DEVELOPMENTAL DYNAMICS 233:946 –953, 2005
RESEARCH ARTICLE
NeuroD1 in the Endocrine Pancreas:
Localization and Dual Function as an
Activator and Repressor
P. Itkin-Ansari,1* E. Marcora,2 I. Geron,1 B. Tyrberg,1 C. Demeterco,1 E. Hao,1 C. Padilla,1
C. Ratineau,3 A. Leiter,3 J.E. Lee,2 and F. Levine1
The basic helix–loop– helix transcription factor NeuroD1 regulates cell fate in the nervous system but
previously has not been considered to function similarly in the endocrine pancreas due to its reported
expression in all islet cell types in the newborn mouse. Because we found that NeuroD1 potently represses
somatostatin expression in vitro, its pattern of expression was examined in both strains of mice in which
lacZ has been introduced into the NeuroD1 locus by homologous recombination. Analysis of adult
transgenic mice revealed that NeuroD1 is predominantly expressed in ␤-cells and either absent or
expressed below the limit of lacZ detection in mature ␣-, ␦-, or PP cells. Consistent with a previous report,
NeuroD1 colocalizes with glucagon as well as insulin in immature islets of the newborn mouse. However, no
colocalization of NeuroD1with somatostatin was detected in the newborn. In vitro, ectopic expression of
NeuroD1 in TRM-6/PDX-1, a human pancreatic ␦-cell line, resulted in potent repression of somatostatin
concomitant with induction of the ␤-cell hormones insulin and islet amyloid polypeptide. Additionally,
NeuroD1 induced expression of Nkx2.2, a transcription factor expressed in ␤- but not ␦-cells. Transfection
studies using insulin and somatostatin promoters confirm the ability of NeuroD1 to act as both a
transcriptional repressor and activator in the same cell, suggesting a more complex role for NeuroD1 in the
establishment and/or maintenance of mature endocrine cells than has been recognized previously.
Developmental Dynamics 233:946 –953, 2005. © 2005 Wiley-Liss, Inc.
Key words: NeuroD1; PDX-1; ␤-cell; ␦-cell; differentiation; pancreas; islet; insulin; somatostatin
Received 2 December 2004; Revised 15 March 2005; Accepted 19 March 2005
INTRODUCTION
Pancreatic ␤-cells, the only cell type
that exhibits glucose-responsive insulin secretion, are destroyed or dysfunctional in all major forms of diabetes mellitus (Mathis et al., 2001;
Robertson et al., 2004). Thus, understanding the process by which these
cells arise from precursors in the pan-
creas has been a major focus of research, with the long-term goal of developing a cell replacement therapy
for diabetes.
During embryonic development,
pancreatic endocrine cell precursors
transiently express the basic helix–
loop– helix (bHLH) factor ngn-3 (Kim
and Hebrok, 2001; Gu et al., 2002).
1
There is controversy about the subsequent pattern of hormone expression.
Whereas some studies, predominantly
using immunohistochemical techniques, have detected coexpression of
multiple major hormones (Teitelman
et al., 1993), lineage tracing studies
have provided a different perspective.
These analyses, in which a hormone
Cancer Center, University of California, San Diego, Stem Cell Program, The Burnham Institute, La Jolla, California
Department of Molecular, Cellular, and Developmental Biology, University of Colorado at Boulder, Boulder, Colorado
3
Division of Gastroenterology, GRASP Digestive Disease Research Center, and Tupper Institute, New England Medical Center-Tufts
University School of Medicine, Boston, Massachusetts
Grant sponsor: NIDDK; Grant numbers: DK 55065; DK55283; DK61248; DK43673; DK52870; DK34928; Grant sponsor: JDRF.
*Correspondence to: Dr. Pamela Itkin-Ansari, Cancer Center, UCSD School of Medicine-0816, La Jolla, CA. 92093-0816.
E-mail: [email protected]
2
DOI 10.1002/dvdy.20443
Published online 19 May 2005 in Wiley InterScience (www.interscience.wiley.com).
© 2005 Wiley-Liss, Inc.
NeuroD1 IN ENDOCRINE DIFFERENTIATION 947
promoter drives cre recombinase to
mark cells that activate the promoter
at any point during development, indicate that ␣- and ␤-cells diverge
early, before the onset of hormone expression (Herrera, 2002), in a cell fate
choice controlled by reciprocal expression of the transcription factors Arx,
which is important in ␣-cell formation,
and Pax4, which is important for ␤and ␦-cell formation (Collombat et al.,
2003). The mechanism by which selective hormone expression between ␦and ␤-cells is achieved remains obscure, given that transcription factors
known to activate somatostatin expression, e.g., PDX-1 and Pax6, are
also present in ␤-cells (Miller et al.,
1994; Serup et al., 1996; Sander et al.,
1997). Therefore, for any model of the
␤-cell development to be correct, it
must incorporate not only a mechanism to specifically activate the insulin gene, but also a mechanism to repress other pancreatic hormones.
Overexpression and targeted disruption studies have established a
crucial role for NeuroD1 in regulating
brain and pancreatic islet development (Lee et al., 1995; Naya et al.,
1995, 1997; Miyata et al., 1999;
Schwitzgebel et al., 2000). The NeuroD1 null mouse exhibits a mild reduction of ␦-cells, a moderate loss of
␣-cells, and a severe reduction of
␤-cells (Naya et al., 1997). This profound effect on ␤-cell number signifies
a critical role for NeuroD1 in addition
to insulin gene transactivation in the
␤-cell, because insulin is not required
for the ␤-cell formation or maintenance (Duvillie et al., 1997). However,
the relatively modest effect on ␣- and
␦-cells was puzzling in light of the reported expression of NeuroD1 in those
cells in newborn animals, prompting
the analysis of NeuroD1 expression
and function in the adult pancreas.
Although NeuroD1 has no known
function in ␦-cells, in the ␤-cell, it
binds as a heterodimer with Class A
bHLH proteins, such as E12, E47, or
HEB, to E-boxes within the insulin
promoter (Rudnick et al., 1994). The
bHLH complex acts synergistically
with PDX-1 to induce insulin gene expression (German et al., 1992b; Glick
et al., 2000). The studies presented
here address the role of the tissue restricted bHLH transcription factor
NeuroD1 in ␤- and ␦- cells.
RESULTS
NeuroD1 Expression Is
Predominantly Restricted to
␤-Cells in the Adult
Pancreas In Vivo
Previously published data from a NeuroD/lacZ knock-in mouse, in which
one allele of the NeuroD1 coding sequence was replaced by lacZ, reported
NeuroD1 expression in all four endocrine cell types of the newborn mouse
(Naya et al., 1997). At birth, the null
mouse exhibits a marked reduction in
␤-cells, but a more modest reduction
in the number of glucagon- or somatostatin-expressing cells. This finding
predicted that NeuroD1 is either absent
from or unimportant in maintaining ␣and ␦-cells. To address this issue, a
careful analysis of the coexpression of
lacZ (representing NeuroD1) with
pancreatic hormones was performed
in both the original strain of lacZ/
NeuroD1 knock-in mice (Naya et al.,
1997) as well as an independently derived strain of lacZ/NeuroD1 knock-in
mice (Miyata et al., 1999).
The only published study of NeuroD1 expression was performed in the
newborn pancreas. Consistent with
that study, we found NeuroD1 in virtually all insulin-expressing cells
(Figs. 1, 2). However, we did not find
NeuroD1 in most somatostatin-expressing cells (Fig. 1, 2). The profile of
glucagon and NeuroD1 expression
was more complex. In islets with a
mature structure in which ␣- and
␦-cells were peripherally situated,
NeuroD1 did not colocalize with glucagon. However, in islets with an immature structure, in which the endocrine tissue appeared as streams
rather than defined, round clusters,
colocalization of glucagon with lacZ
was consistently observed in a minority of cells (Fig. 1). This concurs with
single-cell transcript analysis from
the embryonic day (E) 10.5 mouse fetal pancreas, which identified NeuroD1-positive cells expressing glucagon (Chiang and Melton, 2003).
In the adult mouse, confocal microscopy of thin sections demonstrated
that lacZ (NeuroD1) was unambiguously absent (or expressed at levels
below the detection limit of the assay)
from at least 90% of somatostatin, glucagon, or pancreatic polypeptide pro-
ducing cells, and present in greater
than 90% of ␤-cells in the adult pancreas (Fig. 2A). Additional analysis by
deconvolution microscopy with a second set of hormone antibodies combined with ␤-galactosidase immunostaining further confirmed the
selective expression of NeuroD1 in
␤-cells (⬍5% colocalization of somatostatin or glucagon with ␤-galactosidase; Fig. 2B). Thus, the pattern of
NeuroD1 expression in the endocrine
pancreas differs between the newborn
and adult.
NeuroD1 and PDX-1 Induce
Expression of the
Endogenous Insulin Gene in
a Human ␦-Cell Line
TRM-6 is a human cell line that expresses high levels of somatostatin in
response to PDX-1 and cell– cell contact (Itkin-Ansari et al., 2000). Reverse
transcriptase-polymerase chain reaction (RT-PCR) analysis revealed that
NeuroD1 is not expressed in those cells
(Fig. 3A). To investigate the role of NeuroD1 in an in vitro model, NeuroD1 was
introduced into both TRM-6 and TRM6/PDX-1 cells by retrovirus-mediated
gene transfer (Fig. 3A).
Coexpression of the transcriptional
activators NeuroD1 and PDX-1 elicited insulin gene expression as detected by RT-PCR in cultures of TRM6/PDX-1/NeuroD1 cells (Fig. 3B).
Neither NeuroD1 nor PDX-1 alone
was sufficient to induce detectable levels of insulin mRNA. The stimulatory
effect of NeuroD1 on the endogenous
insulin gene was apparent in both cultures of TRM-6/PDX-1 cells infected at
high multiplicities of infection (MOI)
with NeuroD1 retrovirus as well as
clones stably expressing NeuroD1.
However, insulin protein could not be
detected consistently in TRM-6/PDX1/NeuroD1 cells. This finding is not
surprising, given that important
␤-cell transcription factors such as
MafA, which binds to the RIPE3b1 element in the insulin promoter, are not
expressed in TRM-6 (Itkin-Ansari et
al., 2003). Glucokinase expression was
not detected in TRM-6/PDX-1 with or
without NeuroD1 (not shown), despite
reports that PDX-1 activates the glucokinase gene (Watada et al., 1996).
948 ITKIN-ANSARI ET AL.
Fig. 1. Colocalization of NeuroD1 with islet hormones in newborn mice. Confocal analysis of
␤-galactosidase activity identified by 5-bromo4-chloro-3-indolyl-␤-D-galactopyranoside (Xgal) histochemistry to detect NeuroD1 (left), immunohistochemistry
for
islet
hormones
(middle), and merged images (right). Analyses
of glucagon/NeuroD1 colocalization are represented in fully formed islets (row 3) and an
immature stream of endocrine tissue (row 4).
Scale bar ⫽ 10 ␮m.
Fig. 2. Colocalization of NeuroD1 and pancreatic islet hormones in adult mice. A: Confocal
analysis of ␤-galactosidase activity identified by
5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) histochemistry to detect NeuroD1 (left), immunohistochemistry for islet hormones (middle), and merged images (right). B:
Deconvolution imaging of double immunostaining for ␤-galactosidase to identify NeuroD1expressing cells (left column, green) and islet
hormones (middle column, red) in islets of adult
knockin mice heterozygous for NeuroD1/lacZ.
In overlay images (right column), yellow reflects
colocalization. Insulin and somatostatin images
are ⫻40, glucagon and pancreatic polypeptide
images are ⫻20 original optical magnification.
Scale bar ⫽ 10 ␮m.
Fig. 1.
Fig. 2.
NeuroD1 IN ENDOCRINE DIFFERENTIATION 949
Fig. 3. Activation of the endogenous insulin
gene by NeuroD1. A: Reverse transcriptase-polymerase chain reaction (RT-PCR) for NeuroD1 on
TRM-6 (T6, lane 1), TRM-6/PDX-1(T6P, lane 2),
TRM-6/NeuroD1 (T6N, lane 3), TRM-6/PDX-1/
NeuroD1 (T6PN, lane 4), and NeuroD1 plasmid
(lane 5). Because NeuroD1 primers do not cross
an intron, samples without reverse transcriptase
were also analyzed and demonstrated the absence of any contaminating genomic DNA (not
shown). RT-PCR for the housekeeping gene porphobilinogen deaminase (PD) provides evidence
of equivalent samples. B: RT-PCR analysis of insulin gene expression in TRM-6/PDX-1/NeuroD1
(T6PN, lane 1), TRM-6/PDX-1 (T6P, lane 2), pancreas (lane 3), and Jurkat cells (lane 4).
Fig. 4. Repression of somatostatin by NeuroD1.
Somatostatin mRNA level (⫾ SD) was determined
by real-time reverse transcriptase-polymerase
chain reaction in cell aggregates of TRM-6, TRM6/PDX-1, and TRM-6/PDX-1/NeuroD1 cells. The
parental TRM-6 cells do not express PDX-1 and
so do not exhibit somatostatin transcription.
NeuroD1 Represses
Somatostatin Expression
In the adult pancreas, mature endocrine cells express only a single major
hormone (Alpert et al., 1988; de Krijger
et al., 1992; Lukinius et al., 1992; Upchurch et al., 1994; Fernandes et al.,
1997). Because NeuroD1 activated the
insulin promoter in a ␦-cell line, it was
of interest to determine whether it had
any effect on somatostatin expression.
Infection of somatostatin-expressing
TRM-6/PDX-1 cells with NeuroD1 resulted in a profound reduction in somatostatin mRNA, returning it to within
twofold of the extremely low baseline
levels found in parental TRM-6 cells
(Fig. 4; Itkin-Ansari et al., 2000).
Fig. 5. Transfection studies of basic helix–
loop– helix regulation of somatostatin versus insulin promoters. A,B: HeLa cells transfected
with NeuroD1, MyoD, or control expression
plasmids, in addition to a PDX-1 expression
plasmid were assayed for insulin and somatostatin gene activation using the insulin promoter reporter construct FFCAT (A) or somatostatin promoter CAT (B), respectively. C:
NeuroD1, wild-type E2A (E2A), E2A deleted for
both transactivating domains (E2A-TAD), or
E2A with a mutation in the DNA-binding domain
(E2A-BM) were examined on the somatostatin
promoter. Values of CAT expression (triplicates ⫾ SD) were interpolated from a standard
curve generated in a CAT enzyme-linked immunosorbent assay. The data are representative of
multiple independent experiments.
To determine the mechanism by
which NeuroD1 achieves this bifunctional role, transfection studies were
conducted. HeLa cells were utilized, as
they have been used extensively to examine the activity of pancreatic hormone promoters (Qiu et al., 1998). Consistent with previous reports (Glick et
al., 2000), in combination with PDX-1,
NeuroD1 potently stimulated the insulin mini-enhancer reporter FF-CAT,
which contains multimerized bHLH
(Far) and homeodomain (Flat) ele-
ments. In contrast, MyoD, a musclespecific bHLH protein, was unable to
activate FF-CAT, demonstrating that
the activating effect of NeuroD1 on the
insulin promoter is specific (Fig. 5A). To
confirm the studies performed in TRM6/PDX-1 and HeLa cells, NeuroD1 and
MyoD were stably expressed in RIN
14B cells, a somatostatin-expressing
RIN variant (Bhathena et al., 1984). In
that cell line, NeuroD1, but not MyoD,
activated the endogenous insulin I
gene, whereas NeuroD1 potently, and
MyoD partially, repressed endogenous
somatostatin expression.
To study the repressive function of
NeuroD1, 1,180 bp of the human somatostatin promoter was cloned upstream of a CAT reporter (SOM-CAT).
This region of the human somatostatin promoter contains seven E-box
consensus sequences, although the
relative importance of these in the
control of somatostatin gene expression is not known. As expected, SOMCAT was activated by PDX-1 in a
dose-responsive manner (not shown).
Moreover, in accordance with our findings on the endogenous somatostatin
promoter in TRM-6/PDX-1 cells, cotransfection of NeuroD1 but not control plasmid diminished activation of
the somatostatin promoter by PDX-1
(Fig. 5B). Surprisingly, the muscle determination bHLH factor MyoD also
inhibited somatostatin expression,
suggesting that a conserved bHLH
function rather than a unique effect of
NeuroD1 is responsible for the repressive effect on somatostatin expression.
These findings indicate that the mechanisms by which bHLH transcription
factors act on the insulin and somatostatin promoters differ significantly,
such that specific bHLH factors are
required to activate the insulin promoter, whereas a region common to
multiple bHLH factors rather than a
specific factor may be sufficient to repress somatostatin transcription.
DNA Binding but Not the
Transcriptional Activation
Domain Is Required for
bHLH-Mediated Repression
of Somatostatin Promoter
Activity
To assess the bHLH functional elements required for repression of soma-
950 ITKIN-ANSARI ET AL.
tostatin gene expression, the ability of
deletion mutants to repress somatostatin promoter activity was examined. Mutants in which transactivation domains were deleted retained
substantial repressive activity on the
somatostatin promoter (Fig. 5C).
However, introduction of a point mutation in the DNA binding domain,
which ablates DNA binding while retaining dimerization activity (Engel
and Murre, 1999), eliminated repressive activity on the somatostatin promoter. In fact, a bHLH DNA binding
mutant consistently enhanced somatostatin promoter activity.
PDX-1 and NeuroD1 Act in
Synergy to Induce NKX2.2
and Islet Amyloid
Polypeptide Expression
If NeuroD1 plays a role in lineage determination between ␦- and ␤-cells, as
opposed to a relatively restricted role
in the ␤-cell, specific activation of additional genes expressed in ␤- but not
␦-cells would be expected to occur. To
assess whether NeuroD1 was responsible for a more global effect on TRM-6
cells, the expression of the hormone
islet amyloid polypeptide (IAPP), which
is cosecreted with insulin by ␤-cells,
was examined. IAPP gene expression
was analyzed in PDX-1- and NeuroD1expressing cells and found to be up-regulated by the combination of factors
(Fig. 6A). In contrast to insulin, IAPP
expression was also stimulated, albeit
to a lesser extent, in TRM-6 cells expressing PDX-1 or NeuroD1 alone. Although IAPP induction by PDX-1 previously has been reported (Macfarlane et
al., 2000), it was not known to be a
downstream target of NeuroD1.
Expression of the homeodomain
factor Nkx2.2 was examined because
it is expressed in ␤- but not ␦-cells in
vivo (Sussel et al., 1998). As expected for a ␦-cell line, TRM-6/PDX-1
cells do not express Nkx2.2. However, Nkx2.2 was activated by NeuroD1, consistent with its expression
in primary ␤-cells (Fig. 6B). Thus,
even in the absence of other notable
␤-cell transcription factors, NeuroD1
was able to activate multiple important ␤-cell specific genes.
Fig. 6. Downstream targets of NeuroD1. A,B:
Quantitative reverse transcriptase-polymerase
chain reaction (RT-PCR) analysis of the effect of
NeuroD1 on the expression of islet amyloid
polypeptide (IAPP, A) and NKX2.2 (B). 1. IAPP
RT-PCR was performed on TRM-6 and TRM-6/
PDX-1 cells ⫾NeuroD1. Gene expression of
Nkx2.2 by RT-PCR was analyzed on TRM-6/
PDX-1 cells ⫾NeuroD1. Error bars represent
the standard deviation of independently performed PCRs.
DISCUSSION
The studies presented here provide
new insight into the complement of
genes regulated by NeuroD1 in the
␤-cell. Moreover, the ability of NeuroD1 to repress somatostatin expression suggests an additional role for
NeuroD1 in the establishment and
maintenance of the mature state in
which each endocrine cell type expresses only a single hormone.
Efforts to examine the pattern of
NeuroD1 expression in human islets
have been complicated by a lack of
reliable antibodies. Thus, information
on NeuroD1 localization is dependent
upon analysis of transgenic mice in
which lacZ has been inserted into the
endogenous NeuroD1 locus. In both
strains, the pattern of lacZ expression
faithfully mimics that of the endogenous locus (Naya et al., 1997; Miyata
et al., 1999). In the studies presented
here, extensive analysis of those
strains revealed that NeuroD1 is predominantly restricted to ␤-cells in the
adult pancreas. This finding included
reanalysis by some of the original authors of the first strain reported (Naya
et al., 1997) as well as examination of
an independently derived strain
(Miyata et al., 1999). Furthermore, two
different techniques, hormone staining
with lacZ immunohistochemistry in
conjunction with deconvolution microscopy (Fig. 2B) and histochemical detection using 5-bromo-4-chloro-3-indolyl␤-D-galactopyranoside
(X-gal)
in
conjunction with confocal microscopy
(Figs. 1, 2A) were used, providing a
much more detailed analysis of the
pattern of NeuroD1 expression than
previously undertaken (Naya et al.,
1997).
Our finding that NeuroD1 is expressed in a subset of glucagon-positive cells during development is consistent with the previous study
(performed only in newborn mice) and
suggests that NeuroD1 is induced by
ngn-3 in ␣-cell precursors but is then
repressed in mature ␣-cells. Furthermore, colocalization of NeuroD1 in
some glucagon-expressing cells (Fig.
1) and cell lines (Jensen et al., 1996)
indicates that glucagon is not actively
repressed by NeuroD1. In fact, NeuroD1 has been reported to activate the
glucagon promoter (Dumonteil et al.,
1998). The lack of NeuroD1 in adult
␣-cells demonstrates that it is not required for glucagon expression.
At the same stage when there was
identifiable colocalization between
NeuroD1 and glucagon, we did not observe colocalization of somatostatin
and NeuroD1, suggesting that ␦-cell
precursors either never express NeuroD1 or do so extremely transiently
(Fig. 7). This result differs from that
previously published, but close examination of the published data does not
reveal convincing colocalization between lacZ and somatostatin (Fig. 3 in
Naya et al., 1997).
To further study the role of NeuroD1 in vitro, the human pancreatic
␦-cell line TRM-6/PDX-1 was used (Itkin-Ansari et al., 2000). We found that
NeuroD1 concomitantly induced the
expression of multiple ␤-cell markers,
including insulin, IAPP, and Nkx2.2,
while ablating endogenous somatostatin gene expression. However, a simple model in which NeuroD1 expression is sufficient to mediate a switch
between the ␦- and ␤-cell lineages is
unlikely, because the NeuroD1 null
mutant does not exhibit an increase in
the number of ␦-cells (Naya et al.,
1997). In the brain and the pancreas,
NeuroD1 IN ENDOCRINE DIFFERENTIATION 951
cells lacking NeuroD1 have a greatly
increased rate of apoptosis (Naya et
al., 1997; Miyata et al., 1999). Thus, it
is possible that a different bHLH protein, which does not repress somatostatin, is required to suppress apoptosis in ␦-cells, which do not express
NeuroD1. Alternatively, the principal
role of NeuroD1 may be in maintenance of insulin rather than somatostatin expression in mature ␤-cells
by active repression of the somatostatin promoter, which might otherwise
be activated by the high PDX-1 levels
in ␤-cells.
The existence of E-box–mediated repression of the somatostatin promoter
has been proposed in both pancreatic
endocrine and nonpancreatic epithelial cell lines, but the identity of the
relevant factors binding to those elements was not determined (Vallejo et
al., 1995). Based upon the lack of NeuroD1 in ␦-cells and its potent repression of the somatostatin gene, NeuroD1 is a candidate for that factor. We
used transfection studies to confirm
the effect of NeuroD1 on the somatostatin promoter and elucidate the
molecular mechanism by which NeuroD1 regulates the somatostatin gene.
DNA binding, but not transactivation,
is required for repression of somatostatin expression.
A recent lineage analysis of ␤-cell
regeneration after partial pancreatectomy indicates that new ␤-cells arise
from preexisting ␤-cells rather than
from stem cells (Dor et al., 2004).
However, in a different model of ␤-cell
regeneration after low-dose streptozotocin to destroy existing ␤-cells, it has
been proposed that new ␤-cells arise
by a process of transdifferentiation
from ␦-cells (Teitelman, 1996; Fernandes et al., 1997; Guz et al., 2001).
Streptozotocin-induced destruction of
␤-cells resulted in the initial appearance of a large population of somatostatin-positive, PDX-1–positive cells, suggested to represent an islet precursor
cell type, followed by regenerated insulin (only) -positive ␤-cells (Teitelman
and Lee, 1987; Teitelman, 1996). A
close relationship between ␦- and ␤cells is also demonstrated by the finding
that they but not ␣-cells arise from a
Pax4-positive precursor (Sosa-Pineda et
al., 1997). Furthermore, ␦- and ␤-cells
are strikingly similar to one another but
distinct from ␣-cells in their response to
Fig. 7. Model of the role of NeuroD1 in endocrine cell lineage determination. Pancreatic endocrine
cells develop from ngn-3–positive progenitors. The ␤- and ␦-cells arise from a PAX4-positive
precursor. In our model, the ␤- and ␦-cell lineages diverge when NeuroD1 is selectively expressed
in ␤- rather than ␦-cells. [Color figure can be viewed in the online issue, which is available at
www.interscience.wiley.com.]
secretogogues (Berts et al., 1996). Overall, it is possible that multiple pathways
for ␤-cell differentiation exist in the
adult pancreas, perhaps differentially
activated by distinct stimuli. If so, it will
be important to elucidate which involve
NeuroD1.
In summary, we establish that NeuroD1 is predominantly restricted to
␤-cells in mature islets. In a human
pancreatic ␦-cell line, NeuroD1 acts in
both a positive and negative manner,
activating ␤-cell specific genes, including insulin and IAPP, while repressing somatostatin. Thus, the studies
presented here suggest a role for NeuroD1 in the establishment and/or
maintenance of the ␤- vs. ␦-cell phenotype (Fig. 7).
EXPERIMENTAL
PROCEDURES
Cell Culture
Cell cultures were maintained in
DMEM, 5.5 mM glucose, supplemented with 10% fetal bovine serum.
Cell aggregation was induced as pre-
viously described (Itkin-Ansari et al.,
2000).
Transfection Studies
A 1,190-bp fragment 5⬘ of the start
codon of the somatostatin gene was amplified from genomic DNA using the
primers SP5⬘primer1-gtttacgcgtggagatcaggcagagc and SP3⬘primer1-gtttagatctcgaaagccgagctgg, and cloned upstream of the CAT gene in the pCAT3BASIC vector (Promega). The FFCAT
plasmid contains a multimerized insulin promoter mini-enhancer (German et
al., 1992a). HeLa cell transfections, by
CaPO4 precipitation, contained 4 ␮g of
plasmid: 1.5 ␮g of CAT reporter, 0.6 ␮g
of PDX-1 retroviral plasmid (ItkinAnsari et al., 2000), and 0.2 ␮g of RSVluciferase, 0.7 ␮g of control plasmid,
and 1.0 ␮g of bHLH-expressing or control plasmid. Trichostatin A (1 ␮g/ml)
was added to some internally controlled
transfection studies. This enhanced activity from the otherwise weak somatostatin promoter construct but did not
change the signal to noise ratio. CAT
assays were performed using a CAT en-
952 ITKIN-ANSARI ET AL.
zyme-linked immunosorbent assay kit
(Roche Diagnostics, Indianapolis, IN).
Luciferase assays were performed using the Luciferase Reporter kit (Roche
Diagnostics) to ensure equivalence in
transfection efficiency between the different conditions.
Analysis and Quantitation of
mRNA
PCR primers for NeuroD1 (Dufayet de
la Tour et al., 2001), somatostatin (Itkin-Ansari et al., 2000), and PDX-1 (Beattie et al., 1999) have been described
previously. Primers were as follows for
insulin: forward 5⬘atcagaagaggccatcaagc, reverse 5⬘tggttcaagggctttattcca;
IAPP: forward 5⬘gaaccatctgaaagctacaccc, reverse 5⬘cattgtcctctaaaggggca;
NKX2.2: forward 5⬘gaaccccttctacgacagca, reverse 5⬘gtcattgtccggtgactcgt.
RT-PCR for the housekeeping genes
porphobilinogen
deaminase
(PD;
Shimizu et al., 1994) or glyceraldehyde3-phosphate dehydrogenase were used
for normalization of samples.
RT-PCR used cDNA corresponding
to 100 –300 ng of RNA as previously
described (Itkin-Ansari et al., 2000).
Pancreas RNA was obtained from
Clontech Labs, Inc. (Palo Alto, CA).
Real-time PCR used SYBR Green I
dye (Sigma, St. Louis, MO) on the ABI
PRISM 7700 (Applied Biosystems,
Foster City, CA) or Opticon DNA Engine Real-Time Thermal Cycler (MJ
Research, Waltham, MA).
Immunohistochemistry
Combined immunohistochemistry for
hormones and X-gal histochemistry
(for NeuroD1) in newborn and adult
mice: 5-␮m frozen sections blocked
with 50 mM glycine (Sigma), 2% donkey serum (Jackson ImmunoResearch
Laboratories), 2% bovine serum albumin (Sigma) in phosphate buffered saline (PBS) were stained with rabbit
anti-insulin 1:50 (Santa Cruz, Santa
Cruz, CA), goat anti-somatostatin
1:500 (Santa Cruz), or mouse anti-glucagon (Sigma) and Cy5 fluor-labeled
anti-Goat/mouse/rabbit (Jackson ImmunoResearch 1:200 dilution). For Xgal histochemistry, slides were fixed
with 1.25% glutaraldehyde (Sigma) in
PBS, and incubated in: 400 ␮g/ml Xgal (US Biological, Swampscott, MA),
5 mM ferriferrocyanide, and 1 mM
MgCl2. Specimens were examined in a
laser scanning confocal microscope
system (MRC 1024 MP, Bio-Rad Laboratories, Richmond, CA).
Double immunohistochemistry for hormones and ␤-galactosidase in adult mice
was performed. Adult heterozygous NeuroD1/lacZ knockin mice were anesthetized and perfused with ice-cold fix solution (4% paraformaldehyde in PBS).
Pancreata were incubated in fix solution
overnight at 4°C. Cryostat sections (20
␮m) were stained by four sequential
rounds of fluorescent immunohistochemistry. In each round slides were washed
four times with PBS for 5 min at room
temperature, incubated in blocking solution (PBS with 0.1% Triton X-100/5% normal goat serum) for 30 min at room temperature, and incubated in antibody
solution overnight at 4°C (first and third
round) or for 1 hr at room temperature
(second and fourth round). Slides were
mounted in Vectashield (Vector Labs,
Burlingame, CA). Digital images of
stained sections were captured using a
fluorescence microscope with a digital
camera (Nikon, Tokyo, Japan) and deconvoluted (nearest neighbors) using SlideBook software (Intelligent Imaging Innovations, Denver, CO). The following
antibodies were used: first round, rabbit
anti–␤-galactosidase (1:1,000, 5 Prime–3
Prime, Boulder, CO); second round, fluorescein anti-rabbit (1:100, Vector); third
round, guinea pig anti-insulin (1:100,
Linco, St. Charles, MO), rabbit anti-glucagon (1:50, Zymed, San Francisco, CA),
rabbit anti-pancreatic polypeptide (1:200,
Peninsula Laboratories, San Carlos, CA),
or rabbit anti-somatostatin (1:50,
Zymed); fourth round, Texas red antiguinea pig or Texas red anti-rabbit (1:
100,Vector). Appropriate controls without
third-round antibody ensured specific
staining.
Virus Production
Cells infected with MSCV-PDX-1
were selected in hygromycin to ensure
that all of the cells expressed the
transgene (Itkin-Ansari et al., 2000).
The human NeuroD1 gene, a kind gift
from Dr. Stephen Tapscott (Fred
Hutchinson Cancer Research Center,
Seattle, WA), was cloned into the
MSCV/hph retroviral vector (Hawley
et al., 1994). VSV-G pseudotyped retrovirus was prepared as previously
described (Burns et al., 1993). Infec-
tions were done at high MOI. All of 15
clones derived from this infection expressed the NeuroD1 transgene.
ACKNOWLEDGMENTS
We thank Dr. Andrew Lassar for the
MyoD expression plasmid, Dr. Stephen Tapscott for the human NeuroD1/BETA2 gene, and Drs. Bill Romano and Cornelis Murre for the E2A
expression plasmids. We thank Stuart
Bossie for expert technical assistance.
We thank Jennifer Freund, Robbin
Newlin, and Dr. Edward Monosov of
the Burnham Institute for their expertise in microscopy and Nan Tang for
constructing the Som-CAT plasmid.
F.L. and A.L. were funded by the
NIDDK. F.L. holds equity in PanCel,
Inc. B.T. is a Hillblom Research Fellow.
REFERENCES
Alpert S, Hanahan D, Teitelman G. 1988.
Hybrid insulin genes reveal a developmental lineage for pancreatic endocrine
cells and imply a relationship with neurons. Cell 53:295–308.
Beattie GM, Itkin-Ansari P, Cirulli V, Leibowitz G, Lopez AD, Bossie S, Mally MI,
Levine F, Hayek A. 1999. Sustained proliferation of PDX-1⫹ cells derived from
human islets. Diabetes 48:1013–1019.
Berts A, Ball A, Dryselius G, Gylfe E, Hellman B. 1996. Glucose stimulation of somatostatin-producing islet cells involves
oscillatory Ca2⫹ signaling. Endocrinology 137:693–697.
Bhathena SJ, Awoke S, Voyles NR,
Wilkins SD, Recant L, Oie HK, Gazdar
AF. 1984. Insulin, glucagon, and somatostatin secretion by cultured rat islet
cell tumor and its clones. Proc Soc Exp
Biol Med 175:35–38.
Burns JC, Friedmann T, Driever W,
Burrascano M, Yee JK. 1993. Vesicular
stomatitis
virus
G
glycoprotein
pseudotyped retroviral vectors: concentration to very high titer and efficient
gene transfer into mammalian and nonmammalian cells [see comments]. Proc
Natl Acad Sci U S A 90:8033–8037.
Chiang MK, Melton DA. 2003. Single-cell
transcript analysis of pancreas development. Dev Cell 4:383–393.
Collombat P, Mansouri A, Hecksher-Sorensen J, Serup P, Krull J, Gradwohl G,
Gruss P. 2003. Opposing actions of Arx
and Pax4 in endocrine pancreas development. Genes Dev 17:2591–2603.
de Krijger RR, Aanstoot HJ, Kranenburg
G, Reinhard M, Visser WJ, Bruining GJ.
1992. The midgestational human fetal
pancreas contains cells coexpressing islet hormones. Dev Biol 153:368 –375.
Dor Y, Brown J, Martinez OI, Melton DA.
2004. Adult pancreatic beta-cells are
NeuroD1 IN ENDOCRINE DIFFERENTIATION 953
formed by self-duplication rather than
stem-cell differentiation. Nature 429:41–
46.
Dufayet de la Tour D, Halvorsen T, Demeterco C, Tyrberg B, Itkin-Ansari P, Loy
M, Yoo S-J, Hao E, Bossie S, Levine F.
2001. ␤-cell differentiation from a human pancreatic cell line in vitro and in
vivo. Mol Endocrinol 15:476 –483.
Dumonteil E, Laser B, Constant I, Philippe
J. 1998. Differential regulation of the
glucagon and insulin I gene promoters
by the basic helix-loop-helix transcription factors E47 and BETA2. J Biol
Chem 273:19945–19954.
Duvillie B, Cordonnier N, Deltour L, Dandoy-Dron F, Itier JM, Monthioux E, Jami
J, Joshi RL, Bucchini D. 1997. Phenotypic alterations in insulin-deficient mutant mice. Proc Natl Acad Sci U S A
94:5137–5140.
Engel I, Murre C. 1999. Ectopic expression
of E47 or E12 promotes the death of E2Adeficient lymphomas. Proc Natl Acad Sci
U S A 96:996 –1001.
Fernandes A, King LC, Guz Y, Stein R,
Wright CV, Teitelman G. 1997. Differentiation of new insulin-producing cells is
induced by injury in adult pancreatic islets. Endocrinology 138:1750 –1762.
German MS, Moss LG, Wang J, Rutter WJ.
1992a. The insulin and islet amyloid
polypeptide genes contain similar cellspecific promoter elements that bind
identical beta-cell nuclear complexes.
Mol Cell Biol 12:1777–1788.
German MS, Wang J, Chadwick RB, Rutter WJ. 1992b. Synergistic activation of
the insulin gene by a LIM-homeo domain
protein and a basic helix-loop-helix protein: building a functional insulin
minienhancercomplex.GenesDev6:2165–
2176.
Glick E, Leshkowitz D, Walker MD. 2000.
Transcription factor BETA2 acts cooperatively with E2A and PDX1 to activate
the insulin gene promoter. J Biol Chem
275:2199 –2204.
Gu G, Dubauskaite J, Melton DA. 2002.
Direct evidence for the pancreatic lineage: NGN3⫹ cells are islet progenitors
and are distinct from duct progenitors.
Development 129:2447–2457.
Guz Y, Nasir I, Teitelman G. 2001. Regeneration of pancreatic beta cells from intra-islet precursor cells in an experimental model of diabetes. Endocrinology 142:
4956 –4968.
Hawley RG, Lieu FHL, Fong AZC, Hawley
TS. 1994. Versatile retroviral vectors for
potential use in gene therapy. Gene Ther
1:136 –138.
Herrera PL. 2002. Defining the cell lineages
of the islets of Langerhans using transgenic mice. Int J Dev Biol 46:97–103.
Itkin-Ansari P, Demeterco C, Bossie S, de
la Tour DD, Beattie GM, Movassat J,
Mally MI, Hayek A, Levine F. 2000.
PDX-1 and cell-cell contact act in synergy to promote delta-cell development
in a human pancreatic endocrine precursor cell line. Mol Endocrinol 14:814 –822.
Itkin-Ansari P, Geron I, Hao E, Demeterco
C, Tyrberg B, Levine F. 2003. Progress
towards a transplantable human beta-cell
line. Ann N Y Acad Sci 1005:138 –147.
Jensen J, Serup P, Karlsen C, Nielsen TF,
Madsen OD. 1996. mRNA profiling of rat
islet tumors reveals nkx 6.1 as a betacell- specific homeodomain transcription
factor. J Biol Chem 271:18749 –18758.
Kim SK, Hebrok M. 2001. Intercellular signals regulating pancreas development
and function. Genes Dev 15:111–127.
Lee JE, Hollenberg SM, Snider L, Turner
DL, Lipnick N, Weintraub H. 1995. Conversion of Xenopus ectoderm into neurons by NeuroD, a basic helix-loop-helix
protein. Science 268:836 –843.
Lukinius A, Ericsson JLE, Grimelius L,
Korsgren O. 1992. Ultrastructural studies
of the ontogeny of fetal human and porcine
endocrine pancreas, with special reference
to colocalization of the four major islet hormones. Dev Biol 153:376 –385.
Macfarlane WM, Shepherd RM, Cosgrove
KE, James RF, Dunne MJ, Docherty K.
2000. Glucose modulation of insulin
mRNA levels is dependent on transcription factor PDX-1 and occurs independently of changes in intracellular Ca2⫹.
Diabetes 49:418 –423.
Mathis D, Vence L, Benoist C. 2001. betaCell death during progression to diabetes. Nature 414:792–798.
Miller CP, McGehee Jr. RE, Habener JF.
1994. IDX-1: a new homeodomain transcription factor expressed in rat pancreatic islets and duodenum that transactivates the somatostatin gene. EMBO J
13:1145–1156.
Miyata T, Maeda T, Lee JE. 1999. NeuroD
is required for differentiation of the
granule cells in the cerebellum and hippocampus. Genes Dev 13:1647–1652.
Naya FJ, Stellrecht CMM, Tsai M-J. 1995.
Tissue-specific regulation of the insulin
gene by a novel basic helix-loop-helix transcription factor. Genes Dev 9:1009 –1019.
Naya FJ, Huang HP, Qiu Y, Mutoh H, DeMayo FJ, Leiter AB, Tsai MJ. 1997. Diabetes, defective pancreatic morphogenesis, and abnormal enteroendocrine
differentiation in BETA2/neuroD-deficient mice. Genes Dev 11:2323–2334.
Qiu Y, Sharma A, Stein R. 1998. p300 mediates transcriptional stimulation by the
basic helix-loop-helix activators of the insulin gene. Mol Cell Biol 18:2957–2964.
Robertson RP, Harmon J, Tran PO, Poitout
V. 2004. Beta-cell glucose toxicity, lipotoxicity, and chronic oxidative stress in
type 2 diabetes. Diabetes 53(Suppl 1):
S119 –S124.
Rudnick A, Ling TY, Odagiri H, Rutter WJ,
German MS. 1994. Pancreatic beta cells
express a diverse set of homeobox genes.
Proc Natl Acad Sci U S A 91:12203–
12207.
Sander M, Neubuser A, Kalamaras J, Ee
HC, Martin GR, German MS. 1997. Genetic analysis reveals that PAX6 is required for normal transcription of pancreatic hormone genes and islet
development. Genes Dev 11:1662–1673.
Schwitzgebel VM, Scheel DW, Conners JR,
Kalamaras J, Lee JE, Anderson DJ, Sussel L, Johnson JD, German MS. 2000.
Expression of neurogenin3 reveals an islet cell precursor population in the pancreas. Development 127:3533–3542.
Serup P, Jensen J, Andersen FG, Forgensen MC, Blume N, Holst JJ, Madsen
OD. 1996. Induction of insulin and islet
amyloid polypeptide production in pancreatic islet glucagonoma cells by insulin
promoter factor 1. Proc Natl Acad Sci
U S A 93:9015–9020.
Shimizu H, Shimizu C, Burns JC. 1994.
Detection of novel RNA viruses: morbilliviruses as a model system. Mol Cell
Probes 8:209 –214.
Sosa-Pineda B, Chowdhury K, Torres M,
Oliver G, Gruss P. 1997. The Pax4 gene
is essential for differentiation of insulinproducing beta cells in the mammalian
pancreas. Nature 386:399 –402.
Sussel L, Kalamaras J, Hartigan-O’Connor
DJ, Meneses JJ, Pedersen RA, Rubenstein JL, German MS. 1998. Mice lacking the homeodomain transcription factor Nkx2.2 have diabetes due to arrested
differentiation of pancreatic beta cells.
Development 125:2213–2221.
Teitelman G. 1996. Induction of beta-cell
neogenesis by islet injury. Diabetes
Metab Rev 12:91–102.
Teitelman G, Lee JK. 1987. Cell lineage
analysis of pancreatic islet cell development: glucagon and insulin cells arise
from
catecholaminergic
precursors
present in the pancreatic duct. Dev Biol
121:454 –466.
Teitelman G, Alpert S, Polak JM, Martinez
A, Hanahan D. 1993. Precursor cells of
mouse endocrine pancreas coexpress insulin, glucagon and the neuronal proteins tyrosine hydroxylase and neuropeptide Y, but not pancreatic
polypeptide. Development 118:1031–
1039.
Upchurch BH, Aponte GW, Leiter AB.
1994. Expression of peptide YY in all
four islet cell types in the developing
mouse pancreas suggests a common peptide YY-producing progenitor. Development 120:245–252.
Vallejo M, Miller CP, Beckman W, Habener JF. 1995. Repression of somatostatin gene transcription mediated by two
promoter silencer elements. Mol Cell Endocrinol 113:61–72.
Watada H, Kajimoto Y, Miyagawa J, Hanafusa T, Hamaguchi K, Matsuoka T,
Yamamoto K, Matsuzawa Y, Kawamori
R, Yamasaki Y. 1996. PDX-1 induces insulin and glucokinase gene expressions
in alphaTC1 clone 6 cells in the presence
of betacellulin. Diabetes 45:1826 –1831.