Alternative splicing analysis in human monocytes and macrophages reveals MBNL1 as major regulator

Abstract

We report the detailed transcriptomic profiles of human innate myeloid cells using RNA sequencing. Monocytes migrate from blood into infected or wounded tissue to differentiate into macrophages, and control inflammation via phagocytosis or cytokine secretion. We differentiated culture primary monocytes with either GM- or M-CSF to obtain pro- or anti-inflammatory macrophages, and respectively activated them with either LPS/IFNγ or anti-inflammatory cytokines. We also treated the THP-1 monocytic cell line with PMA and similar cytokines to mimic differentiation and activation. We detected thousands of expression and alternative-splicing changes during monocyte-to-macrophage differentiation and activation, and a net increase in exon inclusion. MBNL1 knockdown phenocopies several alternative-splicing changes and strongly impairs PMA differentiation, suggesting functional defects in monocytes from Myotonic Dystrophy patients. This study provides general insights into alternative splicing in the monocyte-macrophage lineage, whose future characterization will elucidate their contribution to immune functions, which are altered in immunodeficiencies, autoimmunity, atherosclerosis and cancer.

Full text

Published online 16 May 2018 Nucleic Acids Research, 2018, Vol. 46, No. 12 6069–6086
doi: 10.1093/nar/gky401
Alternative splicing analysis in human monocytes and
macrophages reveals MBNL1 as major regulator
Hongfei Liu1, Paolo A. Lorenzini1,2, Fan Zhang3, Shaohai Xu1, Mei Su M. Wong1, Jie Zheng3
and Xavier Roca1,*
1School of Biological Sciences, Nanyang Technological University, 637551 Singapore, 2Nanyang Institute of
Technology in Health and Medicine, Interdisciplinary Graduate School (IGS), Nanyang Technological University,
637551 Singapore and 3School of Computer Science and Engineering, Nanyang Technological University, 637551
Singapore
Received December 19, 2017; Revised March 13, 2018; Editorial Decision March 30, 2018; Accepted May 01, 2018
ABSTRACT
We report the detailed transcriptomic profiles of
human innate myeloid cells using RNA sequenc-
ing. Monocytes migrate from blood into infected or
wounded tissue to differentiate into macrophages,
and control inflammation via phagocytosis or cy-
tokine secretion. We differentiated culture primary
monocytes with either GM- or M-CSF to obtain
pro- or anti-inflammatory macrophages, and respec-
tively activated them with either LPS/IFN␥or anti-
inflammatory cytokines. We also treated the THP-1
monocytic cell line with PMA and similar cytokines
to mimic differentiation and activation. We detected
thousands of expression and alternative-splicing
changes during monocyte-to-macrophage differen-
tiation and activation, and a net increase in exon
inclusion. MBNL1 knockdown phenocopies several
alternative-splicing changes and strongly impairs
PMA differentiation, suggesting functional defects in
monocytes from Myotonic Dystrophy patients. This
study provides general insights into alternative splic-
ing in the monocyte–macrophage lineage, whose fu-
ture characterization will elucidate their contribution
to immune functions, which are altered in immunod-
eficiencies, autoimmunity, atherosclerosis and can-
cer.
INTRODUCTION
The human immune system exhibits a remarkable diversity
and versatility in ontogeny, morphology and function, thus
offering great opportunities for regulatory network analy-
ses. Monocytes are myeloid cells of the mononuclear phago-
cytic system which are critical for cellular innate immunity,
as they contribute to homeostasis by clearing cellular debris
and toxic particles, and also ght pathogens (1,2). Upon
infection or wound, blood monocytes migrate into tissues
and differentiate into macrophages, which directly combat
infection through a pro-inammatory response, or subse-
quently control it by anti-inammatory (wound healing) re-
sponse (3).
Cultured primary monocytes are differentiated by
Granulocyte-Macrophage Colony Stimulating Factor or
GM-CSF (CSF-2) or by Macrophage Colony Stimulating
Factor or M-CSF (CSF-1), respectively adopting mostly
pro- or anti-inammatory properties, and these cells are of-
ten referred to as M1 or M2 macrophages (Figure 1A) (3).
The THP-1 cell line, which recapitulates some monocyte
functions, can be differentiated into macrophage-like cells
by phorbol myristic acid (PMA) (4,5). Most tissue-resident
macrophages are deposited at early developmental stages,
are maintained by self-renewal, and may be generated
without a monocytic stage like brain microglia (2,6,7),
yet monocyte-derived macrophages are important for
defence against infection. Recent efforts are cataloguing
the heterogeneity of macrophages in both humans and
rodents (8).
Monocyte or macrophage receptor binding to microor-
ganisms elicits phagocytosis or cytokine secretion via
deep changes in gene expression (6). The best known
stimulus pertains to the Gram-negative bacteria-derived
lipopolysaccharide (LPS) binding to Toll-Like Receptor
4 (TLR4), which results in inammatory response that
is enhanced by gamma-interferon (IFN␥)(9,10). Anti-
inammatory macrophages are polarized by cytokines such
as IL4 for tissue repair, allergic and anti-parasitic responses,
IL1␤for humoral immunity and T-helper 2 response, or
IL10 for anti-inammatory response and tissue repair (Fig-
ure 1a) (11,12). These polarization protocols exemplify the
great plasticity of macrophages in culture which is even
higher in the body.
Monocytes and macrophages are deeply associated with
many types of human disease at multiple levels. Their de-
fective responses result in enhanced infections, while an ex-
*To whom correspondence should be addressed. Tel: +65 65927561; Fax: +65 67913856; Email: xroca@ntu.edu.sg
C
The Author(s) 2018. Published by Oxford University Press on behalf of Nucleic Acids Research.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License
(http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work
is properly cited. For commercial re-use, please contact journals.permissions@oup.com
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

6070 Nucleic Acids Research, 2018, Vol. 46, No. 12
Figure 1. Human primary monocyte differentiation protocol and differentially expressed genes (DEGs). (A) Diagram of differentiation and polarization
of human primary monocytes (Mo) to macrophages (M␾) in culture. We performed RNA-seq for the seven monocyte and macrophage conditions. (B)
Flow cytometry measurements of four representative surface markers on monocytes and macrophages (more in Supplementary Figure S2b). Right, median
uoresce intensity (MFI) of four donors. As expected, CD14 and the haemoglobin scavenger receptor CD163 in monocytes were downregulated only in
GM-CSF macrophages. The T-cell co-stimulator CD80 and the Immunoglobulin G Fc receptor CD16 were upregulated only in GM-CSF or M-CSF
macrophages, respectively, while CD80 is further increased by LPS/IFN␥.(C) Venn diagrams with numbers of differentially expressed genes (DEGs)
with fold change ≥2(q-value ≤0.05) for GM-CSF and M-CSF treated macrophages versus monocytes (left), and for the four polarizations (right). The
enrichment of common M/GM-CSF DEGs is highly signicant (P∼0, one-tailed Fisher’s Exact test). (D) Bar charts with the top six enriched clusters
of DEGs upon M/GM-CSF stimulations, using DAVID bioinformatics resources. List hits, number of genes in our list that belong to each gene category;
Population hits (pop hits), total number of genes that belong to each gene category in the whole DAVID database. Barcharts display gene category and
percentage of our list hits in the population hits in the y- and x-axis, respectively. Each bar includes our list hits number and the benjamini q-value of the
cluster.
cessive or uncontrolled function causes allergy or autoim-
munity, such as in acute inammation or sepsis (3), in
obese adipose tissue leading to insulin resistance and di-
abetes (3,13), and in atherosclerosis (14). Tumour associ-
ated macrophages play a dual role in cancer, as they ght
solid tumors early on, and later they promote metastasis
and confer poor prognosis (3,15). Finally, Myelodysplas-
tic Syndromes (MDS), which are diverse pre-malignancies
with reduced myeloid differentiation, cytopenia and suscep-
tibility for acute myeloid leukemia, are caused by frequent
and recurrent Splicing Factor (SF) mutations whose prog-
nostic value and pathogenesis mechanisms are currently un-
der evaluation (16,17). The frequent SF mutations in MDS
illustrate that splicing plays a crucial role in the develop-
ment and function in the myeloid lineage.
Immune-cell differentiation and activation involve
widespread transcriptomic changes, reecting regulation
via transcription and mRNA decay (18), yet alternative
splicing (AS) is becoming important (19,20). The com-
plexity of the human transcriptome and proteome, across
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

Nucleic Acids Research, 2018, Vol. 46, No. 12 6071
and within tissues, is largely expanded by AS virtually
regulating all intron-containing transcripts (21,22). The
AS types, such as cassette exon inclusion, intron retention
or alternative splice sites, result in mRNAs which can
encode protein isoforms with altered, antagonistic or
unrelated functions. AS patterns are dictated by cis-acting
elements, comprising exonic or intronic splicing enhancers
or silencers, which activate or repress splicing via binding to
trans-acting SFs, and each SF regulates up to thousands AS
events (23). AS also depends on RNA secondary structure
(24), transcription and chromatin (25). RNA-sequencing
(RNA-seq) studies showed a prevalence of tissue-specic
AS (26–28).
AS has been largely studied in lymphocytes but less so in
myeloid cells (29). AS and transcription affect distinct gene
subsets upon T-cell activation (19), and AS changes from
cell lines and primary T cells differ in the response path-
ways and stimuli dependence. Recent transcriptomic studies
characterized AS in Hematopoietic Stem Cells (30)aswell
as in terminal rodent and human erythropoiesis (31,32),
terminal murine granulopoiesis (33) and megacaryopoiesis
(31), revealing increased intron retention as a feature of
late-stage red blood cells and mature granulocytes. Even
though the RNA proles of THP-1 and primary mono-
cytes, resting and LPS-activated macrophages are well char-
acterized and master transcriptional and epigenetic regu-
lators are known, there is insufcient information on AS
(4,5,9,10). LPS-activated mouse macrophages slow down
splicing in certain pre-mRNAs thus favoring AS (34,35).
Furthermore, splicing factor 3a in murine macrophages reg-
ulates AS events involved in TLR signalling (36). For hu-
man monocytes/macrophages, a recent study established
the role of the quaking (QKI) RNA-binding protein in reg-
ulating AS of wound-healing macrophages and promotion
of atherogenic plaques (37), and another AS study of LPS
activation revealed the role of CELF1 (38). Just a few AS
events are functionally studied in monocytes/macrophages,
in genes like CD44 in inammatory disorders such as
rheumatoid arthritis (39), Tissue Factor in thrombogene-
sis, wound healing, angiogenesis and metastasis (40), and
MyD88 in signalling (41). SRSF3 was shown to control IL-
1␤secretion upon E. coli exposure (42), yet many more SFs
await characterization.
Here, we present a comprehensive RNA-seq analysis
of THP-1 and primary cells to identify thousands of AS
changes during differentiation to pro- or anti-inammatory
macrophages, and hundreds during inammatory activa-
tion, and most of these are totally uncharted. We identi-
ed MBNL1 as an important AS regulator during differ-
entiation. This study should help understand the regulation
of these important cells and their connection to multiple
pathologies, including immunodeciencies, autoimmunity
and cancer (3).
MATERIALS AND METHODS
Ethics
Informed consent to all healthy donors was obtained for
all blood samples from which monocytes and macrophages
were derived. Approval for the presented studies with hu-
man samples was provided by Nanyang Technological Uni-
versity’s Institutional Review Board ethics committee (IRB-
2014-10-002), and donor anonymity was ensured.
Human monocyte and macrophage culture
For primary monocyte purication, we isolated Periph-
eral Blood Mononuclear Cells (PBMCs) from four healthy
donors by Histopaque-1077 (Sigma-Aldrich) gradient
(Supplementary Figure S1A). We subsequently incubated
the samples with negative selection microbeads (Pan Mono-
cyte Isolation Kit, cat.no130-096-537, Miltenyi Biotech) to
obtain puried monocytes.
We added 60 mL blood from each donor to phosphate
buffered saline (PBS) with a ratio of 1:3. We gently poured
30 mL of mixture into ltered Leucosep tubes (cat. no.
227290, Greiner Bio One), followed by centrifugation at
800 ×g, 15 min at room temperature and zero deceler-
ation. Then, we carefully harvested the PBMC layer and
washed it twice with MACS buffer (PBS, pH 7.2 sup-
plemented with 0.5% BSA and 2 mM EDTA, degassed
before use and stored at 4◦C). We resuspended 1 ×107
PBMC cells in 30 ␮l MACS buffer, added by FcR block-
ing reagent and incubated mixtures with Biotin-Antibody
Cocktail and Anti-Biotin MicroBeads. We passed mixtures
through LS columns standing on magnetic separator, and
we collected the ow-through with the unlabeled mono-
cytes. After washing with MACS buffer once, we collected
the monocyte sample directly for RNA isolation.
We differentiated monocytes to macrophages by cultur-
ing puried monocytes with the different cytokines for 7
days in RPMI 1640 medium (cat. no. 30-2001, ATCC) sup-
plemented with 20% Fetal Bovine Serum (FBS; Gibco), 100
U/ml Penicillin and 100 ␮g/ml Streptomycin (Gibco) at
37◦Cand5%CO
2. We used 100 ng/ml GM-CSF to in-
duce inammatory macrophages, and 100 ng/ml M-CSF
to induce anti-inammatory macrophages. Every 3 days we
changed the culture media and added fresh cytokines. On
day 7, we polarized GM-CSF macrophages by 100 ng/ml
LPS plus 50 ng/ml IFN-␥for 8 h. In turn, we separately
incubated M-CSF macrophages for 8 h with 10 ng/ml of
either IL-4, IL-1␤or IL-10 for the three polarizations.
THP-1 culture and nucleofection
We cultured the human monocytic cell line THP-1 (cat. no.
TIB-202, ATCC) in RPMI 1640 Medium (cat. no. 30-
2001, ATCC) supplemented with 10% FBS, 0.05 mM
2-mercaptoethanol, 100 U/ml Penicillin and 100 ␮g/ml
Streptomycin at 37◦Cand5%CO
2. We separately treated
THP-1 monocyte-like cells for 3 days with 100 ng/ml phor-
bol ester 12-O-tetradecanoylphorbol-13-acetate (PMA,
Sigma-Aldrich) or 0.5 ␮M 1,25-dihydroxyvitamin D3 (vita-
min D3, Sigma-Aldrich) to obtain PMA3D and VD3 cells.
We enhanced the differentiation of PMA3D cells by remov-
ing the PMA-containing medium at day 3, and further in-
cubating cells in fresh medium for another 5 days to derive
the PMA5Dr samples, as described (43). We stimulated the
PMA3D cells with 100 ng/ml LPS plus 20 ng/ml IFN-␥for
8 h, and with 20 ng/ml IL-4 plus 20 ng/ml IL-13 for 8 h.
For microscopy, we plated primary monocytes or THP-1
cells in tissue-culture treated ␮-Dish (81156, Ibidi GmbH,
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

6072 Nucleic Acids Research, 2018, Vol. 46, No. 12
Germany) at a density of 1 ×105/cm2, followed by the cor-
responding treatments. We took images with the Zeiss Live
Cell Observer.
The nucleofection of untreated and PMA3D treated
THP-1 cells with dsiRNA (Integrated DNA Technologies,
IDT, USA) generally followed the protocol described be-
fore (44), with dsiRNA target sequences in Supplemen-
tary Table S1. For PMA3D cells, we treated THP-1 cells
with 100 ng/ml PMA for 16 h and detached cells by Ac-
cutase (A6964, Sigma-Aldrich). We mixed 1 ×106cells
with 200 pmol dsiRNA in SG nucleofector solution (Lonza,
Switzerland) and electroporated cells subsequently using
the FI-100 program on Amaxa 4D-Nucleofector (Lonza,
Switzerland). After that, we recovered the cells in THP-
1 nucleofection medium (Lonza Mouse T cell nucleo-
fection medium supplemented with 20% human serum,
1% nonessential amino acids, 1% sodium pyruvate and
1% penicillin/streptomycin/L-glutamine) for 4 h. Then
we changed the medium back to normal differentiation
medium with 100 ng/ml PMA, and we cultured the cells
for 2 days. For untreated THP-1 nucleofection, we collected
cells directly for electroporation, and after recovery we cul-
tured them for another 3 days before RNA or protein iso-
lation.
Lentiviral transduction
We inserted the RBFOX2 and MBNL1 shRNAs (Supple-
mentary Table S1) into the pLKO.1 vector for gene knock-
down. We packaged the recombinant lentiviruses using
a second generation system (Addgene, Lentiviral Guide:
https://www.addgene.org/viral-vectors/lentivirus/lenti-
guide/#second-generation). We used Lipofectamine 2000
Reagent (Life Technologies, USA) to transfect HEK293T
cells with three plasmids encoding pPol/Gag, pVSVG,
and pLKO.1 harboring the corresponding shRNA. We
changed the medium, and harvested the packaged lentiviral
particles 72 h later. We introduced the lentiviral particles
into THP-1 cells by centrifugation at 800 ×g, 32◦Cfor30
min in the presence of 10 ␮g/ml Hexadimethrine bromide
(polybrene, cat no H9268, Sigma). Then we cultured
cells for 3 days and selected transfectants with 2 ␮g/ml
puromycin (Sigma-Aldrich).
Western blotting
We washed cells twice with PBS and incubated with the pro-
tein lysis buffer [50 mM Tris–HCl, pH8; 150 mM NaCl; 1%
Triton X-100; 10% glycerol; 1 mM EDTA; 1×cOmplete
EDTA-free protease inhibitor cocktail (Roche)] for 10 min
before centrifugation. We determined the protein concen-
tration by Bradford assay (Bio-Rad). We ran 30 ␮gofpro-
tein by SDS-PAGE and transferred them to PVDF mem-
brane (Bio-Rad, USA). We detected the specic proteins by
the following primary antibodies: anti-MBNL1 (ab108519,
Abcam), anti-RBFOX2 (A300-864A, Bethyl), anti-SNRPE
(sc-21057, Santa Cruz), anti-RAVER2(ab174321, Abcam),
anti-GEMIN2 (sc-32806, Santa Cruz), anti-hnRNP A1
(sc-32301), anti-KSRP (A302-021A, Bethyl), anti-GAPDH
(sc-25778, Santa Cruz), anti-␣Tubulin (sc-8035, Santa
Cruz) and anti-␤-Actin (A5441, Sigma-Aldrich). We quan-
tied the protein samples from three biological replicates
using the Quantity One software (Bio-Rad, Life Science Re-
search). We derived the relative expression of each protein
from its ratio over reference protein level.
RNA pull-down
We generated double-stranded DNA templates by anneal-
ing and ligation of oligodeoxyribonucleotides (IDT) with
complementary overhangs and a T7 promoter. We in vitro
transcribed the RNA probes from these DNA templates
with T7 mMESSAGE mMACHINE Kit (cat. no. AM1344,
Life Technologies) at 37◦C for 4 h, followed by DNase
treatment for 15 min and RNA purication and concen-
tration (Zymo Research). Next we end-labeled the RNAs
with desthiobiotin, by attaching a single biotinylated nu-
cleotide to the 3terminus of the RNA probe using T4
RNA ligase in its reaction buffer (50 mM Tris•HCl, 10 M
MgCl2, 10 M DTT, 1 mM ATP; pH 7.8) at 16◦C overnight,
followed by RNA purication and ethanol precipitation
(Pierce RNA 3End Desthiobiotinylation Kit, Thermo Sci-
entic). We performed the RNA pull-down assay accord-
ing to the Pierce Magnetic RNA-Protein Pull-Down Kit
(Thermo Scientic). We captured the biotin-labelled RNA
by streptavidin magnetic beads incubated with the RNA in
capture buffer (20 mM Tris of pH 7.5, 1 M NaCl, 1 mM
EDTA) for 30 min at room temperature. After two 20 mM
Tris buffer (pH 7.5) washes, we added nuclear extract to
the RNA-Protein binding buffer (20 mM Tris of pH 7.5, 50
mM NaCl, 2 mM MgCl2, 0.1% Tween-20) and incubated
them with the RNA at 4◦C for 2 h. We washed the beads
three times with the wash buffer (20 mM Tris of pH 7.5,
10 mM NaCl, 0.1% Tween-20) before incubating them with
the default biotin elution buffer at 37◦C for 30 min. We col-
lected the eluates and analysed them with the ow-through
by Western Blotting.
RNA extraction and RT-PCR
We washed monocytes/macrophages with cold PBS three
times, followed by addition of 600 ␮l lysis buffer with 2-
mercaptoethanol directly to 1 ×106cells and then mixing
until cells lysed. We extracted total RNA using PureLink
RNA Mini Kit (Life Technologies, USA). We extracted to-
tal RNA strictly following the kit’s manual as stated above
and processed for RT-qPCR freshly, or immediately stored
in –20◦C for short period (less than 2 weeks) and –80◦C
for longer time (less than 6 months). We removed resid-
ual DNA by TURBO DNase (Ambion) treatment for 30
min, followed by heat-inactivation and RNA purication.
We measured RNA concentration by spectrophotometry
(NanoDrop, Thermo Scientic) and checked RNA integrity
by agarose gel. We used only high quality RNA for down-
stream experiments. We processed cDNA directly for real-
time PCR or stored it immediately at –20◦C for less than
2 weeks, or at –80◦C for up to 6 months. We reverse tran-
scribed 1 ␮g RNA to cDNA with M-MuLV reverse tran-
scriptase (New England Biolabs), dNTPs and oligo-dT in
10 ␮l of volume at 42◦C for 1 h, followed by 15 min of
inactivation at 65◦C. We performed quantitative RT–PCR
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

Nucleic Acids Research, 2018, Vol. 46, No. 12 6073
using the SYBR Green Master Mix (Thermo Fisher Sci-
entic) on CFX96 Touch Real-Time PCR Detection Sys-
tem (Bio-Rad). We indicate primer sequences in Supple-
mentary Table S2, with all amplicons shorter than 200 bp.
We used hACTB (␤-Actin) as qPCR reference gene. We cal-
culated the relative expression of three biological replicates
of each target gene by the equation 2(Cq(␤-Actin)-Cq(target gene)).
We plotted means and standard deviation.
We performed semi-quantitative RT-PCR to validate or
detect alternative splicing (AS) events. We designed primer
pairs to map in the anking exons of each target AS event,
with sequences in Supplementary Table S3. We radioac-
tively labeled each forward primer with ␥-32P-ATP (Perkin-
Elmer, USA) using T4 PolyNucleotide Kinase (New Eng-
land Biolabs, USA), followed by purication with Mi-
crospin G-25 columns (GE Healthcare). We carried out the
PCRs using Go-Taq DNA polymerase (Promega, USA) for
25–33 cycles. We separated the PCR products with 8% na-
tive polyacrylamide gel electrophoresis (PAGE) at 200 V for
5 h in TBE buffer, followed by vacuum-drying and exposure
to phosphorimaging screen. We used Typhoon imager (GE
Healthcare) for signal scanning, and we quantied the band
intensities by ImageQuant TL software (GE Healthcare).
RNA-sequencing library preparation
After removing medium, we washed macrophages twice
with cold PBS to get rid of oating cells. We washed THP-
1 cells twice in PBS by centrifugation. We isolated total
RNA using RNeasy Mini Kit together with QIAshredder
(Qiagen). We removed the residual DNA by 30-min DNase
treatment (TURBO DNA-free Kit, Ambion) at 37◦C. Sub-
sequently, we puried RNA with RNA Clean and Con-
centrator (Zymo Research), and checked its concentration
and integrity using ribogreen assay and bioanalyzer (Ag-
ilent Technologies 2100 Bioanalyzer). We only processed
RNA samples with RIN above 8. We prepared RNA li-
braries using the TruSeq Stranded mRNA Library Prep Kit
(Illumina). In brief, we puried poly-A containing mRNA
molecules by poly-T oligos attached to magnetic beads. We
cleaved and fragmented the mRNA into small pieces, fol-
lowed by reverse transcription of the rst cDNA strand
using random primers. We replaced dTTP with dUTP in
the second strand synthesis by using DNA Polymerase I
and RNase H, so as to track the strand information in the
following steps. We ligated the cDNA fragments with the
adapter and puried them. We enriched the molecules by
PCR to generate the nal cDNA library. We used HiSeq
2500 System Rapid mode (Illumina) for RNA-sequencing,
giving rise to paired-end reads of 101 bp. We show total read
numbers of each sample in Supplemental Table S4.
RNA-seq data analysis
We rst conrmed the original read quality of our RNA-seq
data by FastQC (Simon Andrews, Babraham Bioinformat-
ics). For gene expression analysis, we aligned the paired-end
reads to human genome (UCSC-hg19) using Tophat2 (45)
with Bowtie2 (46). In RNA-seq data analysis, we used the
default parameters for each tool unless otherwise stated.
For Tophat2, we allowed a maximum of two mismatches
and set two parameters as –library-type =fr-rststrand and
–read-realign-edit-dist =0. We used Cufinks, Cuffmerge
and Cuffdiff (47) for transcripts assembly and quanti-
cation of the Tophat2 output. We masked out annotated
transfer-messenger RNA, ribosomal RNA and mitochon-
drial transcripts in Cuffdiff analysis by using the parameter
–mask-le with corresponding GTF le. We obtained 87–
92% of aligned reads to the genome with Tophat2 (Supple-
mentary Table S4, except one replica). The Pearson corre-
lation (R2) for Log10(FPKM) (Fragments Per Kilobase of
transcript per Million fragments mapped) for each individ-
ual condition across donor pairs was on the range of 0.90–
0.96, highlighting the strong reproducibility of our data
(Supplementary Figure S3; samples from different donors
are not identical). We used the common FPKM (fragments
per kilobase of transcript per million fragments mapped) as
the expression-levels measure for each gene or transcript.
To detect differential AS events (DASEs), we used three
softwares, rMATS (48), MISO (49) and SpliceTrap (50). To
run rMATS, we adopted the alignment outputs of Tophat2.
Our read type is paired-end (-t paired) and with length of
101bp (-len 101), while the analysis type was the default
unpaired. In SpliceTrap running, we rst mapped the orig-
inal paired-end reads to SpliceTrap database hg19, with
low cutoff. We performed postAnalysis subsequently, with
2 junction reads required per junction (-j 2) and low cut-
off (-c L). Finally, we compared exon inclusion ratio be-
tween each two samples using SpliceChange. In MISO anal-
ysis, rst we used the Tophat2 alignment results to esti-
mate isoform expression levels within each sample, with
the command miso –run. We included the –paired-end op-
tion, and we set the mean and standard deviation of insert
length to 350 and 150, respectively. Then we summarized
the MISO outputs for each sample, with the command sum-
marize miso –summarize-samples, to obtain condence in-
tervals for the probabilistic framework. Next, we made pair-
wise comparisons between samples, using compare miso
–compare-samples, to detect differentially expressed iso-
forms. In our MISO analysis, we covered the following
AS types: skipped exons (SE), alternative 3/5splice sites
(A3SS, A5SS), mutually exclusive exons (MXE) and Re-
tained introns (RI).
For MISO, we took the ‘sample posterior mean’ aver-
age among different replicates as the nal exon inclusion
level for each condition, and used Bayes factor min and me-
dian as a criteria of statistical signicance among the repli-
cates. We obtained the nal alternative spliced exon list of
MISO for each comparison by applying the following cut-
offs: inclusion level difference ≥0.05; Bayes factor Min ≥1
and Bayes factor Median ≥10. For SpliceTrap, we also used
the inclusion ratio average of all replicates as the nal exon
inclusion level, and we calculated the P-value using paired
Student’s t-test. We adopted the following cutoffs to get the
nal AS list: inclusion level difference ≥0.05 and P-value
≤0.001. For the rMATS AS list we just applied the cut-offs
of inclusion level difference ≥0.05 and P-value ≤0.05.
For the co-detected DASEs among the three tools, we re-
trieved the coordinate annotations of each exon from each
tools output, including the information of event type (AA,
AD, CA, IR, MXE), chromosome, strand, start and end po-
sition of AS exons and anking exons on the chromosome.
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

6074 Nucleic Acids Research, 2018, Vol. 46, No. 12
DASEs harboring the same information detected by two or
three tools were named tri- or co-detected DASEs, as we
labeled them in Supplementary Figure S5.
Flow cytometry
We harvested cells and washed them with 1X Phosphate
buffered saline (PBS) before incubation with 2 ␮lof1
mg/ml goat serum (Sigma-Aldrich) for 5 min for Fc re-
ceptors blocking on ice prior to antibody staining. We
stained cells according to manufacturer’s protocol with
uorochrome-coupled antibodies listed in Supplementary
Table S5. We incubated each sample with primary anti-
bodies or the matching isotypes for 15 min on ice. Then
we washed cells twice in buffer and resuspended them in
300 ␮l of PBS. We took ow cytometric measurements us-
ing ve lasers LSR Fortessa X-20 (Becton Dickinson). We
performed data analysis by Cyogic (v.1.2.1).
RESULTS
Major changes in gene expression upon monocyte-to-
macrophage differentiation and activation
We have completed a high-depth RNA-seq proling in hu-
man blood-derived monocyte differentiation and activa-
tion. We used Ficoll gradient and negative selection to iso-
late human monocytes from peripheral blood mononuclear
cells of four donors, yielding >96% purity as judged by
FACS analysis of CD14 (Supplementary Figure S1). Then
we used different cytokines to differentiate these mono-
cytes to macrophages, and activated them with different sig-
nalling molecules (Figure 1A). In brief, we differentiated
monocytes to macrophages with either GM-CSF or M-CSF
for seven days, and cells adhered to plate (Supplementary
Figure S2A). We polarized the GM-CSF macrophages with
LPS/IFN␥, and the M-CSF macrophages with either IL1␤,
IL10 or IL4, each for 8 h. We conrmed the efciency of
these protocols by cell morphology and extensive surface-
marker proling by FACS (Figure 1B and Supplementary
Figure S2B, and antibodies in Supplementary Table S5).
From all four donors, we obtained ∼300 million paired-
end reads per each of the seven conditions (Supplemen-
tary Figure S3, details in Supplementary Methods). Us-
ing a cutoff of ≥2-fold increase or decrease in FPKM,
we identied thousands of differentially expressed genes
(DEGs) in macrophages compared to monocytes (Figure
1C, left Venn diagram; Supplementary Table S6). A to-
tal of 2153 and 1277 DEGs are specic to GM-CSF and
M-CSF macrophages respectively, while the 3497 com-
mon changes are highly enriched. Gene Ontology (GO)
analysis with DAVID for common and M-CSF or GM-
CSF-specic DEGs revealed most enriched clusters in
immunology-relevant categories such as immunity, mem-
brane and chemotaxis (Figure 1D). DEGs specic to ei-
ther M- or GM-CSF macrophages show enrichment in dif-
ferent pathways (Figure 1D, left), highlighting that certain
changes in gene expression drive some of the different prop-
erties of these macrophages, while other DEGs might be a
consequence of the differentiation process.
Compared to the corresponding resting macrophages,
LPS/IFN␥-activation induced many more DEGs than the
three anti-inammatory polarizations (Figure 1C, right di-
agram). GO enrichment analysis showed that LPS/IFN␥
responses clustered in innate immunity and specic path-
ways such as pleckstrin-homology domain, while anti-
inammatory activations were enriched in different immu-
nity pathways like c-type lectin-like and chemotaxis.
In parallel, we treated THP-1 cells with PMA to mimic
macrophage differentiation, for three days (PMA3D) and
for three days followed by ve days rest with fresh medium
(PMA5dr) (43). Also we incubated THP-1 cells with vita-
min D3 (VD3) which activates certain immune pathways
(51). We also induced PMA3D cells with either LPS plus
IFN␥, or with IL4 plus IL13, to mimic the LPS/IFN␥
and IL4 treatments of blood-derived macrophages. These
six conditions showed the expected morphological changes
(Supplementary Figure S2C) and FACS proles for ve
markers (Supplementary Figure S2D). RNA-seq revealed
common DEGs for THP-1 differentiation and polarization
(Supplementary Figure S4A; Supplementary Table S6).
Many RNA binding protein genes change their expression
upon monocyte-to-macrophage differentiation and activation
Several genes encoding RNA Binding Proteins (RBPs)
change expression by ≥2-fold during differentiation and
LPS/IFN␥-activation of primary cells (Figure 2A). These
RBPs included several SFs such as RBM6, SRRM2,
and SRSF5 for M/GM-CSF, and RBM7/11/17, TIA1,
SRSF12, FUBP1, MBNL2 and RBFOX2 for LPS/IFN␥
activation. Among the RBPs as DEGs in THP-1 differ-
entiation and activation (Supplementary Figure S4A and
B), we conrmed by Western Blotting the PMA-induced
downregulation of SNRPE, GEMIN2 and RAVER2, while
we found that hnRNPA1 or KSRP levels did not change
(Supplementary Figure S4C). This experiment shows that
changes in RNA levels lead to differences in protein levels
which should establish the splicing patterns in each condi-
tion.
We found 1824 DEGs in common with M/GM-CSF
in primary cells and PMA/VD3 in THP-1 cells, which is
about one-third of the DEGs in each category (Figure 2B).
A number of RBPs consistently change expression upon
primary-cell and THP-1 differentiation (Figure 2C, left),
which might account for their analogous morphological
and functional features such as cell adherence and expan-
sion, enhanced granularity, etc. GM-CSF and PMA dif-
ferentiated macrophages exhibit upregulation of CPEB1
and CPEB2 translation regulators, RBFOX2, and other
RBPs like IGF2BP3 and MEX3B. In turn, commonly up-
regulated genes by M-CSF and PMA include RNA degra-
dation enzyme RNASE1. Importantly, M/GM-CSF and
PMA macrophages strongly downregulate SFs MBNL1
and PTBP2, degradation enzymes RNASE2 and RNASE3,
scavenger decapping enzyme DCPS, mRNA destabilizing
factors ZFP36 (also known as TTP) and ZFP36L2, and the
putative RBP ZNF74.
Among the DEGs upon LPS/IFN␥activation of both
GM-CSF and PMA3D macrophages (Figure 2C, right),
we detected upregulation of translation regulators CPEB2,
CPEB3 and SAMD4A, splicing factor PRPF3, antivi-
ral protein ZC3HAV1, adenosine-to-inosine editing en-
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

Nucleic Acids Research, 2018, Vol. 46, No. 12 6075
Figure 2. DEGs common in primary and THP-1 cells, and identication of RBPs important for macrophage differentiation or polarization. (A)Heatmap
of mRNA expression (FPKM) for the top 20 RBPs as DEGs in primary cells, either upon M- or GM-CSF differentiation, or induced by LPS/IFN␥.The
four columns under each condition represent the RNA-seq values for four donors. (B) Venn diagrams showing the proportion of DEGs with fold change
≥2(q-value ≤0.05) between the indicated conditions in primary and THP-1 cells. We used genes instead of transcripts for this comparison. The overlap
for analogous stimuli between primary and THP-1 cells is highly signicant (P<10−50 for all three tests, one-tailed Fisher’s Exact test). (C)Heatmapof
mRNA expression (FPKM) levels for the top 15 RBPs as DEGs both in differentiation or activation of primary cells and THP-1, as labelled on top.
zyme ADAR, decapping enzyme DCP1A, exosome subunit
for RNA degradation PNPT1, ARE-destabilizing factor
ZFP36, microRNA biogenesis factor MOV10, and RBPs
with unknown function like RBM43. In turn, we also saw
downregulation of antiviral protein ZC3HAV1L, tRNA
processing factors RPP25 (RNaseP) and TSEN2 (tRNA
splicing endonuclease), and translation repressor MSI2.
Some of these RBPs might be responsible for ne tuning
the anti-inammatory responses triggered by TLR4 stimu-
lation.
Alternative splicing changes upon monocyte-to-macrophage
differentiation and activation
To detect differential AS events (DASEs) by comparing
percentage spliced in (PSI) between conditions, we used
rMATS (48), MISO (49) and SpliceTrap (50) altogether
with a high statistical threshold to minimize the false posi-
tives. We used three tools not to miss many true AS changes
that are only detected by one of them, because just few
DASEs are co-detected by two or three tools for each condi-
tion (Supplementary Figure S5A, top). We found thousands
of DASEs in GM- or M-CSF induced macrophages (Figure
3A, Supplementary le AS blood gene list), with a strong
enrichment of shared events, with 2435 in 1549 genes. In
turn 1933 and 1262 DASEs in 1327 and 903 genes are spe-
cic to GM- and M-CSF, respectively (Figure 3A, left). In-
terestingly, both M/GM-CSF macrophages exhibit higher
levels of long isoforms, or higher inclusion of cassette ex-
ons (Table 1), suggesting that either an increase in splicing
activator/s or a decrease in repressor/s is associated with
macrophages as opposed to monocytes. The most common
type of DASEs in differentiation and activation is cassette
exon followed by mutually exclusiveexons, and by intron re-
tention and alternative 5or 3splice sites with similar levels
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

6076 Nucleic Acids Research, 2018, Vol. 46, No. 12
Figure 3. Alternative splicing in primary monocyte differentiation and polarization. (A) Venn diagrams with the numbers of DASEs for differentiation and
activation as indicated (PSI change ≥0.05; SpliceTrap P-value ≤0.001, MISO BFmin ≥1 & BFmedian ≥10, MATS FDR ≤0.05). In parentheses, number
of genes affected by DASEs. The enrichment of common M/GM-CSF DASEs is highly signicant (P∼0, one-tailed Fisher’s Exact test). (B) Bar charts
with the top enriched clusters of DASEs upon M- and/or GM-CSF stimulation using DAVID, as in Figure 1D. (C) Heat map showing PSI of top ten
RBPs (mostly splicing regulators) as DASEs upon both M/GM-CSF treatments. Only the DASE with the highest PSI change is plotted for each RBP. (D)
Heat maps for the PSI changes of DASEs in different enriched clusters of monocyte differentiation. Only DASEs with the highest PSI change (one per
gene) are shown.
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

Nucleic Acids Research, 2018, Vol. 46, No. 12 6077
(Supplementary Figure S5B, two left diagrams; Supplemen-
tary Table S7).
Certain pathways are commonly enriched and others are
specic to DASEs in M- or GM- CSF macrophages (Figure
3B), which are different from the pathways affected in the
DEGs (Figure 1D), as seen in other cell types (19,27,28).
DASEs in both M/GM-CSF macrophages versus mono-
cytes frequently affect RBPs and SFs in particular, such
as TRA2A/B isoforms, TIA1 or HNRNPLL (Figure 3C).
Other pathways enriched in both or either M- or GM-CSF
macrophage DASEs reveal known functions in ‘immunity’
or ‘coated pit’ (related to endocytosis) (Figure 3D), or clus-
ter in sets of genes whose functions and coordinated AS
changes remain to be characterized in these cells (‘WD’,
‘DENN’ and ‘zinc-nger PHD-type’ protein families). Fi-
nally, ∼800 AS changes upon LPS/IFN␥activation also
demand characterization, compared to the very few de-
tected changes by either IL4, IL1␤or IL10 treatments, and
with only 13 DASEs in common between the LPS/IFN␥
and at least one anti-inammatory activation.
We also measured the DASEs for THP-1 differentiation
by PMA and VD3, and the activation by LPS/IFN␥or
IL4/13 (Supplementary le AS thp1 gene list). We found
hundreds of DASEs for differentiation, with many more
in common between PMA3D and PMA5Dr than between
these two and VD3 (Figure 4A, left, and Table 1). We also
detected nearly twice as many DASEs for LPS/IFN␥com-
pared to IL4/13 (Figure 4A, right, and Table 1). The DASE
co-detection by more than one tool is very low (Supplemen-
tary Figure S5A, bottom), and the AS type distribution in
THP-1 conditions shows a similar pattern as above, with the
exception of higher frequency of mutually exclusive exons in
LPS/IFN␥stimulation (Supplementary Figure S5B, right).
By comparing these THP-1 DASEs with those of analo-
gous primary cells, we found a small yet enriched overlap
for macrophage differentiation and for LPS/IFN␥activa-
tion, but no overlap for IL4 (Figure 4B).
We tested as many as 45 DASEs in THP-1 (Figure 4C
for 17 DASEs not counting SAP25, and 28 more in Sup-
plementary Figure S6), plus 14 more in both THP-1 and
primary cells (Figure 5D, not counting NUMB as it is re-
peated), for a total of 59. For all the PSI data points, we used
the correlation coefcient between RNA-seq and RT-PCR,
so as to monitor the trend in PSI changes across condi-
tions rather than precise PSI values which are biased by am-
plication cycles (Supplementary Figure S7A–C). We con-
sidered a total of 47 (80%) as ‘validated’ as their correla-
tion coefcient exceeds the cutoff of 0.4, while 33 (56%) are
highly reliable with a coefcient >0.8 (Supplementary Fig-
ure S7D). A good example of validated DASE is INF2 cas-
sette exon which, in both RNA-seq and RT-PCR, switches
from predominant inclusion in untreated or VD3-treated
THP-1 cells to strong skipping in both PMA treatments and
in LPS/IFN␥or IL4/13 polarization. The 80% validation
rate shows an overall high concordance between RNA-seq
and RT-PCR.
Overall, we identied a large number of DASEs dur-
ing monocyte-to-macrophage differentiation and activa-
tion, which are regulated by one or likely more SFs. The
best candidate AS regulators are the SFs with strongest ex-
pression changes during each process. Below we character-
ize two of these SFs.
Changes in RBFOX2 and MBNL1 expression are associated
with differential AS between monocytes and macrophages
Next we sought out to establish the role of RBFOX2 and
MBNL1 in regulating AS during monocyte-to-macrophage
differentiation. Our RNA-seq data revealed an increase in
RBFOX2 expression upon GM-CSF or PMA treatment
of primary or THP-1 cells, respectively, and a decrease
in MBNL1 in M/GM-CSF and PMA treated cells (Fig-
ure 2C). While MBNL2 did not change expression dur-
ing monocyte-to-macrophage differentiation, MBNL3 was
also downregulated but did not pass the statistical cutoff,
so we focused on MBNL1 only. We conrmed the MBNL1
and RBFOX2 ndings by real-time RT-PCR (Figure 5A).
Furthermore, these mRNA changes led to the correspond-
ing changes of RBFOX2 and MBNL1 protein levels upon
PMA treatment of THP-1 cells as seen by western blotting
(Figure 5B), suggesting that these SFs might regulate a frac-
tion of the DASEs during monocyte differentiation.
Consistent with a role of MBNL1 in monocyte AS,
we found a large overlap of our DASEs during mono-
cyte differentiation and the DASEs for a previous study
of MBNL1/2 knockdown in HEK293 and HeLa cell lines
(52)(Figure5C). Specically, among the 134 DASEs upon
shMBNL1/2 in this previous study, we identied 60 in
our primary monocyte differentiation under GM/M-CSF
treatment, with 12 also detected in THP-1 differentiation.
In addition, we found hundreds of DASEs upon primary
monocyte differentiation with RBFOX2 binding sites (UG-
CAUG) in the regulated exons or 300 nt of the anking in-
trons, with more than 80 DASEs in common with THP-1
differentiation. We validated 15 DASEs by radioactive RT-
PCR (Figure 5D). Among these, ten were affected by the
above-mentioned knockdown of MBNL1/2 in the previous
study (green gene names), three harbour one or more RB-
FOX2 binding sites (red gene names), and ve additional
ones were not previously documented as MBNL1 targets
but had RBFOX2 binding sites (blue gene names). Loca-
tions of RBFOX2 binding sites are indicated in Supple-
mentary Figure S8. All these DASEs showed a consistent
change in AS in both primary monocytes treated with M-
and/or GM-CSF, and in THP-1 cells incubated with PMA.
From these, inclusion of the cassette exon increased upon
differentiation in seven DASEs (NUMB, SBF1, ADAM15,
GOLIM4, GSNK1G3, PPP1R12A and FNBP1), decreased
in seven DASEs (MGRN1, VPS29, MBNL1, SCARB1,
LRRFIP2, NUMA1 and PLD1), and switched 5ss usage
in NCOR2. This AS proles suggest that the two SFs might
regulate these DASEs in monocyte-to-macrophage differ-
entiation.
These DASEs might be biologically relevant in the con-
text of monocyte-to-macrophage differentiation, such as in
GOLIM4 (golgi integral membrane protein 4) involved in
trafcking between the cis Golgi and endosomes (53), or
NCOR2 (Nuclear Receptor Corepressor 2) as transcrip-
tional repressor via chromatin remodelling (Figure 5e) (54).
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

6078 Nucleic Acids Research, 2018, Vol. 46, No. 12
Table 1. Binary comparisons of differential AS events
Inclusion events Skipping events
Human primary monocytes Mo GM-CSF 2574 1794
Mo M-CSF 2264 1433
GM-CSF LPS/IFN␥448 357
M-CSF IL4 34 27
M-CSF IL1␤25 13
M-CSF IL10 17 12
THP-1 cell line THP-1 PMA3D 305 267
THP-1 PMA5Dr 349 211
THP-1 VD3 1142 519
PMA3d LPS/IFN␥224 131
PMA3d IL4/IL13 133 75
MBNL1 regulates AS during monocyte-to-macrophage dif-
ferentiation
To test the role of MBNL1 and RBFOX2 in AS dur-
ing monocyte differentiation, we managed to knock these
two SFs down in THP-1 cells by two different methods:
by stably expressing a short-hairpin RNA (shRNA) using
lentiviral transduction, and by transient nucleofection of
dicer-substrate RNA (dsiRNA), targeting two different ex-
ons (Supplementary Table S1). As MBNL1 and RBFOX2
are respectively down- and upregulated upon monocyte-to-
macrophage differentiation, we knocked these SFs down
when they are highly expressed, i.e. in untreated THP-1
for MBNL1 and in PMA3D cells for RBFOX2. By real-
time RT-PCR and Western Blotting, we observed that these
knockdowns reduced the target RNAs and proteins to dif-
ferent extents (Figure 6A and B). In the western blot, each
band in the MBNL1 doublet probably contains more than
one MBNL1 isoform, but based on the AS of exon 7 and 8
in our RNA-Seq/RT-PCR, the lower band most likely con-
tains isoform(s) with exon 7, while the upper band might
contain isoforms with exon 6 spliced directly to 8 (55).
We expected that the MBNL1 targets change AS
upon knockdown in THP-1 in the same direction than
upon PMA differentiation. Among the eight predicted
MBNL1 targets we tested, six showed the same change
in AS upon MBNL1 knockdown (especially in sh2-
MBNL1) and PMA, which included NCOR2, MGRN1,
NUMB, SCARB1, LRRFIP2, NUMA1, but not SBF1 and
ADAM15 (Figure 6C). Interestingly, ve out of the other
six DASEs also clearly responded to MBNL1 knockdown
in the expected direction, including GOLIM4, CSNK1G3,
PPP1R12A, FNBP1, PLD1, ARAP1 but not CSNK1G3.
Even though sh2-MBNL1 slightly reduces RBFOX2 pro-
tein, all conrmed DASEs show consistent effects with dsi1-
MBNL1, which does not have such cross-effect. Our nding
that eleven of the fourteen tested DASEs are regulated by
MBNL1 in monocytes suggest that this SF is an important
AS regulator in these cells.
In turn, we expected that the RBFOX2 targets change AS
upon knockdown in the reverse direction as in PMA treat-
ment of THP-1 cells, but only GOLIM4 among the nine
predicted targets showed this pattern (Figure 6C). Knock-
down of RBFOX2 had no consistent effect on the AS of
ADAM15, CSNK1G3, FNBP1, and PPP1R12A. In addi-
tion, dsi2-RBFOX2 changed AS of LRRFIP2, NUMA1,
PLD1 and ARAP1 in the opposite direction as the expected
one, but these effects are less reliable because they are not
shared with sh1-RBFOX2, and because this knockdown
also slightly upregulates MBNL1 protein. Thus, most of the
RBFOX2 AS targets in macrophages remain to be identi-
ed.
Next, we performed RNA pulldowns to investigate
whether MBNL1 and RBFOX2 regulate DASEs by di-
rect binding to the target RNAs. Based on predictions of
MBNL1 and RBFOX2 motifs by SpliceAid2 (56), we de-
signed probes for exon 47 of NCOR2 and exon 19 of LR-
RFIP2 because both are clearly regulated by MBNL1 (Fig-
ure 6C), as well as exon 7 of GOLIM4, regulated by both
SFs (Supplementary Figure S8). RNA pulldowns show that
RBFOX2 binds to probes with its cognate motif (Figure 6D,
left), with a stronger signal for GOLIM4 probe 1 than for
LRRFIP2 probe 4. This binding is specic because it is dra-
matically reduced by mutating the two motifs in GOLIM4
(Figure 6D, GOLIM4-1mut2). Consistent with the presence
of MBNL1 binding sites in all probes, MBNL1 is pulled
down by all except NCOR2 probes 2 and 3 (Figure 6D, left),
at different efciencies. MBNL1 binding was abrogated in
the mutant LRRFIP2 probe 2 (Figure 6D, right) but not
in other mutant probes, suggesting that MBNL1 might
bind these RNAs at multiple sites. These RNA pulldowns
are consistent with MBNL1-mediated regulation of DASEs
during monocyte-to-macrophage differentiation via direct
binding to these targets.
MBNL1 but not RBFOX2 is essential for monocyte-to-
macrophage differentiation
Finally, we addressed the functional signicance of these
two SFs in THP-1 differentiation by PMA. Our abovemen-
tioned results suggest that MBNL1 knockdown in THP-1
cells might recapitulate some of the phenotypic changes in-
duced by PMA. To test this hypothesis, we induced PMA
differentiation in the stable MBNL1 and RBFOX2 knock-
down cells, in comparison to differentiation in the origi-
nal and control shRNA THP-1 cells. Surprisingly, we found
a very strong impairment of PMA differentiation upon
MBNL1 knockdown, as most PMA-treated THP-1 cells
remained round-shaped (Figure 7A) like monocytes, and
the macrophage-specic cell adhesion marker CD54 was
far less induced in these cells compared to controls (Figure
7B). Consistently, MBNL1 knockdown with dsi1 showed a
smaller yet signicant reduction of PMA-triggered CD54
compared to control (Figure 7C). The fact that dsi1 and
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

Nucleic Acids Research, 2018, Vol. 46, No. 12 6079
Figure 4. Alternative splicing in THP-1 cells. (A) Venn diagrams with the numbers of DASEs (PSI change ≥0.05; SpliceTrap P-value ≤0.001, MISO BFmin
≥1 & BFmedian ≥10, MATS FDR ≤0.05) for the indicated treatments of PMA/VD3 differentiation (left) and polarization (right). In parentheses, number
of genes affected by DASEs. The PMA shared events are highly enriched, while the overlap with VD3 DASEs is also higher than expected (P∼0 and P
<10−5, one-tailed Fisher’s Exact test). (B) Venn diagrams showing the numbers of DASEs that are common in primary and THP-1 cells upon analogous
stimuli. In parentheses, number of genes affected by DASEs. The overlaps for differentiation and LPS activation are signicantly enriched (both P<10−10 ,
Fisher’s Exact test). (C) Radioactive RT-PCR validation of DASEs identied in THP-1 differentiation. Bar chart on top shows the RNA-seq-derived PSI
of each DASE with its type in parenthesis, with cassette exons (CA), intron retention (IR) and alternative 3splice sites (AA). The radioactive PAGE shows
the RT-PCR results with their PSI quantication at the bottom of each lane. The lack of some graph bars in SAP25 is due to the lack of detection in
certain conditions, yet it shows a strong reduction in cassette exon inclusion by PMA. All DASEs shown here are validated except MATK and ZNF678
(Supplementary Figure S7A).
sh2 target different MBNL1 sequences strongly argue for
the specicity of these results, and the degree of knockdown
corresponds to the magnitude of the phenotypic change. In
turn, we found no visible phenotype upon RBFOX2 down-
regulation (Figure 7A and B). We also performed RNA-seq
on the THP-1 cells with MBNL1 knockdown, and found an
enriched overlap with the DASEs induced by PMA, and in
either with PMA and/or M/GM-CSF macrophages, which
should account for the impaired differentiation phenotype
(Figure 7D; Supplementary les AS shMBNL1 gene list
and DEG gene list). In summary, these results establish the
key role of MBNL1 in PMA differentiation of THP-1 cells,
which suggests that this factor is also crucial for primary
monocyte differentiation.
DISCUSSION
We unveiled thousands of DASEs during human monocyte-
to-macrophage differentiation and activation in culture pri-
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

6080 Nucleic Acids Research, 2018, Vol. 46, No. 12
Figure 5. Regulation of MBNL1 and RBFOX2 expression upon primary monocyte-to-macrophage differentiation and their potential AS targets. ( A)
Quantitative real-time RT-PCR conrms the changes in mRNA levels of MBNL1 (top) and RBFOX2 (bottom) during THP-1 and primary cell differen-
tiation. Y axis indicates CT values of target relative to ␤-Actin control. (B) Western blotting shows that MBNL1 and RBFOX2 proteins are respectively
down- and upregulated upon differentiation of primary monocytes and THP-1 cells, with tubulin as loading control. (C) Venn diagram with the numbers
of DASEs for primary monocyte and THP-1 differentiation versus previously identied DASEs affected by siMBNL1+2 in HeLa and HEK293T (52). The
overlap is signicantly enriched (P<10−25, one-tailed Fisher’s Exact test). (D) Radioactive RT-PCR validation of potential MBNL1 and/or RBFOX2
DASEs conserved in THP-1 and primary cell differentiation. AS type in parenthesis, including cassette exons (CA) and alternative 5splice sites (AD).
Green genes indicate DASEs affected by siMBNL1+2, blue genes have one or more RBFOX2 binding sites within the target exon and/or 300nt in anking
introns, and red genes have both. Top bar chart shows the RNA-seq-derived PSI of each DASE, while PAGE shows the RT-PCR results with PSI quan-
tication of three replicates at the bottom of each lane. All these DASEs are validated (Supplementary Figure S7B). (E) RNA-seq screenshot proles for
two DASEs, with the read density for each sequence in gray aligned to the exon(box)-intron(line) structure below.
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

Nucleic Acids Research, 2018, Vol. 46, No. 12 6081
Figure 6. MBNL1 protein regulates 11 out of 14 DASEs during monocyte-to-macrophage differentiation. (A) Quantitative real-time RT-PCR shows
effective knockdown of MBNL1 mRNA by dsi1-MBNL1 nucleofection and sh2-MBNL1 transduction (left), and knockdown of RBFOX2 mRNA by dsi2-
RBFOX2 nucleofection and sh1-RBFOx2 transduction (right), all in THP-1. Knockdown of MBNL1 or RBFOX2 has no major effect on one another’s
mRNA. (B) Western blotting shows MBNL1 protein downregulation by sh2-MBNL1 and slightly by dsi1-MBNL1, as well as RBFOX2 downregulation by
both sh1-RBFOX2 and dsi2-RBFOX2. Each knockdown did not affect the other protein with the exception of sh2-MBNL1 cells showing some RBFOX2
down-regulation, which was not seen in dsi1-MBNL1 and shRNA control. Lane labels at the bottom. (C) Radioactive RT-PCR showing that MBNL1
knockdown in THP-1 cells changed 10 DASEs (green labels) in the same direction as PMA3D treatment, while both MBNL1 and RBFOX2 knockdowns
in PMA3D cells changed the DASE of GOLIM4 (red) in the expected direction. AS type indicated in parenthesis, including cassette exons (CA) and
alternative 5splice sites (AD). Bottom right corner, mini-table summary of the 14 DASE changes upon SF knockdown. (D) Western blotting analysis of
RNA pulldowns with 139-nt probes for NCOR2 (probes 1–3), LIRRFIP2 (probes 1–4), and GOLIM4 (probe 1), numbered from 5to 3(Supplementary
Figure S8). Wild-type probes on the left and mutants on the right. FT, ow-through.
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

6082 Nucleic Acids Research, 2018, Vol. 46, No. 12
Figure 7. Knockdown of MBNL1 causes a strong deciency in PMA differentiation of THP-1 cells. (A) Left, phase-contrast microscope images showing
THP-1 differentiation upon PMA3D treatment, with control and shRNA-expressing cells. Right, quantication of PMA-differentiated cells. We counted
the extended irregular shaped cells as differentiated, while small round cells as un-differentiated. Differentiated cells ratio =number of differentiated cells
divided by total. Bar chart derived from three biological replicates with ∼100 counted cells per replicate. (B) Left, FACS proles of CD54 expression in
control, MBNL1 or RBFOX2 shRNA knockdown in untreated THP-1 and PMA3D. Right, median uoresce intensity (MFI) of three biological replicates.
(C) MFI of THP-1 and PMA3D cells treated with indicated dsiRNAs. (D) Venn diagrams of RNA-seq-detected DASEs (only with rMATS) upon MBNL1
knockdown of THP-1 cells in common with those in PMA differentiation (left) and in M/GM-CSF differentiation of primary cells (right). The overlaps
between MBNL1 DASEs and those for PMA conditions or both PMA and M/GM-CSF are signicantly enriched (both P<10−30 , one-tailed Fisher’s
Exact test). (E) Heatmap of RNA levels of MBNL1/3, CELF1, CNBP and DMPK in primary monocyte-to-macrophage differentiation and LPS/IFN␥
activation, whose expression patterns suggest a possible monocyte dysfunction in myotonic dystrophies.
mary cells and THP-1 cell line. Importantly, from the many
RBPs and SFs with strong expression changes, MBNL1 ap-
pears to be a major regulator of AS in monocytes and is es-
sential for PMA differentiation of THP-1 cells. This study
should be a major stepping stone for the detailed charac-
terization of AS regulation in these important myeloid cells
and their connection to disease and therapy.
AS is a major regulatory process in human monocytes and
macrophages
We found that AS plays a prominent role in both M-
and GM-CSF differentiation of primary monocytes, with
many DASEs in common and many that are specic to
these stimuli. The differential pathways that are regulated
by DASEs upon these treatments strongly suggest that
AS contributes to the distinct properties of these cells. As
GM- and M-CSF macrophages have mostly pro- and anti-
inammatory properties and are associated with different
disease states, their differential regulation is crucial to un-
derstand and manipulate these cells (3). For instance, re-
pressing tumor associated macrophages with a wound heal-
ing phenotype (like M-CSF macrophages) might have ther-
apeutic benet (3,15). QKI was shown to regulate AS in M-
CSF macrophages (37), consistent with our RNAseq data
showing a ∼30% increase in its RNA levels specic to these
cells. We only detected a reduction in expression of SF3A1-
3 and SRSF3 in macrophages, despite the reported role of
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

Nucleic Acids Research, 2018, Vol. 46, No. 12 6083
the SF3A subcomplex in AS during TLR4 stimulation of
mouse macrophages (36), and the role of SRSF3 in IL1␤
production (42). Our study adds MBNL1 as an important
AS regulator in monocytes.
We also revealed hundreds of DASEs in polarization
of GM-CSF macrophages by LPS/IFN␥, but very few in
the IL4/IL1␤/IL10 polarizations of M-CSF macrophages,
with almost none in common. This nding might simply
reect that AS does not play a prominent role in anti-
inammatory responses of macrophages, as opposed to dif-
ferentiation and LPS/IFN␥activation. Alternatively, even
though we used well-established protocols, the lack of AS
changes in anti-inammatory polarizations could be due
to the different doses (100 ng/ml LPS versus 10 ng/ml
IL4/IL1␤/IL10), to treatment times (8 h in all), or to M-
CSF macrophages being already fully activated. In THP-
1,alowerPMAdoseof10ng/mL results in stronger
IL4/IL13 activation (57), suggesting that an adjustment of
cytokine treatments might reveal more DASEs. Finally, we
conrmed the upregulation of CELF1 by LPS/IFN␥(38)
(Figure 7E).
Known and new RBPs as regulators of monocytes and
macrophages
Our study revealed a large number of RBPs that strongly
change expression (≥2-fold up or down) in the different pri-
mary monocytes and macrophages. RBPs that also change
in analogous THP-1 conditions may contribute to the com-
mon function of these cells. Encouragingly, we found many
RBPs previously implicated in monocyte/macrophage
functions. For differentiation, the previously known DEGs
include (i) CPEB1 depletion in macrophages, which en-
hances IL6 production upon LPS treatment, by regulat-
ing translation of other mRNAs (58); (ii) macrophage-
enhanced downregulation of RNASE2 and RNASE3,
which encode mostly eosinophil-derived secretory ribonu-
cleases that ght infection (59); (iii) ZFP36 downregulation,
which in turn mediate the AU-Rich Element-dependent
degradation of TNF-␣mRNA in liver macrophages, among
other targets (60); ZFP36 and ZFP36L2 are also respec-
tively up and downregulated by LPS treatment of mouse
macrophages (61). The remaining conserved RBPs are un-
characterized in monocytes/macrophages, and might also
regulate their functions.
For LPS/IFN␥activation, the RBPs as conserved DEGs
in primary and THP-1 cells that have been previously char-
acterized include: (i) the p150 ADAR isoform upregulated
by the interferon-inducible promoter (62); (ii) ZC3HAV1
upregulated in macrophages along other antiviral genes
(63); (iii) increased DCP1A expression which regulates IL6
secretion in THP-1 (64); (iv) PNPT1 upregulation which
induces pro-inammatory cytokines IL6 and IL8 (65);
(v) MSI2 downregulation, which controls progression of
Chronic Myeloid Leukemia via NUMB (66). The biological
signicance of the LPS/IFN␥-induced expression changes
of the other RBPs remains to be elucidated.
Several transcription factors that control monocytes
and/or macrophages are acquiring diagnostic, prognostic
or therapeutic value, like Nuclear Factor kappa B (NF-␬B)
associated with cancer inammation and tumor progression
(67), or Peroxisome Proliferator-activated Receptor gamma
(PPAR␥) and Kruppel-like factor 4 (KLF4) associated with
protection from obesity-induced insulin resistance (68,69).
Furthermore, microRNAs play a role in endotoxin toler-
ance (70). Certain RBPs/SFs, other than QKI in atheroge-
nesis (37), should be soon linked to monocyte/macrophage
disease.
Role of RBFOX2 and MBNL1 during monocyte-to-
macrophage differentiation
Several ndings initially supported a role of RBFOX2 and
MBNL1 as AS regulators in human monocyte differentia-
tion, including their respective up- and downregulation dur-
ing differentiation of both primary monocytes and THP-1
cells, and the overlap between our differentiation DASEs
and known MBNL1 targets in other cells (52). However, our
analysis of the role of these SFs upon knockdown in THP-1
only claried the picture for MBNL1. For RBFOX2, only
one DASE (GOLIM4) changed AS in the expected direc-
tion upon knockdown, indicating that the RBFOX2 targets
in macrophages remain to be identied. Nevertheless, RB-
FOX2 knockdown in PMA-treated THP-1 cells had no vis-
ible phenotype, as these cells remained differentiated and
attached, strongly suggesting that this factor is dispensable
for the establishment or maintenance of the PMA differ-
entiated state. It could also be that the slight knockdown
is not sufcient or that other paralogs compensate for the
RBFOX2 reduction. Thus, further studies are required to
establish the role of RBFOX2 in macrophages.
Remarkably, many DASEs changed AS upon MBNL1
knockdown in the same direction as during monocyte dif-
ferentiation (Figures 6Cand7C), arguing that the over-
all MBNL1 reduction, rather than the changes in the
relative expression of MBNL1 splice isoforms, largely
caused the differential AS patterns between monocytes
and macrophages. Our dataset includes known and new
MBNL1 targets, and for the two DASEs we checked by
RNA pulldowns the regulation may be via direct bind-
ing (Figure 6D), to be conrmed by other approaches to
monitor binding within cells such as Cross-Linking and
ImmunoPrecipitation (CLIP). Most importantly, MBNL1
knockdown in THP-1 very strongly impaired PMA differ-
entiation, indicating that the MBNL1-mediated regulation
of AS or other processes such as mRNA localization (71)is
essential for the differentiation of this cell line and probably
of primary monocytes.
MBNL1 in human monocyte/macrophage physiology and
disease
MBNL1 increases expression and regulates three develop-
mental transitions, which are embryonic stem cell differ-
entiation (52), the transition between fetal and adult tis-
sues, majorly including muscle, heart and brain (72–75),
and terminal erythroid differentiation (76). In contrast,
our human primary monocyte-to-macrophage differentia-
tion model shows >2-fold downregulation of total MBNL1
RNA and at least one protein isoform (Figure 6A and B).
Albeit monocytes and macrophages are both mature cells,
monocytes proliferate much more than macrophages, hence
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

6084 Nucleic Acids Research, 2018, Vol. 46, No. 12
in these myeloid cells low MBNL1 is associated with lower
proliferation, as opposed to stem cells. Furthermore, the
loss of PMA-induced THP-1 differentiation upon MBNL1
knockdown could be caused by a putative MBNL1 function
in ‘priming’ these monocytes for the differentiation pro-
cess. An alternative explanation based on MBNL1’s role
in developmental transitions is that its knockdown might
turn THP-1 cells into a more de-differentiated state which
renders these cells unresponsive to PMA. Consistent with
this notion, the DASEs of PMA-treated THP-1 cells and
upon MBNL1 knockdown do not show a perfect over-
lap, highlighting the role of other SFs in PMA differentia-
tion, and that MBNL1 knockdown might have turned these
cells into a new state, perhaps reminiscent of precursors
in the myeloid lineage. Future experiments should extend
this nding to primary cells and fully characterize the iden-
tity of these MBNL1 knockdown cells. Furthermore, such
follow-up should address whether MBNL1 overexpression
in macrophages disrupts their identity and/or function.
The important AS regulation by MBNL1 in human
monocytes and its role during differentiation revealed a
putative unexpected connection to Myotonic Dystrophies
DM1 and DM2. DM1/2 are caused by repeat expansions
in RNAs which sequester MBNL1, thus resulting in a
loss of function of this SF and turning the AS in adult
tissues to that of undifferentiated cells, mainly resulting
in muscular dystrophy, myotonia as well as cardiac and
neurological problems (73,77). The causative genes with
the expansions, DMPK for the severe DM1 and CNBP
(ZNF9) for the milder DM2, need to be expressed in the
cells together with MBNL1 for the loss of function to
occur. Indeed, both DMPK and CNBP are highly ex-
pressed in human monocytes compared to macrophages
(Figure 7E), suggesting that human monocytes in DM pa-
tients suffer from a loss of function of MBNL1, and that
monocyte-derived macrophages would be defective in these
patients. Furthermore, CELF1, whose gain of function
upon phosphorylation-induced stabilization contributes to
DM1 (78), is strongly induced by LPS (38), suggesting
that activated inammatory macrophages might also ex-
hibit phenotypic defects in DM1 patients. DM patients
show abnormally high blood insulin together with its low
binding to INSR (insulin receptor) in monocytes (79,80).
Even though MBNL1 supports strong binding to insulin
by inducing inclusion of INSR exon 11 (81), this cassette
exon showed very low inclusion level in all our samples
from primary and THP-1 cells, as seen before (82), suggest-
ing that the high insulin levels in DM patients are caused
by cells other than monocytes. DM monocytes also exhibit
abnormal expression of IgG-Fc receptors resulting in hy-
percatabolism of IgG (83), which could be caused by aber-
rant AS (or other processes) due to MBNL1 loss of func-
tion in monocytes and perhaps other leukocytes. Indeed, we
did detect a few DASEs in FCGR1B/2A/2B/3A genes in
primary monocytes and macrophages to be characterized
in future studies. Beyond these disruptions, other mono-
cyte functions and probably their ability to differentiate to
macrophages could be compromised in DM patients lead-
ing to immunosuppression, underlying the characteristic
muscle, cardiac and neurological disorders. This immuno-
suppression should be mild if any, otherwise it would have
been clinically described, or perhaps the monocytic defects
are compensated by the two paralogs MBNL2/3(84)or
by other factors. Indeed, a recent study showed differen-
tial alterations of AS patterns upon different degrees of
MBNL1 knockdown (85), which highlights the high sensi-
tivity of deregulated AS events to free MBNL1 levels which
could also be tissue specic. Further studies should clarify
the functional properties of monocytes and macrophages in
DM patients.
In brief, here we report an extensive and detailed char-
acterization of AS in human monocytic differentiation and
activation, and established an important yet intriguing role
of MBNL1 in AS of monocytes.
DATA AVAILABILITY
We deposited the raw fastq les in the Sequence Read
Archives (SRA) of the National Center for Biotechnology
Information (NCBI) under accession number SRP139891
of Bioproject PRJNA449980.
SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online.
ACKNOWLEDGEMENTS
We thank members of the Roca lab for useful advice. We are
grateful to Daniela Moses for RNA library preparation and
for running the Illumina Hi-Seq machine. We also thank
Christiane Ruedl and Sin Tiong Ong for advice.
Author contributions: H.L. performed most of the experi-
ments and in silico analyses, with critical contributions from
P.A.L. F.Z. was instrumental in setting the bioinformatics
pipeline with supervision from J.Z. M.S.M.W. performed
many of the A.S. validations, and S.X. did some of the
last experiments with MBNL1. X.R. conceived the study
and wrote the manuscript with help from coauthors, mainly
from H.L.
FUNDING
Academic Research Fund Tier 2 [MOE2013-T2-1-101
ARC 45/13, MOE2016-T2-2-104(S)], both from Singa-
pore’s Ministry of Education (https://www.olga.moe.gov.sg/
T2/default.aspx). The funders played no role in study de-
sign, data collection and analysis, decision to publish or
preparation of the manuscript. Funding for open access
charge: Academic Research Fund Tier 2 [MOE2016-T2-2-
104(S)].
Conict of interest statement. None declared.
REFERENCES
1. Auffray,C., Sieweke,M.H. and Geissmann,F. (2009) Blood
monocytes: development, heterogeneity, and relationship with
dendritic cells. Annu. Rev. Immunol.,27, 669–692.
2. Ginhoux,F. and Jung,S. (2014) Monocytes and macrophages:
developmental pathways and tissue homeostasis. Nat. Rev. Immunol.,
14, 392–404.
3. Biswas,S.K., Chittezhath,M., Shalova,I.N. and Lim,J.Y. (2012)
Macrophage polarization and plasticity in health and disease.
Immunol. Res.,53, 11–24.
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

Nucleic Acids Research, 2018, Vol. 46, No. 12 6085
4. Kohro,T., Tanaka,T., Murakami,T., Wada,Y., Aburatani,H.,
Hamakubo,T. and Kodama,T. (2004) A comparison of differences in
the gene expression proles of phorbol 12-myristate 13-acetate
differentiated THP-1 cells and human monocyte-derived
macrophage. J. Atheroscler. Thromb.,11, 88–97.
5. Qin,Z. (2012) The use of THP-1 cells as a model for mimicking the
function and regulation of monocytes and macrophages in the
vasculature. Atherosclerosis,221, 2–11.
6. Lavin,Y., Mortha,A., Rahman,A. and Merad,M. (2015) Regulation
of macrophage development and function in peripheral tissues. Nat.
Rev. Immunol.,15, 731–744.
7. Varol,C., Mildner,A. and Jung,S. (2015) Macrophages: development
and tissue specialization. Annu. Rev. Immunol.,33, 643–675.
8. Ginhoux,F., Schultze,J.L., Murray,P.J., Ochando,J. and Biswas,S.K.
(2016) New insights into the multidimensional concept of
macrophage ontogeny, activation and function. Nat. Immunol.,17,
34–40.
9. Alasoo,K., Martinez,F.O., Hale,C., Gordon,S., Powrie,F.,
Dougan,G., Mukhopadhyay,S. and Gaffney,D.J. (2015)
Transcriptional proling of macrophages derived from monocytes
and iPS cells identies a conserved response to LPS and novel
alternative transcription. Sci. Rep.,5, 12524.
10. Sharif,O., Bolshakov,V.N., Raines,S., Newham,P. and Perkins,N.D.
(2007) Transcriptional proling of the LPS induced NF-kappaB
response in macrophages. BMC Immunol.,8,1.
11. Gordon,S. and Martinez,F.O. (2010) Alternative activation of
macrophages: mechanism and functions. Immunity,32, 593–604.
12. Mantovani,A., Sozzani,S., Locati,M., Allavena,P. and Sica,A. (2002)
Macrophage polarization: tumor-associated macrophages as a
paradigm for polarized M2 mononuclear phagocytes. Trends
Immunol.,23, 549–555.
13. Lumeng,C.N., Bodzin,J.L. and Saltiel,A.R. (2007) Obesity induces a
phenotypic switch in adipose tissue macrophage polarization. J. Clin.
Invest.,117, 175–184.
14. Moore,K.J. and Tabas,I. (2011) Macrophages in the pathogenesis of
atherosclerosis. Cell,145, 341–355.
15. Qian,B.Z. and Pollard,J.W. (2010) Macrophage diversity enhances
tumor progression and metastasis. Cell,141, 39–51.
16. Inoue,D., Bradley,R.K. and Abdel-Wahab,O. (2016) Spliceosomal
gene mutations in myelodysplasia: molecular links to clonal
abnormalities of hematopoiesis. Genes Dev.,30, 989–1001.
17. Yoshida,K., Sanada,M., Shiraishi,Y., Nowak,D., Nagata,Y.,
Yamamoto,R., Sato,Y., Sato-Otsubo,A., Kon,A., Nagasaki,M. et al.
(2011) Frequent pathway mutations of splicing machinery in
myelodysplasia. Nature,478, 64–69.
18. Stumpo,D.J., Lai,W.S. and Blackshear,P.J. (2010) Inammation:
cytokines and RNA-based regulation. Wiley Interdiscip. Rev. RNA,1,
60–80.
19. Martinez,N.M., Pan,Q., Cole,B.S., Yarosh,C.A., Babcock,G.A.,
Heyd,F., Zhu,W., Ajith,S., Blencowe,B.J. and Lynch,K.W. (2012)
Alternative splicing networks regulated by signaling in human T cells.
RNA,18, 1029–1040.
20. O’Connell,R.M., Rao,D.S. and Baltimore,D. (2012) microRNA
regulation of inammatory responses. Annu. Rev. Immunol.,30,
295–312.
21. Lee,Y. and Rio,D.C. (2015) Mechanisms and regulation of alternative
pre-mRNA splicing. Annu. Rev. Biochem.,84, 291–323.
22. Nilsen,T.W. and Graveley,B.R. (2010) Expansion of the eukaryotic
proteome by alternative splicing. Nature,463, 457–463.
23. Busch,A. and Hertel,K.J. (2012) Evolution of SR protein and hnRNP
splicing regulatory factors. Wiley Interdiscip. Rev. RNA,3, 1–12.
24. Warf,M.B. and Berglund,J.A. (2010) Role of RNA structure in
regulating pre-mRNA splicing. Trends Biochem. Sci.,35, 169–178.
25. Naftelberg,S., Schor,I.E., Ast,G. and Kornblihtt,A.R. (2015)
Regulation of alternative splicing through coupling with transcription
and chromatin structure. Annu. Rev. Biochem.,84, 165–198.
26. Licatalosi,D.D. and Darnell,R.B. (2010) RNA processing and its
regulation: global insights into biological networks. Nat. Rev. Genet.,
11, 75–87.
27. Barbosa-Morais,N.L., Irimia,M., Pan,Q., Xiong,H.Y.,
Gueroussov,S., Lee,L.J., Slobodeniuc,V., Kutter,C., Watt,S., Colak,R.
et al. (2012) The evolutionary landscape of alternative splicing in
vertebrate species. Science,338, 1587–1593.
28. Merkin,J., Russell,C., Chen,P. and Burge,C.B. (2012) Evolutionary
dynamics of gene and isoform regulation in Mammalian tissues.
Science,338, 1593–1599.
29. Martinez,N.M. and Lynch,K.W. (2013) Control of alternative
splicing in immune responses: many regulators, many predictions,
much still to learn. Immunol. Rev.,253, 216–236.
30. Chen,L., Kostadima,M., Martens,J.H., Canu,G., Garcia,S.P.,
Turro,E., Downes,K., Macaulay,I.C., Bielczyk-Maczynska,E., Coe,S.
et al. (2014) Transcriptional diversity during lineage commitment of
human blood progenitors. Science,345, 1251033.
31. Edwards,C.R., Ritchie,W., Wong,J.J., Schmitz,U., Middleton,R.,
An,X., Mohandas,N., Rasko,J.E. and Blobel,G.A. (2016) A dynamic
intron retention program in the mammalian megakaryocyte and
erythrocyte lineages. Blood,127, e24–e34.
32. Pimentel,H., Parra,M., Gee,S.L., Mohandas,N., Pachter,L. and
Conboy,J.G. (2016) A dynamic intron retention program enriched in
RNA processing genes regulates gene expression during terminal
erythropoiesis. Nucleic Acids Res.,44, 838–851.
33. Wong,J.J., Ritchie,W., Ebner,O.A., Selbach,M., Wong,J.W.,
Huang,Y., Gao,D., Pinello,N., Gonzalez,M., Baidya,K. et al. (2013)
Orchestrated intron retention regulates normal granulocyte
differentiation. Cell,154, 583–595.
34. Bhatt,D.M., Pandya-Jones,A., Tong,A.J., Barozzi,I., Lissner,M.M.,
Natoli,G., Black,D.L. and Smale,S.T. (2012) Transcript dynamics of
proinammatory genes revealed by sequence analysis of subcellular
RNA fractions. Cell,150, 279–290.
35. Pandya-Jones,A., Bhatt,D.M., Lin,C.H., Tong,A.J., Smale,S.T. and
Black,D.L. (2013) Splicing kinetics and transcript release from the
chromatin compartment limit the rate of Lipid A-induced gene
expression. RNA,19, 811–827.
36. O’Connor,B.P., Danhorn,T., De Arras,L., Flatley,B.R.,
Marcus,R.A., Farias-Hesson,E., Leach,S.M. and Alper,S. (2015)
Regulation of toll-like receptor signaling by the SF3a mRNA splicing
complex. PLoS Genet.,11, e1004932.
37. de Bruin,R.G., Shiue,L., Prins,J., de Boer,H.C., Singh,A., Fagg,W.S.,
van Gils,J.M., Duijs,J.M., Katzman,S., Kraaijeveld,A.O. et al. (2016)
Quaking promotes monocyte differentiation into pro-atherogenic
macrophages by controlling pre-mRNA splicing and gene expression.
Nat. Commun.,7, 10846.
38. Lin,J., Hu,Y., Nunez,S., Foulkes,A.S., Cieply,B., Xue,C., Gerelus,M.,
Li,W., Zhang,H., Rader,D.J. et al. (2016) Transcriptome-Wide
analysis reveals modulation of human macrophage inammatory
phenotype through alternative splicing. Arterioscler. Thromb. Vasc.
Biol.,36, 1434–1447.
39. Grisar,J., Munk,M., Steiner,C.W., Amoyo-Minar,L.,
Tohidast-Akrad,M., Zenz,P., Steiner,G. and Smolen,J.S. (2012)
Expression patterns of CD44 and CD44 splice variants in patients
with rheumatoid arthritis. Clin. Exp. Rheumatol.,30, 64–72.
40. Srinivasan,R. and Bogdanov,V.Y. (2012) Splice variants of Tissue
Factor and integrin-mediated signaling. Thromb. Res.,129(Suppl. 2),
S34–S37.
41. Burns,K., Janssens,S., Brissoni,B., Olivos,N., Beyaert,R. and
Tschopp,J. (2003) Inhibition of interleukin 1 receptor/Toll-like
receptor signaling through the alternatively spliced, short form of
MyD88 is due to its failure to recruit IRAK-4. J. Ex p. M e d . ,197,
263–268.
42. Moura-Alves,P., Neves-Costa,A., Raquel,H., Pacheco,T.R.,
D’Almeida,B., Rodrigues,R., Cadima-Couto,I., Chora,A.,
Oliveira,M., Gama-Carvalho,M. et al. (2011) An shRNA-based
screen of splicing regulators identies SFRS3 as a negative regulator
of IL-1beta secretion. PLoS One,6, e19829.
43. Daigneault,M., Preston,J.A., Marriott,H.M., Whyte,M.K. and
Dockrell,D.H. (2010) The identication of markers of macrophage
differentiation in PMA-stimulated THP-1 cells and monocyte-derived
macrophages. PLoS One,5, e8668.
44. Maess,M.B., Wittig,B. and Lorkowski,S. (2014) Highly efcient
transfection of human THP-1 macrophages by nucleofection. J. Vi s.
Exp., e51960.
45. Kim,D., Pertea,G., Trapnell,C., Pimentel,H., Kelley,R. and
Salzberg,S.L. (2013) TopHat2: accurate alignment of transcriptomes
in the presence of insertions, deletions and gene fusions. Genome
Biol.,14, R36.
46. Langmead,B. and Salzberg,S.L. (2012) Fast gapped-read alignment
with Bowtie 2. Nat. Methods,9, 357–359.
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018

6086 Nucleic Acids Research, 2018, Vol. 46, No. 12
47. Trapnell,C., Roberts,A., Goff,L., Pertea,G., Kim,D., Kelley,D.R.,
Pimentel,H., Salzberg,S.L., Rinn,J.L. and Pachter,L. (2012)
Differential gene and transcript expression analysis of RNA-seq
experiments with TopHat and Cufinks. Nat. Protoc.,7, 562–578.
48. Shen,S., Park,J.W., Lu,Z.X., Lin,L., Henry,M.D., Wu,Y.N., Zhou,Q.
and Xing,Y. (2014) rMATS: robust and exible detection of
differential alternative splicing from replicate RNA-Seq data. Proc.
Natl. Acad. Sci. U.S.A.,111, E5593–E5601.
49. Katz,Y., Wang,E.T., Airoldi,E.M. and Burge,C.B. (2010) Analysis
and design of RNA sequencing experiments for identifying isoform
regulation. Nat. Methods,7, 1009–1015.
50. Wu,J., Akerman,M., Sun,S., McCombie,W.R., Krainer,A.R. and
Zhang,M.Q. (2011) SpliceTrap: a method to quantify alternative
splicing under single cellular conditions. Bioinformatics,27,
3010–3016.
51. Neme,A., Nurminen,V., Seuter,S. and Carlberg,C. (2016) The
vitamin D-dependent transcriptome of human monocytes. J. Steroid
Biochem. Mol. Biol.,164, 180–187.
52. Han,H., Irimia,M., Ross,P.J., Sung,H.K., Alipanahi,B., David,L.,
Golipour,A., Gabut,M., Michael,I.P., Nachman,E.N. et al. (2013)
MBNL proteins repress ES-cell-specic alternative splicing and
reprogramming. Nature,498, 241–245.
53. Linstedt,A.D., Mehta,A., Suhan,J., Reggio,H. and Hauri,H.P. (1997)
Sequence and overexpression of GPP130/GIMPc: evidence for
saturable pH-sensitive targeting of a type II early Golgi membrane
protein. Mol. Biol. Cell,8, 1073–1087.
54. Chen,J.D. and Evans,R.M. (1995) A transcriptional co-repressor that
interacts with nuclear hormone receptors. Nature,377, 454–457.
55. Tran,H., Gourrier,N., Lemercier-Neuillet,C., Dhaenens,C.M.,
Vautrin,A., Fernandez-Gomez,F.J., Arandel,L., Carpentier,C.,
Obriot,H., Eddarkaoui,S. et al. (2011) Analysis of exonic regions
involved in nuclear localization, splicing activity, and dimerization of
Muscleblind-like-1 isoforms. J. Biol. Chem.,286, 16435–16446.
56. Piva,F., Giulietti,M., Burini,A.B. and Principato,G. (2012) SpliceAid
2: a database of human splicing factors expression data and RNA
target motifs. Hum. Mutat.,33, 81–85.
57. Jin,C., Wu,L., Li,J., Fang,M., Cheng,L. and Wu,N. (2012) Multiple
signaling pathways are involved in the interleukine-4 regulated
expression of DC-SIGN in THP-1 cell line. J. Biomed. Biotechnol.,
2012, 357060.
58. Ivshina,M., Alexandrov,I.M., Vertii,A., Doxsey,S. and Richter,J.D.
(2015) CPEB regulation of TAK1 synthesis mediates cytokine
production and the inammatory immune response. Mol. Cell. Biol.,
35, 610–618.
59. Acharya,K.R. and Ackerman,S.J. (2014) Eosinophil granule proteins:
form and function. J. Biol. Chem.,289, 17406–17415.
60. Carballo,E., Lai,W.S. and Blackshear,P.J. (1998) Feedback inhibition
of macrophage tumor necrosis factor-alpha production by
tristetraprolin. Science,281, 1001–1005.
61. Wang,K.T., Wang,H.H., Wu,Y.Y., Su,Y.L., Chiang,P.Y., Lin,N.Y.,
Wang,S.C., Chang,G.D. and Chang,C.J. (2015) Functional regulation
of Zfp36l1 and Zfp36l2 in response to lipopolysaccharide in mouse
RAW264.7 macrophages. J. Inamm. (Lond.),12, 42.
62. Blango,M.G. and Bass,B.L. (2016) Identication of the long, edited
dsRNAome of LPS-stimulated immune cells. Genome Res.,26,
852–862.
63. Oshiumi,H., Mifsud,E.J. and Daito,T. (2016) Links between
recognition and degradation of cytoplasmic viral RNA in innate
immune response. Rev. Med. Virol.,26, 90–101.
64. Seto,E., Yoshida-Sugitani,R., Kobayashi,T. and
Toyama-Sorimachi,N. (2015) The assembly of EDC4 and Dcp1a into
processing bodies is critical for the translational regulation of IL-6.
PLoS One,10, e0123223.
65. Sarkar,D. and Fisher,P.B. (2006) Molecular mechanisms of
aging-associated inammation. Cancer Lett.,236, 13–23.
66. Ito,T., Kwon,H.Y., Zimdahl,B., Congdon,K.L., Blum,J., Lento,W.E.,
Zhao,C., Lagoo,A., Gerrard,G., Foroni,L. et al. (2010) Regulation of
myeloid leukaemia by the cell-fate determinant Musashi. Nature,466,
765–768.
67. Biswas,S.K., Sica,A. and Lewis,C.E. (2008) Plasticity of macrophage
function during tumor progression: regulation by distinct molecular
mechanisms. J. Immunol.,180, 2011–2017.
68. Liao,X., Sharma,N., Kapadia,F., Zhou,G., Lu,Y., Hong,H.,
Paruchuri,K., Mahabeleshwar,G.H., Dalmas,E., Venteclef,N. et al.
(2011) Kruppel-like factor 4 regulates macrophage polarization. J.
Clin. Invest.,121, 2736–2749.
69. Odegaard,J.I., Ricardo-Gonzalez,R.R., Goforth,M.H., Morel,C.R.,
Subramanian,V., Mukundan,L., Red Eagle,A., Vats,D.,
Brombacher,F., Ferrante,A.W. et al. (2007) Macrophage-specic
PPARgamma controls alternative activation and improves insulin
resistance. Nature,447, 1116–1120.
70. Nahid,M.A., Satoh,M. and Chan,E.K. (2011) MicroRNA in TLR
signaling and endotoxin tolerance. Cell. Mol. Immunol.,8, 388–403.
71. Wang,E.T., Cody,N.A., Jog,S., Biancolella,M., Wang,T.T.,
Treacy,D.J., Luo,S., Schroth,G.P., Housman,D.E., Reddy,S. et al.
(2012) Transcriptome-wide regulation of pre-mRNA splicing and
mRNA localization by muscleblind proteins. Cell,150, 710–724.
72. Kalsotra,A., Xiao,X., Ward,A.J., Castle,J.C., Johnson,J.M.,
Burge,C.B. and Cooper,T.A. (2008) A postnatal switch of CELF and
MBNL proteins reprograms alternative splicing in the developing
heart. Proc. Natl. Acad. Sci. U.S.A.,105, 20333–20338.
73. Konieczny,P., Stepniak-Konieczna,E. and Sobczak,K. (2014) MBNL
proteins and their target RNAs, interaction and splicing regulation.
Nucleic Acids Res.,42, 10873–10887.
74. Lin,X., Miller,J.W., Mankodi,A., Kanadia,R.N., Yuan,Y.,
Moxley,R.T., Swanson,M.S. and Thornton,C.A. (2006) Failure of
MBNL1-dependent post-natal splicing transitions in myotonic
dystrophy. Hum. Mol. Genet.,15, 2087–2097.
75. Miller,J.W., Urbinati,C.R., Teng-Umnuay,P., Stenberg,M.G.,
Byrne,B.J., Thornton,C.A. and Swanson,M.S. (2000) Recruitment of
human muscleblind proteins to (CUG)(n) expansions associated with
myotonic dystrophy. EMBO J.,19, 4439–4448.
76. Cheng,A.W., Shi,J., Wong,P., Luo,K.L., Trepman,P., Wang,E.T.,
Choi,H., Burge,C.B. and Lodish,H.F. (2014) Muscleblind-like 1
(Mbnl1) regulates pre-mRNA alternative splicing during terminal
erythropoiesis. Blood,124, 598–610.
77. Thornton,C.A., Wang,E. and Carrell,E.M. (2017) Myotonic
dystrophy: approach to therapy. Curr. Opin. Genet. Dev.,44, 135–140.
78. Kuyumcu-Martinez,N.M., Wang,G.S. and Cooper,T.A. (2007)
Increased steady-state levels of CUGBP1 in myotonic dystrophy 1 are
due to PKC-mediated hyperphosphorylation. Mol. Cell,28, 68–78.
79. Mishra,S.K., Chetty,S. and Kataoka,M. (1992) Abnormal insulin
receptor binding in cultured monocytes in myotonic muscular
dystrophy. Biochem. Med. Metab. Biol.,47, 161–167.
80. Moxley,R.T. 3rd, Livingston,J.N., Lockwood,D.H., Griggs,R.C. and
Hill,R.L. (1981) Abnormal regulation of monocyte insulin-binding
afnity after glucose ingestion in patients with myotonic dystrophy.
Proc. Natl. Acad. Sci. U.S.A.,78, 2567–2571.
81. Echeverria,G.V. and Cooper,T.A. (2014) Muscleblind-like 1 activates
insulin receptor exon 11 inclusion by enhancing U2AF65 binding and
splicing of the upstream intron. Nucleic Acids Res.,42, 1893–1903.
82. Benecke,H., Flier,J.S. and Moller,D.E. (1992) Alternatively spliced
variants of the insulin receptor protein. Expression in normal and
diabetic human tissues. J. Clin. Invest.,89, 2066–2070.
83. Banerjee,D., McClintock,J., Silver,M.M. and Hudson,A.J. (1982)
Monocyte IgG-Fc receptors in myotonic dystrophy. Clin. Exp.
Immunol.,50, 572–578.
84. Fardaei,M., Rogers,M.T., Thorpe,H.M., Larkin,K., Hamshere,M.G.,
Harper,P.S. and Brook,J.D. (2002) Three proteins, MBNL, MBLL
and MBXL, co-localize in vivo with nuclear foci of expanded-repeat
transcripts in DM1 and DM2 cells. Hum. Mol. Genet.,11, 805–814.
85. Wagner,S.D., Struck,A.J., Gupta,R., Farnsworth,D.R.,
Mahady,A.E., Eichinger,K., Thornton,C.A., Wang,E.T. and
Berglund,J.A. (2016) Dose-dependent regulation of alternative
splicing by MBNL proteins reveals biomarkers for myotonic
dystrophy. PLoS Genet.,12, e1006316.
Downloaded from https://academic.oup.com/nar/article-abstract/46/12/6069/4996580
by guest
on 20 July 2018