BRD7389 paper

Small-molecule inducers of insulin expression in
pancreatic α-cells
Dina Fomina-Yadlina,b,c,1
, Stefan Kubiceka,b,1
, Deepika Walpitab
, Vlado Dancikb,d
, Jacob Hecksher-Sørensenb,2
,
Joshua A. Bittkerb
, Tanaz Sharifniab,e
, Alykhan Shamjib
, Paul A. Clemonsb
, Bridget K. Wagnerb
, and
Stuart L. Schreibera,b,f,3
a
Howard Hughes Medical Institute, b
Broad Institute of Harvard and Massachusetts Institute of Technology, Cambridge, MA 02142; c
Department of Molecular
and Cellular Biology, Harvard University, Cambridge, MA 02138; d
Mathematical Institute, Slovak Academy of Sciences, Košice, 040 01, Slovakia; e
Department
Biological and Biomedical Sciences, Harvard Medical School, Boston, MA 02115; and f
Department of Chemistry and Chemical Biology, Harvard University,
Cambridge, MA 02138
Contributed by Stuart L. Schreiber, July 15, 2010 (sent for review June 22, 2010)
High-content screening for small-molecule inducers of insulin
expression identified the compound BRD7389, which caused α-cells
to adopt several morphological and gene expression features of
a β-cell state. Assay-performance profile analysis suggests kinase
inhibition as a mechanism of action, and we show that biochemical
and cellular inhibition of the RSK kinase family by BRD7389 is likely
related to its ability induce a β-cell-like state. BRD7389 also in-
creases the endocrine cell content and function of donor human
pancreatic islets in culture.
BRD7389 | pancreatic islets | Rsk kinase | transdifferentiation | beta cells
Type 1 diabetes is an autoimmune disease characterized by the
loss of insulin-producing β-cells in pancreatic islets of Lang-
erhans. Islet transplantation into the liver can effectively cure the
disease (1), but is not an ideal treatment due to limited donor
material and immunological complications. An alternative ap-
proach, not yet feasible, is to create new β-cells (2), either by
stepwise differentiation of undifferentiated stem or stem-like cells
(3), or by transdifferentiation (4), the heritable change of cell
identity to an insulin-producing (β-like) cell. The latter approach
could result in a replacement source for the deficient cell type
directly from patient material (either in vivo or ex vivo). Increasing
β-cell mass by small-molecule drug-induced transdifferentiation is
a speculative but exciting approach to treating diabetes—one that
is significantly different from currently available small-molecule
drugs that increase insulin secretion in existing β-cells and are
therefore ineffective in the later stages of type 1 diabetes, in which
most β-cell mass has been lost.
Cell-type specification in the pancreas is regulated by a set of
master regulatory transcription factors that control the transition
from one progenitor cell state to the next, ultimately yielding
mature endocrine cell types in islets (5). Recently, it has been
shown that misexpression of these master regulatory transcription
factors causes direct transdifferentiation between cell types. For
example, ectopic overexpression of a single transcription factor
(Arx) is sufficient for in vivo conversion of β-cells to α-cells in the
adult mouse pancreas (6). Similarly, viral delivery of three tran-
scription factors (Pdx1, Ngn3, MafA) to an adult mouse pancreas
causes the transdifferentiation of acinar cells to β-cells (7). Finally,
in vivo conversion of α-cells to β-cells has recently been achieved in
mature mouse α-cells by ectopic overexpression of Pax4 (8).
Results
Because a single gene is sufficient to induce transdifferentiation of
α-cells to β-cells, we sought to determine whether a small molecule
could have the same effect. Possible readouts for induction of a β-
cell state include insulin production and insulin secretion. We
chose to target the production of insulin protein because we
imagined that this would be more feasible to achieve in the course
of a 3-d small-molecule treatment than insulin secretion. To that
end, we developed a high-content, cell-based assay to detect in-
sulin protein expression in the mouse α-cell line αTC1. Normal
mouse α-cells are insulin negative, but have the ability to adopt
a β-cell phenotype after extreme β-cell loss (9). Similarly, the α-cell
line we used spontaneously reexpressed small but detectable levels
of insulin, despite being a subclone selected for low insulin protein
(10). During assay development and optimization, we could show,
by spiking in β-cells and by antibody competition, that our assay
was sensitive enough to reliably detect insulin levels in as few as
3% of cells, and at 15-fold lower levels than in β-cells (Fig. S1).
We screened 30,710 compounds for induction of insulin pro-
duction using this assay and identified a molecule, BRD7389 (Fig.
1A), that after 3-d treatment induced insulin expression in mouse
α-cells. BRD7389 induced a dose-dependent up-regulation of
Ins2 mRNA, peaking at ≈0.85 μM; 5-d treatment with BRD7389
resulted in greater induction of insulin gene expression, about 50-
fold at 0.85 μM (Fig. 1B), which could not be further increased by
longer treatments up to 21 d. This compound appears to be spe-
cific to α-cells, because a pancreatic ductal cell line (PANC-1)
showed no induction, and a mouse β-cell line (βTC3) no further
increase of insulin expression. In addition to insulin expression,
BRD7389 significantly up-regulated expression of Pdx1 (Fig. 1C),
a master regulatory transcription factor that specifies pancreatic
progenitors and directly activates the insulin promoter (11). We
also observed a dose-dependent increase in the expression of
other β-cell markers, including Pax4, Iapp, and Npy, after a 5-d
treatment with BRD7389 (Fig. S2).
Treatment with BRD7389 caused a stable change in cell shape
from a fibroblast-like morphology, characteristic of α-cells, to
a clustered state resembling β-cells in culture (Fig. 1 D–F, Left).
Finally, we detected low levels of insulin protein in compound-
treated α-cells by immunofluorescence (Fig. 1 D–F, Right). Rela-
tive to background fluorescence in DMSO-treated α-cells, insulin
staining is induced 1.5-fold following 5-d treatment with BRD7389,
compared with 4-fold higher levels in β-cells. Both insulin mRNA
and protein levels are significantly increased from a basal α-cell
state in compound-treated cells, but do no reach levels detected in
mature β-cells. Therefore, although these cells have not achieved
a β-cell state, they have adopted several features of β-cells.
To identify the mechanism of action of BRD7389, we used
screening data in ChemBank (12) to compare assay performance
Author contributions: D.F.-Y., S.K., J.H.-S., B.K.W., and S.L.S. designed research; D.F.-Y.,
S.K., and T.S. performed research; D.W. and V.D. contributed new reagents/analytic tools;
D.F.-Y., S.K., J.A.B., and P.A.C. analyzed data; and D.F.-Y., S.K., A.S., B.K.W., and S.L.S wrote
the paper.
The authors declare no conflict of interest.
Freely available online through the PNAS open access option.
1
D.F.-Y. and S.K. contributed equally to this work.
2
Present address: Hagedorn Research Institute, DK-2820 Gentofte, Denmark.
3
To whom correspondence should be addressed. E-mail: stuart_schreiber@harvard.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1010018107/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1010018107 PNAS | August 24, 2010 | vol. 107 | no. 34 | 15099–15104
CELLBIOLOGY

of BRD7389 with 9,995 other small molecules in a total of 32 assays
involving both BRD7389 and other compounds. This computa-
tional method looks for similarity of biological assay-performance
profiles among a diverse set of compounds, including many known
“bioactives.” We uncovered multiple connections of BRD7389 to
known kinase inhibitors. Accordingly, we profiled this compound
at 10 μM against a panel of 219 kinases, selected to represent
a diverse subset of the human kinome (13). We observed signifi-
cant inhibition of a number of kinases, including FLT3, DRAK2,
and the RSK family (Fig. 2A and Table S1). To validate these
profiling results, we obtained dose–response curves and deter-
mined half-maximal inhibitory concentration (IC50) values for
BRD7389 and the most potently inhibited kinases (Fig. S3). The
compound was most active against the entire RSK family of ki-
nases, with IC50 values of 1.5 μM, 2.4 μM, and 1.2 μM for RSK1,
RSK2, and RSK3, respectively (Fig. 2B). Therefore, we focused on
investigating the role of RSK kinases in α-cells.
In addition to measuring the biochemical in vitro inhibition
of RSKs, we also determined the functional consequences of
BRD7389 on Rsk activity in mouse α-cells. All Rsk kinases consist
of two functional domains, which are activated through a series
of consecutive phosphorylation events (14). Kinase activity was
measured using pan- and phospho-specific antibodies to detect
total and active Rsk protein in αTC1 cells. Western blot analy-
sis revealed a 50% decrease in kinase activity, as measured by
autophosphorylation of both N-terminal and C-terminal domains,
at concentrations above 3.4 μM (Fig. 2 C and E). Phosphorylation
of ribosomal protein S6 at serines 235 and 236, direct targets of
the Rsk kinases (15), was reduced by a similar amount after
compound treatment (Fig. 2 D and F). These findings confirm that
BRD7389 has activity as an Rsk family kinase inhibitor in vitro
and in cell culture.
We then sought to determine whether knockdown of Rsk
family members would have an effect on insulin production in
α-cells. We observed 2- to 4-fold increases in insulin expres-
sion upon RNAi of individual Rsk proteins, especially Rsk2 and
Rsk3, but the effect is not as strong as compound treatment with
BRD7389 (Fig. 2G). The knockdown efficiency was at least 50%
for all constructs (Fig. 2H), and better knockdown did not cor-
relate with stronger induction of insulin expression. Similar to
compound treatment, which causes maximum induction of insulin
expression at concentrations around the biochemical IC50 for
Rsks, only partial knockdown of the enzymes seems optimal for
insulin induction.
Though mouse α-cells are useful for screening, species differ-
ences and potential microenvironmental factors make testing
compounds in human pancreatic cells essential. Using human
donor-derived pancreatic islets, we tested BRD7389 in dissociated
islet cells cultured on an extracellular matrix (16) designed to
preserve the functional characteristics of β-cells. Though we did
observe donor-to-donor variability in the response to BRD7389,
some observations were shared among islets from donors with
a low body-mass index (BMI) (Fig. 3 and Figs. S4–S8). For ex-
ample, 5-d treatment with BRD7389 enhanced glucose-stimulated
insulin secretion (GSIS) in both high- (16.7 mM) and low-glucose
(1.67 mM) conditions (Fig. 3A), as well as glucose-stimulated
glucagon secretion (GSGS) in low-glucose (1.67 mM) conditions
(Fig. 3B). Moreover, we detected a dose-dependent increase in the
expression of endocrine hormones and transcription factors fol-
lowing 5-d compound treatment (Fig. 3C). Microscopy revealed
Fig. 1. BRD7389 stimulates insulin production in a mouse α-cell line. (A) Structure of BRD7389. Quantitative PCR analysis of (B) insulin (Ins2) and (C) Pdx1
expression following 3- and 5-d treatment with the indicated concentrations of BRD7389. Gene expression was normalized by actin (Actb) expression, and
scored relative to the DMSO-treated controls. Data represent the mean ± SDs of three independent experiments. Bright field microscopy and immunoflu-
orescence for insulin protein in the Cy5 channel was performed on (D) DMSO-treated αTC1 α-cells (average Cy5 intensity per cell: 212 ± 43), (E) αTC1 cells
treated for 5 d with 3.4 μM BRD7389 (average Cy5 intensity per cell: 320 ± 67), and (F) DMSO-treated βTC3 β-cells (average Cy5 intensity per cell: 865 ± 177).
(Scale bar: 50 μm.)
15100 | www.pnas.org/cgi/doi/10.1073/pnas.1010018107 Fomina-Yadlin et al.

Fig. 2. BRD7389 inhibits the RSK family of kinases. (A) Inhibition of the AGC family of kinases (figure modified from ref. 13). Biochemical inhibition of
selected kinases by 10 μM BRD7389 was tested at an ATP concentration within 15 μM of the apparent KM, resulting in 75% activity remaining (gray), 51–75%
remaining (yellow), 25–50% remaining (orange), or 25% remaining (maroon). (B) Dose-dependent inhibition of RSK kinases by BRD7389. Each protein was
incubated with the indicated concentration of BRD7389, and kinase activity was determined in a radiometric filter-binding assay. Activity was scored relative
to DMSO. (C) Western blot analysis for levels and activity of intracellular RSKs upon 5-d compound treatment. Each blot was simultaneously probed with the
indicated primary antibody (Rsk1/2/3, Rsk pT359/pS363, or Rsk pT573) and β-actin antibody, followed by incubation with IRDye-labeled secondary antibodies.
Blots were scanned on an infrared imaging system. (D) Western blot analysis for levels and phosphorylation of intracellular ribosomal protein S6 upon
compound treatment. Each blot was simultaneously probed with the indicated primary antibody (S6rp, S6rp pS235/pS236, S6rp pS240/pS244) and β-actin
antibody, followed by incubation with IRDye-labeled secondary antibodies. Blots were scanned on an infrared imaging system. (E) Fraction of active Rsk by
quantification of Western blots in C. Each specific band was quantified using Odyssey software, normalized to the β-actin signal, and phosphorylation was
plotted as the ratio of normalized phospho-specific to normalized pan-specific antibody signal. (F) Fraction of phosphorylated ribosomal protein S6 by
quantification of Western blots in D as described. (G) Insulin (Ins2) mRNA levels following treatment with BRD7389 or knockdown of the indicated Rsk kinases.
For knockdowns, αTC1 cells were infected with lentiviruses carrying expression cassettes that encode short hairpin RNAs directed against the indicated Rsk.
The following day, infected cells were selected with puromycin, and RNA was prepared after 4 additional d. Significant difference from the average of empty
vector controls, **P 0.01. (H) Knockdown efficiency of Rsk hairpins. Remaining mRNA levels following knockdown of individual Rsk enzymes assayed by
qPCR with corresponding primer sets.
Fomina-Yadlin et al. PNAS | August 24, 2010 | vol. 107 | no. 34 | 15101
CELLBIOLOGY

that the total number of cells in culture decreased, with ≈50% of
cells remaining at 3.4 μM BRD7389 (Fig. 3 D and E). Nonetheless,
the β-cell population remained essentially unchanged, decreasing
only slightly at higher compound concentrations, whereas the
α-cell population decreased dramatically at high concentrations
(Fig. 3D). Staining for cleaved caspase 3, an indicator of apoptosis,
revealed an increase in the fraction of total cells undergoing ap-
optosis (Fig. S9). Whereas other cell types start undergoing apo-
ptosis at 1.7 μM BRD7389, β-cells are only marginally affected at
the highest concentration tested. These differences in viability and
the resulting changes in the ratios of cell types are likely too small
to account for the increases in expression of β-cell-specific genes,
suggesting that treatment with BRD7389 either induces β-cell-like
characteristics in non–β-cells, or enhances existing β-cell function
in human pancreatic islet culture.
Discussion
In summary, we have identified a unique small molecule that up-
regulates insulin expression, normally a defining property of
pancreatic β-cells, in terminally differentiated α-cells. A mecha-
nism potentially involving the inhibition of RSK kinases is sup-
ported by the increase in insulin expression following knockdown
of individual RSK kinases. Our findings raise the possibility that
BRD7389 functions by inhibiting multiple RSK family members
simultaneously. Interestingly, previously described RSK inhib-
itors (17) FMK and BI-D1870 did not induce insulin expression in
α-cells. These compounds inhibit not only RSK enzymes, but also
members of several other kinase families (18). These data suggest
that a tight specificity profile for different kinases might be nec-
essary for optimal induction of insulin expression in α-cells.
Therefore, a systematic evaluation of the entire kinome by both
small-molecule and knockdown approaches will better define the
roles of on- and off-target effects and may lead to the identifica-
tion of conditions for complete transdifferentiation to β-cells.
BRD7389 also increases β-cell–specific gene expression in
primary human islet cells. These experiments could in principle be
confounded by differences in donor age, sex, BMI, and the purity
and viability of islet batches. We found that differences in BMI
appear to influence compound effects; there was an increase in
endocrine hormone secretion in islets from lower BMI donors,
whereas islets from high-BMI donors had attenuated responses.
Interestingly, primary human islet cells tolerate higher concen-
trations of BRD7389 than the mouse α-cell line used here. Al-
though we observed pronounced compound effects on endocrine
cell numbers and function, it is not clear whether these effects are
mediated through effects on α-cells or other cell types in this
Fig. 3. BRD7389 affects primary human islets. (A) Glucose-stimulated insulin secretion after 5-d treatment with the indicated concentration of BRD7389.
Data represent mean ± SD of four replicates. Single asterisk indicates significant difference from DMSO-treated control (*P 0.05 and **P 0.01).
(B) Glucose-stimulated glucagon secretion after 5-d treatment with the indicated concentration of BRD7389. Data represent mean ± SD of four replicates.
Significant difference from DMSO-treated control, **P 0.01. (C) Relative gene-expression changes of endocrine specific (colored lines) and control (gray
lines) genes following 5-d treatment with the indicated concentration of BRD7389. Data represent mean ± SD of four replicates. (D) Quantification of relative
cell numbers compared with DMSO-treated controls from immunofluorescence samples. Data represent mean ± SD of four replicates. (E) Representative
images of dissociated human-islet cells treated for 5 d with the indicated concentrations of BRD7389. Immunofluorescence staining was performed with
insulin and glucagon antibodies; DNA was stained with Hoechst 33342. (Scale bar: 50 μm.)

culture system. Future experiments involving in vivo β-cell abla-
tion, lineage tracing in animal models, and purified human α-cells
will help illuminate the effects of BRD7389 in greater detail.
These findings show the feasibility of identifying compounds that
induce insulin expression in α-cells and suggest small-molecule
approaches to increase β-cell mass by transdifferentiation in vivo.
Materials and Methods
Reagents. Compound BRD7389 (kbsa-0113758) was obtained from Aurora
Fine Chemicals Ltd. All other reagents were obtained from Sigma Aldrich
unless otherwise stated. Primers were bought from Eurofins MWG Operon,
except for Rsk2 and Rsk3 primers, which were ordered from Applied Bio-
systems. Antibodies used in this study were insulin (Sigma I8510), glucagon
(Sigma G2654), RSK1/RSK2/RSK3 (32D7; Cell Signaling Technology, CST 9355),
phospho-p90RSK (Thr359/Ser363; CST 9344), phospho-p90RSK (Thr573; CST
9346), S6 ribosomal protein (CST 2217), phospho-S6 ribosomal protein (Ser235/
236; CST 2211), phospho-S6 ribosomal protein (Ser240/244; CST 2215), β-actin
(Sigma A1978), and cleaved-caspase 3 (Abcam, ab13847). Fluorescently la-
beled secondary antibodies were purchased from Jackson ImmunoResearch.
IRDye antibodies for Western blots were purchased from Odyssey.
Cell and Human Islet Culture. Mouse pancreatic cell lines αTC1 and βTC3 were
grown in low-glucose DMEM supplemented with 10% FBS, 50 U/mL peni-
cillin, and 50 μg/mL streptomycin.
Human islets were obtained through the Islet Cell Resource Consortium
(http://icr.coh.org/) and through the National Disease Research Interchange
(http://www.ndriresource.org/). The purity and viability of human islets are
reported to be 70–93% and 70–98%, respectively, and the average age of
cadaveric donors was 40.7 ± 9.0 y (range 32–57 y; n = 6). Specific data on
individual donors is reported in Table S2. Islets were washed with PBS and
incubated in CMRL medium supplemented with 10% FBS, 2 mM glutamine,
100 U/mL penicillin, and 100 μg/mL streptomycin. Islets were gently dissoci-
ated into a cell suspension by incubating in Accutase (37 °C, 10 min), and
seeded in 96-well plates containing extracellular matrix secreted by the
HTB9 human bladder carcinoma cell line [adapted from Beattie et al. (16)].
Compound treatments for both cell lines and primary human islet cultures
were performed as follows: cells were plated and allowed to adhere over-
night, after which compound solutions in DMSO were added to achieve the
indicated concentrations in 0.1% DMSO. For 5-d treatment, media was
changed and new compound added on day 3.
High-Content Screening. A total of 10,000 αTC1 cells per well were plated in
50 μL media in black, optical bottom, tissue-culture-treated 384-well plates
(Corning) and allowed to attach overnight. Compounds (100 nL per well)
were pin-transferred from concentrated DMSO stocks. Three days after the
beginning of compound treatment, cells were fixed with 1% formaldehyde
in PBS for 30 min at room temperature. Following one wash with PBS, cells
were permeabilized by addition of 50 μL PBS-T (PBS supplemented with
0.1% Triton X-100) for 60 min at room temperature and blocked with 2%
BSA/PBS-T for 60 min. Twenty microliters of primary antiinsulin antibody,
diluted 1:4,000 in 2% BSA/PBS-T, was added per well and incubated over-
night at 4 °C. Following two PBS-T washes, 20 μL Cy-2–labeled donkey-
α-guinea pig antibody diluted 1:500 in 2% BSA, 10 μg/mL Hoechst 33342/
PBS-T was added per well and incubated for 1 h at room temperature in the
dark. After two washes with 50 μL PBS-T, plates were stored in PBS in the
dark at 4 °C until analysis.
Images were acquired on an ImageXpress Micro automated microscope
(Molecular Devices) using a 4× objective (binning 2, gain 2), with laser- and
image-based focusing (offset −130 μm, range ±50 μm, step 25 μm). Images
were exposed for 10 ms in the DAPI channel (Hoechst) and 500 ms in the GFP
channel (insulin). Image analysis was performed using the cell-scoring mod-
ule of MetaXpress software (Molecular Devices). All nuclei were detected
with a minimum width of 1 pixel, maximum width of 3 pixels, and an intensity
of 200 gray levels above background. Cytoplasmic regions around these
nuclei were evaluated for Cy2 staining in the green GFP channel (minimum
width of 5 pixels, maximum width of 30 pixels, intensity 200 gray levels
above background, 10 μm minimum stained area). In total, 75,264 wells were
screened, corresponding to 30,710 unique compounds in duplicate plus
control wells. The compounds screened were selected from a number of su-
blibraries in the Broad Institute compound collection. The screening set was
comprised of 1,920 molecules with previously annotated biological activity,
purchased from commercial vendors Biomol International Inc., Calbiochem,
EMD Biosciences, Microsource Discovery Systems Inc., Prestwick Chemical Inc.,
and Sigma-Aldrich; 1,280 purified natural products from Analyticon Discov-
ery; 15,356 commercial drug-like compounds from ChemDiv Inc., Maybridge,
and TimTec LLC; and 12,154 diversity-oriented synthetic (DOS) compounds
generated at the Broad Institute. The commercial drug-like compounds were
prefiltered by the suppliers to avoid undesired reactive functional groups and
meet physical property filters based on Lipinski’s rule of five. The DOS com-
pounds consisted of a series of stereochemically diverse eight- and nine-
membered macrocycles ranging in molecular mass from 307 to 727 Da, with
an average molecular mass of 572 Da.
Compound purity and identity were determined by UPLC-MS (Waters).
Purity was measured by UV absorbance at 210 nm. Identity was determined
on a SQ mass spectrometer by positive electrospray ionization. Mobile phase
A consisted of 0.1% ammonium hydroxide; mobile phase B consisted of 0.1%
ammonium hydroxide in acetonitrile. The gradient ran from 5% to 95%
mobile phase B over 0.8 min at 0.45 mL/min. An Acquity BEH C18, 1.7 μm, 1.0 ×
50-mm column was used with column temperature maintained at 65 °C.
Compounds were dissolved in DMSO at a nominal concentration of 1 mg/mL,
and 0.25 μL of this solution was injected.
Hits were selected based on the intensity of staining in the Cy2 channel and
the number of Cy2 positive cells, and counterscreened in the same assay
without the use of primary antibody and with Cy3-labeled secondary anti-
body to remove inactive autofluorescent compounds.
In all subsequent immunofluorescence experiments, Cy3 and Cy5 sec-
ondary antibodies were used to avoid effects of compound autofluorescence
in the Cy2 channel.
Gene Expression Analysis. Following compound treatment, cells were lysed
and RNA isolated using the RNeasy Mini Kit (Qiagen) according to the
manufacturer’s protocol. RNA was reverse transcribed with random primers
using the High Capacity cDNA Reverse Transcription Kit with RNase inhibitor
(Applied Biosystems).
Quantitative PCR was performed with Power SYBR Green PCR Master Mix
(Applied Biosystems) on an Applied Biosystems 7900HT real-time PCR ma-
chine using primers in Table S3.
Kinase Profiling. Kinase profiling and dose–response curves were performed
at Millipore’s KinaseProfiler according to the manufacturer’s protocols. ATP
concentrations were within 15 μM of the apparent KM for each enzyme.
Western Blot Analysis. Cell extracts were generated by lysing cells in modified
RIPA buffer containing 1% Nonidet P-40, 0.1% Na deoxycholate, 150 mM
NaCl, 1 mM EDTA, and 50 mM Tris (pH 7.5), and supplemented with protease
and phosphatase inhibitors. A total of 20 μg of each sample were run on E-
Page 48 gels (Invitrogen) and transferred to PVDF membranes using Invi-
trogen iBlot technology. Each blot was simultaneously probed with indicated
primary antibody (all at 1:1,000) and 1:10,000 β-actin antibody, following by
incubation with 1:5,000 IRDye-labeled secondary antibody. Blots were scan-
ned on LI-COR Odyssey Infrared Imaging System and analyzed using Odyssey
software. Each specific band was normalized to the β-actin signal, and
phosphorylation was plotted as a ratio of normalized phospho-specific to
normalized pan-antibody signal.
RNAi Experiments. Lentiviruses resulting in the expression of shRNAs against
RSK family members were obtained from the RNAi Consortion (TRC) (19). The
following hairpins were used: Rsk1 shRNA1: NM_009097.1-559s1c1, Rsk1
shRNA2: NM_009097.1-685s1c1, Rsk2 shRNA1: NM_148945.1-269s1c1, Rsk2
shRNA2: NM_148945.1-1345s1c1, Rsk2 shRNA3: NM_148945.1-1833s1c1,
Rsk3 shRNA1: NM_011299.3-384s1c1, Rsk3 shRNA2: NM_011299.3-627s1c1,
Rsk3 shRNA3: NM_011299.3-2351s1c1. Mouse αTC1 cells were plated in 96-
well plates at 15,000 cells per well in 200 μL of DMEM. The next day, poly-
brene was added to each well (8 μg/mL), and cells were spin-infected with
8 μL virus at 2,250 rpm for 30 min at 30 °C. Media was changed the following
day to fresh, low-glucose DMEM containing 1 μg/mL puromycin and cultured
for 4 additional d. Cells were lysed in RLT buffer and mRNA extracted using
Qiagen RNeasy 96 Kit.
Hormone Secretion in Human Islets. Dissociated human isletscultured in 96-well
plates were washed once with 100 μL per well of PBS and incubated for 1 h in
100 μL low-glucose (1.67 mM) KRB buffer (138 mM NaCl, 5.4 mM KCl, 2.6 mM
MgCl2, 2.6 mM CaCl2, 5 mM NaHCO3, 0.1% BSA), and for an additional hour in
either high-glucose (16.7 mM) or low-glucose KRB buffer. Supernatant from
the first hour was used for glucose-stimulated glucagon secretion using ALPCO
Glucagon (human, mouse, rat) ELISA (following manufacturer’s protocol for
50 μL of sample). Supernatant from the second hour was used to measure
glucose-stimulated insulin secretion using ALPCO Insulin ELISA (human).
Fomina-Yadlin et al. PNAS | August 24, 2010 | vol. 107 | no. 34 | 15103
CELLBIOLOGY

ACKNOWLEDGMENTS. We thank Andrew Stern, Michelle Palmer, Lynn
Verplank, and the entire Chemical Biology Platform at the Broad Institute for
helpful suggestions in assay development and with high-content screening;
Thomas Nieland, Serena Silver, and David Root from the Broad RNAi platform
for lentiviral knockdown constructs and advice for optimization of the infection
protocol; Jack Taunton (University of California, San Francisco) for a sample of
the RSK inhibitor FMK and advice on RSK biology; Yuan Yuan (Chemistry and
Chemical Biology Department, Harvard University) for expression primers;
Robert Gould and the entire CB/NT Diabetes Team for helpful discussion and
advice; and Alejandro Wolf Yadlin (Chemistry and Chemical Biology Depart-
ment, Harvard University) for performing Western blot quantiﬁcation. Funding
for this project was provided by the Juvenile Diabetes Research Foundation and
National Institute for General Medical Sciences Grant GM38627 (to S.L.S.); Na-
tional Institutes of Health Grant RL1-HG004671 for computational work toward
target-hypothesis generation (to V.D. and P.A.C.); Ernst Schering Research Foun-
dation and European Union FP7 Marie Curie Program Grant PIOF-GA-2008-
221135 (to S.K.); an MCO training grant from Harvard University (to D.F.); and
Type 1 Diabetes Pathﬁnder Award DP2-DK083048 from the National Institutes
of Health–National Institute of Diabetes and Digestive and Kidney Diseases (to
B.K.W.). S.L.S. is an Investigator at the Howard Hughes Medical Institute.
1. Shapiro AM, et al. (2000) Islet transplantation in seven patients with type 1 diabetes
mellitus using a glucocorticoid-free immunosuppressive regimen. N Engl J Med 343:
230–238.
2. Borowiak M, Melton DA (2009) How to make beta cells? Curr Opin Cell Biol 21:
727–732.
3. Kroon E, et al. (2008) Pancreatic endoderm derived from human embryonic stem cells
generates glucose-responsive insulin-secreting cells in vivo. Nat Biotechnol 26:
443–452.
4. Zhou Q, Melton DA (2008) Extreme makeover: Converting one cell into another. Cell
Stem Cell 3:382–388.
5. Pearl EJ, Horb ME (2008) Promoting ectopic pancreatic fates: Pancreas development
and future diabetes therapies. Clin Genet 74:316–324.
6. Collombat P, et al. (2007) Embryonic endocrine pancreas and mature beta cells
acquire alpha and PP cell phenotypes upon Arx misexpression. J Clin Invest 117:
961–970.
7. Zhou Q, Brown J, Kanarek A, Rajagopal J, Melton DA (2008) In vivo reprogramming of
adult pancreatic exocrine cells to beta-cells. Nature 455:627–632.
8. Collombat P, et al. (2009) The ectopic expression of Pax4 in the mouse pancreas
converts progenitor cells into alpha and subsequently beta cells. Cell 138:449–462.
9. Thorel F, et al. (2010) Conversion of adult pancreatic alpha-cells to beta-cells after
extreme beta-cell loss. Nature 464:1149–1154.
10. Hamaguchi K, Leiter EH (1990) Comparison of cytokine effects on mouse pancreatic
alpha-cell and beta-cell lines. Viability, secretory function, and MHC antigen
expression. Diabetes 39:415–425.
11. Iype T, et al. (2005) Mechanism of insulin gene regulation by the pancreatic transcription
factor Pdx-1: Application of pre-mRNA analysis and chromatin immunoprecipitation to
assess formation of functional transcriptional complexes. J Biol Chem 280:16798–16807.
12. Seiler KP, et al. (2008) ChemBank: A small-molecule screening and cheminformatics
resource database. Nucleic Acids Res 36 (Database issue):D351–D359.
13. Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam S (2002) The protein kinase
complement of the human genome. Science 298:1912–1934.
14. Carriere A, Ray H, Blenis J, Roux PP (2008) The RSK factors of activating the Ras/MAPK
signaling cascade. Front Biosci 13:4258–4275.
15. Anjum R, Blenis J (2008) The RSK family of kinases: Emerging roles in cellular
signalling. Nat Rev Mol Cell Biol 9:747–758.
16. Beattie GM, Cirulli V, Lopez AD, Hayek A (1997) Ex vivo expansion of human
pancreatic endocrine cells. J Clin Endocrinol Metab 82:1852–1856.
17. Nguyen TL (2008) Targeting RSK: An overview of small molecule inhibitors.
Anticancer Agents Med Chem 8:710–716.
18. Bain J, et al. (2007) The selectivity of protein kinase inhibitors: A further update.
Biochem J 408:297–315.
19. Moffat J, et al. (2006) A lentiviral RNAi library for human and mouse genes applied to
an arrayed viral high-content screen. Cell 124:1283–1298.

Supporting Information
Fomina-Yadlin et al. 10.1073/pnas.1010018107
Fig. S1. High-content screen for small-molecule inducers of insulin production in mouse α-cells. (A) Assay setup. (B) Detection of spiked-in β-cells (cell line βTC3)
in a population of α-cells (cell line αTC1) by automated image analysis following immunoﬂuorescence detection of insulin. (C) Competition experiment sim-
ulating lower insulin levels compared with β-cells.In a population of α-cells, 5% spiked-in β-cells were stained with insulin primary antibody and Cy2-labeled
α-guinea pig secondary antibody and increasing amounts of competing Cy3-labeled α -guinea pig antibody. Staining intensity of insulin positive and negative
cells was measured in the green (Cy2) channel. (D) Representative images from spiked-in and antibody competition experiments. (Scale bar: 50 μm.)
Fomina-Yadlin et al. www.pnas.org/cgi/content/short/1010018107 1 of 6

Fig. S2. Induction of β-cell markers in α-cells and comparison with β-cell levels. Five-day treatment with BRD7389 causes dose-dependent up-regulation of Ins2
(A), Pdx1 (B), Pax4 (C), Npy (D), and Iapp (E) mRNA in mouse α-cells. Levels of expression in mature mouse β-cells are plotted for comparison. Data represent the
mean ± SDs of three independent experiments.

Fig. S3. Dose-responsive kinase inhibition by BRD7389. Human kinases CDK5/p35 (A), DRAK1 (B), FLT3 (C), PIM1 (D), PKG1α (E), RSK1 (F), RSK2 (G), RSK3 (H),
and SGK (I) were incubated with the indicated concentrations of BRD7389, and kinase activity was determined in a radiometric filter-binding assay. Activity was
scored relative to DMSO. Two replicate experiments were used to fit curves and calculate IC50 values using logistic regression implemented in Spotfire.
Fig. S4. Effects of BRD7389 treatment in primary human islets from donor 1. (A) Glucose-stimulated insulin secretion after 5-d treatment with BRD7389.
(B) Relative gene-expression changes of endocrine-specific (colored) and housekeeping (gray) genes following 5-d treatment with BRD7389. (C) Quantification
of relative cell numbers compared with DMSO-treated controls from immunofluorescence samples. (D) Representative images of compound-treated dissoci-
ated islets. (Scale bar: 50 μm.)

Fig. S5. Effects of BRD7389 treatment in primary human islets from donor 3. (A) Glucose-stimulated insulin secretion after 5-d treatment with BRD7389. (B)
Quantification of relative cell numbers compared with DMSO-treated controls from immunofluorescence samples. (C) Representative images of compound-
treated dissociated islets. (Scale bar: 50 μm.)
(B) Glucose-stimulated glucagon secretion after 5-d treatment with BRD7389. (C) Quantification of relative cell numbers compared with DMSO-treated
controls from immunofluorescence samples. (D) Representative images of compound-treated dissociated islets. (Scale bar: 50 μm.)

(B) Relative gene-expression changes of endocrine-specific (colored) and housekeeping (gray) genes following 5-d treatment with BRD7389. (C) Quantification
of relative cell numbers compared with DMSO-treated controls from immunofluorescence samples. (D) Representative images of compound-treated dissoci-
ated islets. (Scale bar: 50 μm.)
(B) Glucose-stimulated glucagon secretion after 5-d treatment with BRD7389. (C) Relative gene-expression changes of endocrine specific (colored) and
housekeeping (gray) genes following 5-d treatment with BRD7389. (D) Quantification of relative cell numbers compared with DMSO-treated controls from
immunofluorescence samples. (E) Representative images of compound-treated dissociated islets. (Scale bar: 50 μm.)

Fig. S9. Activation of caspase 3 in dissociated human islets after 5-d treatment with BRD7389. (A) Quantification of insulin-positive cells and cleaved
caspase 3-positive cells. (B) Quantification of costaining for insulin and cleaved caspase 3, indicating insulin-positive β-cells undergoing apoptosis and insulin-
negative non–β-cells undergoing apoptosis. Significant difference from DMSO-treated control, *P 0.05 and **P 0.01. (C) Representative images of
cleaved caspase 3 staining in compound-treated dissociated islets. (Scale bar: 50 μm.)
Table S1. Inhibition profiling across a panel of 253 kinases with compounds BRD7389 and
BRD6447
Table S1 (DOC)
Values indicate the percent remaining activity for the indicated human (h), mouse (m), rat (r), and yeast (y)
kinases. Assays were performed using 10 μM BRD7389 and BRD6447, and at an ATP concentration within 15 μM of
the apparent Km of each kinase.
Table S2. Donor information for human islet samples analyzed in Fig. 3 and Figs. S3–S7
Table S2 (DOC)
Table S3. Primers used for quantitative RT-PCR gene expression measurements
Table S3 (DOC)
Table lists mouse (mm) and human (hs) primers used, their target transcript, orientation, and sequence.

BRD7389 paper

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to BRD7389 paper

Similar to BRD7389 paper (20)

BRD7389 paper