amount remaining in cells. This behavior appeared
similar to that observed from untransfected INS-1 cells (Fig. 2
By confocal microscopy of INS-1 cells stably expressing
hProCpepGFP, most INS-1 cells are small and round and lack
cytoplasmic processes. Nevertheless, the green fluorescence
distribution from hProCpepGFP in ?75% of cells (n ? 105
cells examined) was quite distinct from the ER marker caln-
exin (in red), instead expressing a peripheral punctate (gran-
ule-like) staining pattern that was comparable to the im-
muofluorescence of insulin (in blue; e.g., Fig. 3A). In
occasional INS-1 cells that extend cytoplasmic processes, a
granular pattern of green fluorescence was seen to accumulate
in the distal tips of processes (Fig. 3C). The data suggest
normal intracellular transport of the hProCpepGFP chimera
with production of human insulin and CpepGFP that are
costored and cosecreted from insulin secretory granules.
Influence of the C(A7)Y Mutation on PI and hProCpepGFP. Introduc-
tion of the C(A7)Y mutation in the insulin coding sequence
cells that extend cytoplasmic processes, hProC(A7)Y-CpepGFP
fluorescence was specifically excluded from these processes, espe-
cially their distal tips (Fig. 3D), suggesting failure of the mutant
hProC(A7)Y-CpepGFP to be delivered to the distal secretory
pathway. In more typical rounded INS-1 cells, none of 115 cells
examined revealed hProC(A7)Y-CpepGFP fluorescence in a se-
cretory granule-like distribution; instead GFP fluorescence
changed to largely overlap with the ER marker calnexin (e.g., Fig.
3B). Curiously, immunofluorescence reflecting endogenous (pro)
insulin distribution changed similarly (Fig. 3B), suggesting a possi-
ble effect of the mutant protein on the fate of wild-type PI. Indeed,
we found by coexpression in 293 cells that, as the hProC(A7)Y-
CpepGFP to wild-type PI ratio was increased, initial synthesis of
labeled PI was largely unaffected but secretion of labeled PI at 1 h
of PI cDNA yet maintaining a high plasmid ratio of hProC(A7)Y-
CpepGFP to wild-type PI, pulse–chase studies confirmed that
secretion of coexpressed PI was impaired while simultaneous
Adenoviral expression of constructs was used in preparation for human insulin-
conditions (gray bars), then incubated for a further 2 h with a secretagogue
mixture (black bars), and finally the cells were lysed and extracted with acid
tion of human insulin and store the remainder intracellularly.
Detection of processed human insulin in 293T cells and INS-1 cells.
another for transiently transfected 293T cells (Right). INS-1 cells were pulse-
chased for the times indicated. 293T cells transfected with cDNAs encoding B
labeled for 60 min without chase. All samples were immunoprecipitated with
anti-GFP before SDS/PAGE fluorography. Chase media (M) bathing stably trans-
same protocol, including 293T control cells expressing cytosolic GFP (G) or Cpep-
GFP (far right) and untransfected INS-1 cells (N). [A low-intensity GFP band
anti-insulin immunoprecipitation from the identical samples and stimulated
insulin secretion and final cell lysate from control INS-1 cells.
Processing and release of hProCpepGFP from INS-1 cells. (Top) Two
www.pnas.org?cgi?doi?10.1073?pnas.0702697104Liu et al.
secretion of coexpressed secreted alkaline phosphatase (SEAP)
was unimpaired (Fig. 3F). Thus, initial secretory defects caused by
hProC(A7)Y-CpepGFP appeared selective for PI.
To understand the extent to which the coexpressed C(A7)Y
mutant perturbs insulin production, we performed pulse–
chase experiments in islets isolated from young Akita and
wild-type mice. Newly synthesized PI from Akita islets exhib-
ited indistinguishable mobility from that of normal islets upon
Tris?tricine?urea? SDS/PAGE under reduced conditions, and
Akita islets exhibited more labeled PI at the zero chase time
than did wild-type islets (Fig. 4A). However, under nonre-
duced conditions, this increase was no longer apparent (Fig.
4A). In Akita islets under nonreduced conditions (Fig. 4A)
cence in INS-1 cells expressing hProCpepGFP (A) or hProC(A7Y)-CpepGFP (B) is
compared with immunofluorescence of calnexin (anti-Clnx) in red, to which it
has been merged. Insulin immunostaining (anti-Ins in blue) is also shown. (C
and D) A minority of INS-1 cells extend neurite-like processes, but in such cells
the intracellular GFP distribution can be compared in cells expressing
hProCpepGFP (C) or hProC(A7Y)-CpepGFP (D). (E) 293T cells were cotrans-
fected, varying the ratio of hProC(A7Y)-CpepGFP plasmid DNA to a fixed
20 min and either lysed without chase (0 h lanes marked with ?) or chased for
encoding SEAP was included. Cells and media were analyzed for both insulin
and SEAP by immunoprecipitation, SDS/PAGE, and autoradiography.
Expression of hProC(A7)Y-CpepGFP. (A and B) Emerald GFP fluores-
Akita islets and INS-1 ? cells expressing hProC(A7)Y-CpepGFP. (A) Islets isolated
35S-labeled amino acids for 30 min. Islets without chase or chased for 2 h in the
were immunoprecipitated with anti-insulin and analyzed by Tris?tricine?urea?
SDS/PAGE under reduced or nonreduced conditions. The arrowhead denotes a
position of aberrant mobility of nonreduced PI with improper disulfide pairing
(12). The migration of fully reduced and oxidized (native) PI is indicated. Note
that, under nonreduced conditions, newly synthesized PI in Akita islets is less
efficiently recovered than under reduced conditions, and there is an increase of
a higher-molecular-mass protein complex (open arrow) that is not detected
under reduced conditions. (B) Lysates of stably transfected INS-1 cells expressing
sion of hProCpepGFP (lane 5) were subjected to Western blotting with anti-GFP
after SDS/PAGE under reduced or nonreduced conditions. Lane 4, uninfected
hProC(A7)Y-CpepGFP is not endoproteolytically processed in ? cells, indicating
islets more efficiently process hProCpepGFP to CpepGFP (?90%) than in INS-1
cells (?50%). Finally, note that most hProC(A7)Y-CpepGFP is not recovered in its
normal position under nonreduced conditions (Right), with increased higher
molecular mass protein complexes (open arrows) that are not detected under
reduced conditions. The positions of molecular mass markers are shown on the
30 min and then lysed in immunoprecipitation buffer containing 0.1% SDS.
GFP-containing peptides were immunoprecipitated, and the samples were ana-
lyzed by Tris?tricine?urea? SDS/PAGE under reducing conditions to detect coim-
munoprecipitation of endogenous PI. Reduced PI was identified by using direct
PI immunoprecipitation from INS-1 control cells (Left).
Protein interactions of nonmutant and C(A7)Y mutant versions of PI in
Liu et al. PNAS ?
October 2, 2007 ?
vol. 104 ?
no. 40 ?
there was only a small but reproducible increase in the
slower-migrating PI monomer (Fig. 4A, arrowhead) and a
portion of immunoprecipitable insulin that was recovered in a
disulfide-linked high-molecular-weight complex (Fig. 4A, ar-
row). By 2 h of chase, the fraction of labeled PI that was
converted to insulin (signifying intracellular transport to se-
cretory granules) was clearly diminished in Akita islets.
It is not straightforward to precisely quantify the mutant fraction
of initial PI translation product in Akita islets. We therefore
blotting with anti-GFP, INS-1 cells or primary pancreatic islets
exhibited only few nonspecific bands under reduced or nonreduced
GFP, both this and CpepGFP were detected as specific bands. In
primary pancreatic islets expressing wild-type hProCpepGFP, the
ratio of processed CpepGFP to unprocessed precursor (Fig. 4B,
lane 5) was far greater than in INS-1 cells; this is expected because
prohormone processing in islets is much more efficient (18). Upon
expression of hProC(A7)Y-CpepGFP (Fig. 4B Left), no processing
to CpepGFP was detected, indicating an inability of the mutant to
undergo transport to secretory granules.
Under nonreduced conditions, hProC(A7)Y-CpepGFP was
largely detected in high-molecular-mass complexes (Fig. 4B, open
arrows) that disappeared under reduced conditions. Upon analysis
after immunoprecipitation of hProC(A7)Y-CpepGFP with anti-
GFP followed by reducing Tris?tricine?urea? SDS/PAGE, coprecipi-
tation of endogenous PI was detected (Fig. 4C). Covalent (but not
noncovalent) complexes containing endogenous PI were preserved
under our immunoprecipitation conditions (data not shown).
Mutant C(A7)Y PI undergoes ER-associated protein degra-
dation (5). Because the mutant is blocked in intracellular
transport (Figs. 3 and 4) and wild-type PI is bound to mutant
(Fig. 4C), this could account for loss of ? cell insulin. To test this,
we transfected INS-832/13 cells with hProCpepGFP or
hProC(A7)Y-CpepGFP. At 48 h after transfection, cells under-
went FACS. As a control, INS-832/13 were transfected only with
secretable GFP (19), which is not expected to augment or
decrease intracellular PI or insulin. Transient expression of
hProCpepGFP augmented by 30% the total PI plus insulin
detectable by RIA that cross-reacts with PI plus insulin of
multiple species (Fig. 5); concomitantly, a 30% increase in actual
human insulin was observed. By contrast, transient expression of
hProC(A7Y)-CpepGFP caused a 28% decrease in the amount of
human insulin stored (Fig. 5) but also caused a 23% decrease in
total PI plus insulin, most of which is derived from endogenous
rat PI. Mutant C(A7)Y human PI (hAkita, with a separately
encoded GFP) also markedly lowered the amounts of PI plus
insulin (Fig. 5). Thus, with or without the GFP moiety, the
C(A7)Y PI mutant exerts dominant negative behavior on en-
dogenous PI and insulin content of pancreatic ? cells.
? Cell Toxicity of the C(A7)Y PI Mutation. By restricting analysis to
expressing the hProC(A7)Y-CpepGFP mutant. INS-1 cells are
nonmotile, allowing daily imaging in culture. In cells expressing
transfection (Fig. 6A Upper), GFP fluorescence intensity changes
over time because of secretion and because gene expression is
transient; nevertheless, the identical cells were clearly identified
each day. By contrast, upon expression of hProC(A7)Y-CpepGFP
(Fig. 6A Lower), cells began to detach from the plate at 2–3 d after
transfection. Quantified over three independent experiments (Fig.
6B), the data indicate ? cell toxicity of hProC(A7)Y-CpepGFP
compared with the nonmutant protein. Taken together, studies of
the C(A7)Y PI mutation highlight a sequence of events, including
nonmutant PI transport, decreased PI and insulin content, and,
ultimately, ? cell death.
ER stress derived largely from secretory protein misfolding has
been implicated in pancreatic ? cell failure. In one example,
transgenic mice and rats develop type 2 DM that can be specifically
tide (20), as occurs in human type 2 DM (21). Rodent islet amyloid
that PI in all species may be predisposed to misfold (2). A primary
reason to date for limited investigation of PI folding in living cells
is the analytical difficulty in distinguishing well folded from mis-
folded PI. Indeed, this report demonstrates the misfolding of
C(A7)Y mutant PI that coexists with native disulfide-paired PI in
islets from Akita mice.
Early work suggested the presence of higher-molecular-weight
forms of PI in Akita islets (9), although a recent study from the
same laboratory failed to demonstrate PI misfolding beyond that
found in normal islets (12). On the other hand, another Cys
mutant of PI (disrupting the intra-A chain disulfide bond) has
DM (23). Indeed, each of the three disulfide bonds of PI impacts
not only final PI structure but also the PI folding pathway, and
S–S disruption creates potential to block PI folding intermedi-
ates. Even for nonmutant PI there is significant potential for
nonnative disulfide isomer formation, either as on-pathway or
off-pathway products (2).
We have wanted to understand why, in Akita islets, there is less
immunoreactive insulin per ? cell with fewer insulin secretory
granules (8, 24). ER stress response in islets and ? cells expressing
the C(A7)Y mutant PI (4, 5) [e.g., increased BiP mRNA (13) and
protein (9)], general deterioration of the secretory pathway (12),
and ? cell death have each been proposed to explain disease
pathogenesis in male Akita mice, yet most of these phenotypes can
be predicted to occur secondary to the deleterious effects of DM
on islet function. By contrast, up-regulated PI biosynthesis reflects
a relatively early stage of disease progression before ? cell failure
and death, yet at this time insulin production (Fig. 5A) is already
cross-reacts with PIs and insulins of multiple species) as well as human insulin
specifically (by RIA). INS-832/13 cells were transiently transfected with secretory
kita in the pCMS vector) as indicated. At 2 d after transfection, fluorescent cells
were sorted and total PI plus insulin content as well as processed human insulin
at 100%, whereas human insulin specifically contained in control cells (39) was
raises both total PI plus insulin and human insulin specifically. Expression of
hProC(A7)Y-CpepGFP (hatched bars) or hAkita (black bars) not only lowers hu-
man insulin, but lowers total PI plus insulin derived primarily from endogenous
rat PI, indicating dominant negative suppression. The mean and standard devi-
ation from three independent experiments are shown.
Steady-state levels of total PI plus insulin (by rat insulin RIA that
www.pnas.org?cgi?doi?10.1073?pnas.0702697104Liu et al.
formation involving the C(A7)Y mutant PI, in conjunction with a
significant decrease in wild-type PI-to-insulin maturation.
(Fig. 4B) and liberates authentic insulin and CpepGFP that is
with the human insulin (Fig. 1) from secretory granules (Fig. 3 A
and C). We understand that not every aspect of PI biosynthesis will
be identical in hProCpepGFP; e.g., the translation time and that
required for native PI disulfide bond formation are likely to be
substantially longer, whereas folding of GFP will consume addi-
tional time. Nevertheless, native PI disulfide pairing is remarkably
tolerant of substitutions within the C peptide (17). Altogether, our
observations support that hProCpepGFP is a functional model of
PI that can serve as a template on which to study effects of the
of processing (Fig. 4B) and intracellular fluorescence distribution
(Fig. 3 B and D) indicating ER retention of hProC(A7)Y-
CpepGFP. This is not surprising because the mutant also clearly
forms high-molecular-weight protein complexes (Fig. 4B). Most
importantly, we directly show that the intermolecular protein
complexes containing C(A7)Y mutant also include wild-type en-
dogenous PI, involving covalent (Fig. 4C) as well as possible
hydrophobic interactions (25).
ER retention of hProC(A7)Y-CpepGFP appears to confer im-
paired transport on nonmutant PI with which it is associated (Fig.
3E) but not on SEAP coexpressed from the same cells (Fig. 3F).
However, as cells get sicker, secretion of other proteins is likely to
be adversely affected (12). Another potential consequence that
requires further investigation is possible ER-associated protein
degradation of nonmutant PI. Indeed, in Akita islets at 2 h after
synthesis, there was a clear loss of labeled PI without a commen-
surate increase in labeled insulin (Fig. 4A). Consistent with the
possibility of ER-associated protein degradation, we establish a
dominant negative decrease in steady-state levels of nonmutant PI
plus insulin induced by expression of the C(A7)Y mutant (Fig. 5).
These data begin to explain insulin deficiency by a mechanism that
sets in motion a chain reaction of additional untoward outcomes.
Indeed, the data in Fig. 6, similar to a previous report (13), indicate
? cells (Fig. 6), which can only further compromise net insulin
production. However, even before ? cell death with resultant loss
of ? cell mass, we propose that the simplest interpretation is that
DM caused by Akita PI is initially a consequence of misfolded PI
causing decreased insulin production from nonmutant PI that is
blocked in the proximal secretory pathway.
Materials and Methods
Materials. We used guinea pig anti-rat insulin RIA (catalog no.
RI-13K; Linco/Millipore); human insulin-specific RIA (catalog
no. HI-14K; Linco/Millipore); rabbit anti-GFP (Immunology
Consultants Labs); Zysorbin (Zymed); [35S]methionine/cysteine
(ICN); Met/Cys-deficient DMEM, DTT, and RIA-grade BSA
(Sigma, St. Louis, MO); Hanks’s balanced salt solution Gibco);
collagenase P and proteinase inhibitor mixture (Roche); nitro-
AtT20/PC2 cells were a gift from R. Mains (University of
Connecticut, Farmington, CT).
Construction of hProCpepGFP and Processing Standards. The cDNA
encoding emerald GFP surrounded by flanking ApaI sites (14)
was amplified by PCR (forward primer, 5?-GGGCCCTATG-
GTGAGCAAGGGCGAGGAGCTG-3?; reverse primer, 5?-
gggccccgcagcagcagccttgtatagctc-3?) and ligated into pTarget
(Promega, Madison, WI) that already contained ApaI-
digested human PI cDNA or that bearing the Akita mutation.
Constructs encoding B chain CpepGFP (forward, 5?-
ggtaccatggccctgtggatgcgcctcctgcc-3?; reverse, 5?-gaattcctactg-
cagggacccctccagggccaa-3?), CpepGFP (forward, 5?-ggtaccAT-
GGAGGCAGAGGACCTGCAGGTGGG-3?; reverse, 5?-
gaattcctactgcagggacccctccagggccaa-3?), and CpepGFP A
chain (forward, 5?-ggtaccATGGAGGCAGAGGACCTG-
CAGGTGGG-3?; reverse, 5?-cctaagctagttgcagtagttctc-
cagctggta-3?) were created by PCR using hProCpepGFP as
template. Each PCR product was ligated into pTarget and
confirmed by DNA sequencing.
Transfection and Labeling of Cells in Culture.293Tcellswerecultured
in high-glucose DMEM plus 10% FBS. Cells were plated into
six-well plates 1 d before transfection. A total of 2 ?g of plasmid
DNA was transfected per well using Lipofectamine. At 48 h after
transfection, cells were pulse-labeled with35S-labeled amino acids
and chased for the times indicated. For Fig. 7, INS832/13 cells (26)
were plated onto 35-mm gridded glass-bottom culture dishes (Mat-
Tek) at 30% confluence 1 d before transfection. Each day, live
GFP-positive cells were specifically identified by their coordinates
on the gridded dish, with fluorescence and phase-contrast images
captured at low power in a tissue culture microscope before
returning cells to the incubator. Statistical significance was calcu-
lated by ANOVA as analyzed by Prism software (GraphPad).
CpepGFP. (A) INS-1 cells plated on gridded glass coverslips were transfected,
and the cells were imaged daily in a tissue culture microscope equipped with
epifluorescence. Three transfected INS-1 cells are highlighted in each field.
The cells are essentially nonmotile, and the relative positions of the cells each
day are indicated. Cells expressing hProC(A7)Y-CpepGFP (Lower, yellow ar-
row) were predisposed to be lost from the culture over time. (B) Quantitation
dramatic at 5 d after transfection, although the data were not included here
because the 5-d time point was not performed in all experiments.
Fate of INS-1 ? cells expressing hProCpepGFP or hProC(A7)Y-
Liu et al. PNAS ?
October 2, 2007 ?
vol. 104 ?
no. 40 ?
Construction and Use of hProCpepGFP Adenovirus. A replication- Download full-text
defective adenoviral construct expressing hProCpepGFP was gen-
erated by established protocol, purified by cesium chloride ultra-
centrifugation, and used with another replication-deficient
islets were infected (24 h at 37°C in RPMI medium 1640 plus 10%
for another 48 h before lysis and detection of hProCpepGFP and
6 h with adenovirus encoding hProCpepGFP at 150 pfu per cell.
Infected cells were washed and returned to regular cell culture
growth medium for another 40 h before experiments.
Stimulation of Insulin Secretion.INS-1cellsinfectedwithhProCpep-
GFP adenovirus were washed with prewarmed RPMI medium
1640 lacking glucose and then incubated for 2 h at 37°C in KRBH
2.8 mM glucose plus 0.2% RIA-grade BSA]. A second 2-h collec-
tion was then made in KRBH plus secretagogue (16.7 mM glucose,
10 mM glutamine, 1 mM tolbutamide, 1 mM isobutylmethylxan-
media were collected and cells were lysed in acid ethanol (28) in
preparation for human insulin-specific RIA.
Immunoprecipitation. Cell lysates and chase media (plus proteinase
inhibitor mixture) were precleared with Zysorbin and subjected to
immunoprecipitation with antibodies described in the text. Anti-
insulin immunoprecipitates were boiled for 5 min in gel sample
buffer [1% SDS, 12% glycerol, and 0.0025% Serva Blue in 50 mM
immunoprecipitates were boiled for 5 min in SDS sample buffer at
a final concentration of 2% SDS, 25% glycerol, 0.01% bromophe-
nol blue, and 62.5 mM Tris (pH 6.8), with or with 100 mM DTT as
SDS/PAGE and Western Blot. Tris?tricine?urea? SDS/PAGE was used
for analysis of PI and insulin, as described previously (2). Conven-
tional SDS/12% PAGE was used for analysis of GFP-containing
proteins. For Western blotting, 30 ?g of total protein (or 100
infected islets) were boiled in SDS sample buffer with or without
100 mM DTT. Proteins were then resolved by SDS/PAGE, elec-
trotransferred to nitrocellulose, and blotted with rabbit anti-GFP
followed by an anti-rabbit secondary antibody conjugated to HRP,
with development of the blot by ECL.
Isolation of Mouse Pancreatic Islets. Pancreata removed from
5-week-old male Akita mice (blood glucose at time of euthanasia:
11 mM), and nonmutant male littermates were digested in 2 mg/ml
collagenase-P in Hanks’s balanced salt solution containing calcium
and magnesium in a shaking water bath for 30 min at 37°C. After
twice in ice-cold buffer, islets were hand-picked for overnight
recovery in RPMI medium 1640 containing 11.1 mM glucose plus
10% FBS and 1% penicillin–streptomycin.
Metabolic Labeling of Mouse Pancreatic Islets. One hundred fifty
wild-type and Akita islets were washed twice in prewarmed
Met/Cys-deficient medium plus 1% BSA and 10 mM Hepes (pH
7.35). Islets were then metabolically labeled with35S-labeled
amino acids in the same medium for 20 min. In Fig. 4A, labeled
islets were briefly washed once with RPMI medium 1640 con-
taining 10% FBS and then split into three equal portions: two
were directly immersed in lysis buffer containing a proteinase
inhibitor mixture for subsequent analysis under nonreduced or
reduced conditions, and the third was chased for 2 h at 37°C in
RPMI medium 1640 (11.1 mM glucose plus 10% FBS). The
media were collected, and the islets were lysed as above. Lysate
aliquots were evaluated for trichloroacetic acid-precipitable
radioactivity to normalize the immunoprecipitations.
Confocal Imaging of INS-1 Cells. INS-1 cells expressing hProCpep-
GFP or hProC(A7)Y-CpepGFP were seeded at low confluency on
polylysine-treated coverslips and grown in complete media for 1 d.
min, washed three times with PBS, and then mounted onto glass
microscope slides for imaging of GFP by confocal fluorescence
microscopy. For immunofluorescence, fixed cells were permeabil-
ized with 0.1% Nonidet P-40 and then incubated with guinea pig
anti-insulin (1:1,000) and rabbit anti-calnexin (1:1,000) and appro-
priate secondary antibodies (1:1,000) conjugated to aminomethyl-
coumarin acetate and Alexa Fluor 555, respectively.
Steady-State PI and Insulin Content of Transiently Transected INS-
832/13 Cells Expressing hProCpepGFP, hProC(A7)Y-CpepGFP, or a Cy-
tosolic GFP. INS-832/13cellsweretransfected1dafterplatingusing
Lipofectamine (3 ?g of plasmid DNA per well of a six-well plate).
After 48 h, trypsinized cells were washed, pelleted, resuspended in
PBS, and counted/sorted by green fluorescence. A total of 100 ?l
Total PI plus insulin was measured by rat insulin RIA that
cross-reacts with PI and insulins of all species. Processed human
insulin was measured by specific RIA.
This work was supported by National Institutes of Health Grant DK48280
with help from both the Molecular Biology Core and the Morphology/
Image Analysis Core of the University of Michigan Diabetes Research and
Training Center (National Institutes of Health Grant DK20572).
1. Min CY, Qiao ZS, Feng YM (2004) Eur J Biochem 271:1737–1747.
2. Liu M, Li Y, Cavener D, Arvan P (2005) J Biol Chem 280:13209–13212.
3. Zhang K, Kaufman RJ (2006) Neurology 66:S102–S109.
4. Nozaki J, Kubota H, Yoshida H, Naitoh M, Goji J, Yoshinaga T, Mori K, Koizumi
A, Nagata K (2004) Genes Cells 9:261–270.
Biophys Res Commun 324:166–170.
6. Ron D (2002) J Clin Invest 109:443–445.
7. Leroux L, Desbois P, Lamotte L, Duvillie B, Cordonnier N, Jackerott M, Jami J,
Bucchini D, Joshi RL (2001) Diabetes 50:S150–S153.
8. Yoshioka M, Kayo T, Ikeda T, Koizumi A (1997) Diabetes 46:887–894.
9. Wang J, Takeuchi T, Tanaka S, Kubo SK, Kayo T, Lu D, Takata K, Koizumi A, Izumi
T (1999) J Clin Invest 103:27–37.
10. Kayo T, Koizumi A (1998) J Clin Invest 101:2112–2118.
11. Guo ZY, Feng YM (2001) Biol Chem 382:443–448.
12. Izumi T, Yokota-Hashimoto H, Zhao S, Wang J, Halban PA, Takeuchi T (2003)
13. OyadomariS,KoizumiA,TakedaK,GotohT,AkiraS,ArakiE,MoriM(2002) JClin
15. Pouli AE, Kennedy HJ, Schofield JG, Rutter GA (1998) Biochem J 331:669–675.
16. Ohara-Imaizumi M, Nakamichi Y, Tanaka T, Ishida H, Nagamatsu S (2002) J Biol
17. Liu M, Ramos-Castan ˜eda J, Arvan P (2003) J Biol Chem 278:14798–14805.
18. Irminger JC, Meyer K, Halban P (1996) Biochem J 320:11–15.
19. Molinete M, Lilla V, Jain R, Joyce PBM, Gorr S-U, Ravazzola M, Halban PA (2000)
20. Matveyenko AV, Butler PC (2006) Diabetes 55:2106–2114.
21. Butler AE, Janson J, Bonner-Weir S, Ritzel R, Rizza RA, Butler PC (2003) Diabetes
22. Westermark P, Engstrom U, Johnson KH, Westermark GT, Betsholtz C (1990) Proc
Natl Acad Sci USA 87:5036–5040.
23. Herbach N, Rathkolb B, Kemter E, Pichl L, Klaften M, Hrabe ´ de Angelis M, Wolf
E, Aigner B, Wanke R (2007) Diabetes 56:1268–1276.
24. Zuber C, Fan JY, Guhl B, Roth J (2004) FASEB J 18:917–919.
25. Yoshinaga T, Nakatome K, Nozaki J, Naitoh M, Hoseki J, Kubota H, Nagata K,
Koizumi A (2005) Biol Chem 386:1077–1085.
26. Hohmeier HE, Mulder H, Chen G, Henkel-Rieger R, Prentki M, Newgard CB (2000)
27. Neerman-Arbez M, Halban PA (1993) J Biol Chem 268:16248–16252.
28. Arakawa K, Ishihara T, Oku A, Nawano M, Ueta K, Kitamura K, Matsumoto M,
Saito A (2001) Br J Pharmacol 132:578–586.
www.pnas.org?cgi?doi?10.1073?pnas.0702697104Liu et al.