Oncogene (2016) 35, 6366–6377

© 2016 Macmillan Publishers Limited, part of Springer Nature. All rights reserved 0950-9232/16
www.nature.com/onc

ORIGINAL ARTICLE
Tumour-processed osteopontin and lactadherin drive the
protumorigenic reprogramming of microglia and glioma
progression
A Ellert-Miklaszewska1, P Wisniewski1, M Kijewska1, P Gajdanowicz, D Pszczolkowska, P Przanowski, M Dabrowski, M Maleszewska
and B Kaminska

Tumour tissue is infiltrated by myeloid cells that are reprogrammed into alternatively activated/regenerative (M2) macrophages.
The contribution of major signalling pathways and their modulators/targets involved in the macrophage reprogramming is poorly
known. Glioblastoma (malignant brain tumour) attracts and reprograms brain-resident microglia and peripheral macrophages into
cells that increase invasion, angiogenesis and suppress antitumour immunity. Using a ‘function-first’ approach and glioma
secretome proteomics we identified osteopontin and lactadherin as proteins that cooperatively activate amoeboid transformation,
phagocytosis and motility of primary microglia cultures via integrins and FAK-Akt (focal adhesion kinase-Akt) signalling. A synthetic
peptide interfering with integrin ligands blocks glioma–microglia communication, functional activation and M2 gene expression.
We found that osteopontin/secreted phosphoprotein 1 (Spp1) produced by non-transformed cells acts as a proinflammatory factor
inducing inflammatory signalling and M1 genes, and counteracts the action of lactadherin. Using constructs encoding functional
mutants of osteopontin, we demonstrated sequential processing of Spp1 by thrombin and matrix metalloproteinase-3 and/or
-7 (MMP-3 and/or -7) in glioma cells, which generates a microglia-activating form devoid of the inflammatory activity, while
retaining the M2 reprogramming potential. A similar form of osteopontin is secreted by human glioma cells but not normal human
astrocytes. Knockdown of osteopontin or lactadherin in glioma cells reduces intracranial glioma growth, blocks amoeboid
transformation of myeloid cells and affects M2 reprogramming of microglia/macrophages. Our findings demonstrate how glioma
cells misuse macrophage-activating signals and redesign primarily proinflammatory signals towards their advantage to induce
M2 reprogramming of tumour-infiltrating brain macrophages.

Oncogene (2016) 35, 6366–6377; doi:10.1038/onc.2016.55; published online 4 April 2016

INTRODUCTION the reduction of glioma invasion in vitro and in animal models,

Tumour-associated macrophages as M1 macrophages exert
antitumoral functions, whereas alternatively activated, M2 macro-
phages exhibit protumoral functions by promoting invasion
and angiogenesis, immunosuppression and enhancing tumour
resistance to chemo- or immunotherapy.1,2 Recently established
guidelines defined a consensus collection of markers to describe
macrophage activation.3 The signals responsible for tumour-
associated macrophage accumulation and their polarization
remain poorly defined. Blocking studies implicated prostaglandin
E2 and interleukin-6 (IL-6) derived from cervical carcinoma in
M2 polarization.4 Chemokine (C–C motif) ligand 2 or IL-65 or
macrophage colony-stimulating factor peptides, produced by
pancreatic cancer cells, induced the M2 differentiation of
monocytes.6 Versican, acting via Toll-like receptor-2/6 complexes
in tumour-associated macrophages, was implicated in Lewis lung
carcinoma metastasis.7
Glioblastoma (GBM)—the most aggressive brain tumour—is
heavily infiltrated with myeloid cells, including brain-resident
microglia, peripheral macrophages and myeloid-derived suppres-
sor cells.8–10 Genetic or chemical ablation of microglia results in
demonstrating their significant role in glioma progression.11–13
Glioma cells promote polarization of microglia and macrophages
to the immunosuppressive, proinvasive phenotype in rodents
and humans.9,13–17 Glioma-associated microglia/macrophages
(GAMs) express the M2 markers: arginase 1 (Arg1), CD206,
chitinase 3-like 3 and MT1-MMP (a membrane type 1-matrix
metalloproteinase) and support tumour growth, invasion and
modulate responses to chemotherapy.11,18–20 GAMs produce
anti-inflammatory cytokines such as IL-10, transforming growth
factor-β1, chemokine CXCL (C–X–C motif ligand) 14,13 have
impaired cytotoxicity and favour the accumulation of T-suppressor
cells.18 Glioma cells support the proliferation of cultured microglia,
increase motility and phagocytosis, but fail to activate inflamma-
tory signalling pathways or to initiate the immune/inflammatory
gene expression.21
Several factors have been implicated in controlling the
recruitment of GAMs, including monocyte chemotactic protein-1
(Mcp1),22 Mcp3,23 Cxcrl1-Cxcr1,24 GDNF (glial cell line-derived
neurotrophic factor),25 soluble macrophage colony-stimulating
factors,26 transforming growth factor-β1, macrophage inhibitory
cytokine 118 and granulocyte–macrophage colony-stimulating

Laboratory of Molecular Neurobiology, Neurobiology Center, The Nencki Institute of Experimental Biology, Warsaw, Poland. Correspondence: Prof B Kaminska, Laboratory of
Molecular Neurobiology, The Nencki Institute of Experimental Biology, 3 Pasteur Street, Warsaw 02-093, Poland.
E-mail: bozenakk@nencki.gov.pl
1
These authors contributed equally to this work.
Received 3 September 2015; revised 17 January 2016; accepted 2 February 2016; published online 4 April 2016

factor.17 Their roles are still disputable and none of the considered
factors is capable of the multifaceted action required for the
M2 polarization.
Herein, a proteomic analysis of the glioma secretome combined
with functional analysis revealed that osteopontin (Spp1, secreted
phosphoprotein 1) and lactadherin (Mgfe8, milk fat globule-EGF
factor 8), highly overexpressed in glioma cells and human GBMs,
induce the M2 reprogramming of microglia via integrin signalling.
Tumour-specific processing of osteopontin by thrombin and MMPs
produced fragments inducing M2 microglia reprogramming.
RESULTS
Proteomic analysis of the glioma secretome identifies osteopontin
and lactadherin as microglia-activating factors
We aimed to identify microglia-activating factors in glioma
secretome using ‘a function-first’ approach, which combines
a proteomic analysis of the glioma secretome with a functional
assay. Glioma-conditioned medium (GCM) was fractionated
by high-performance liquid chromatography (HPLC) into 90
fractions. Each fraction was evaluated for the ability to induce
amoeboid transformation of microglia. Sixteen microglia-
activating fractions (scores 4–6) vs control (score 3) and several
inhibiting fractions (scores 1 and 2) were obtained (Figure 1 and
Supplementary Figure S1). Microglia-activating fractions were
subjected to tandem mass spectrometry. Unprocessed data files
containing tandem mass spectrometry spectra were submitted to
the Mascot search engine for database searching. The protein
sequence analysis resulted in the identification of many proteins
(Supplementary Table 1), with osteopontin (Spp1) and lactadherin
(Mfge8) at the top of this list (Figure 1b). Fractions containing
osteopontin and lactadherin stimulated microglial phagocytosis to
a similar extent as GCM (Figure 1c).
The level of Spp1a, Spp1c and Mfge8 mRNA was higher by
35-, 600- and 4-fold, respectively, in glioma cells compared with
that in non-transformed astrocytes. Secretion of osteopontin was
highly upregulated in C6 glioma cells compared with astrocytes
(Figure 1d). Quantification of SPP1 and MFGE-8 expression in
samples of pilocytic astrocytomas (human benign tumours with
favourable prognosis) and highly malignant GBM was performed.
The level of SPP1 mRNA was significantly elevated in GBM in
comparison with pilocytic astrocytomas (Figure 1e). The analysis
of REMBRANDT database shows that patients with higher SPP1
expression live significantly shorter than patients with low/
intermediate SPP1 expression (Figure 1f). The levels of MFGE-8
mRNA were similar in low- and high-grade gliomas but GBM
patients with ⩾ 2-fold higher MFGE-8 expression live shorter than
patients with low/intermediate expression (not shown).
Interference with osteopontin and lactadherin binding to integrins
abolished the microglial activation induced by GCM
Osteopontin and lactadherin bind to the integrins through the
RGD motif (arginine–glycine–aspartate), and αvβ3 and αvβ5
integrins are common receptors for both proteins.27 We designed
a synthetic 7-amino-acid RGD-containing peptide (RGD) as
a competitive inhibitor of both ligands; a scrambled sequence
peptide (SCR) served as a control. Morphological alterations of
microglia, visualized by the staining of F-actin with phalloidin-
tetramethylrhodamine B isothiocyanate, were visible upon treat-
ment with GCM alone or with RGD or SCR peptides. Preincubation
with 500 μM synthetic RGD peptide, but not with the SCR peptide,
prevented the GCM-induced amoeboid transformation of micro-
glia (Figure 2a). The RGD peptide completely blocked both the
GCM-induced phagocytosis of fluorescent beads (Figure 2b) and
microglial migration (Figure 2c). Silencing the integrin subunits αv,
β3 or both with specific small interfering RNAs (siRNAs) reduced
the number of phagocytosing microglial cells after GCM
© 2016 Macmillan Publishers Limited, part of Springer Nature.
Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6367
stimulation (Figure 2d). The efficacy of knockdown of the integrin
subunits was confirmed by flow cytometry and quantitative
polymerase chain reaction (qPCR) (Supplementary Figure S2).
GCM treatment increased the phosphorylation of focal adhesion
kinase (FAK), a mediator of integrin signalling. Pretreatment with
the RGD peptide abolished the activation of FAK as well
as downstream kinases, such as Akt and ERK, in microglial cells
(Figure 2e). The analysis of selected M1/2 genes showed that GCM
activates the expression of M2-specific genes21 coding for
the c-Myc, inhibitors of differentiation 1 and 3, Smad7, Arg1 and
MT1-MMP, but does not affect the expression of inos (encoding
a nitric oxide synthase). Pretreatment with the RGD peptide (but
not the SCR control) significantly reduced the GCM-induced
upregulation of the M2 phenotype genes (Figure 2f).
Osteopontin produced by non-transformed cells induces the
proinflammatory activation of microglia and counteracts the
action of lactadherin
To determine whether exogenous proteins could mimic the action
of GCM on microglia, we cloned the cDNAs coding for lactadherin
(mfge8) and the two forms of osteopontin (spp1a and spp1c)
(Figure 3a). Ectopic gene expression in murine NIH/3T3 fibroblasts
was verified by qPCR (Figure 3b). Fibroblasts provided the proper
post-translational modifications of osteopontin. Fibroblast-
conditioned media (3T3-CM) had no effect on microglia, but
CM from fibroblasts transfected with pSpp1a or pSpp1c strongly
stimulated microglial phagocytosis, although less efficiently than
GCM (Figure 3c). CM from fibroblasts transfected with an empty
vector or pMfge8 did not increase the microglial phagocytosis,
whereas CM from fibroblasts overexpressing pSpp1a or pSpp1c
alone or combination of pMfge8, pSpp1a and pSpp1c increased
phagocytosis to an extent similar to GCM and induced the
amoeboid transformation of microglia (Figures 3c and d).
CM from fibroblasts expressing Spp1a or/and Spp1c increased
active signal transducer and activator of transcription 1 (STAT1),
STAT3, STAT5 and phosphorylated IκB levels in the microglia,
which is an indication of the proinflammatory signalling. Spp1a
and Spp1c variants have comparable activity in stimulating
microglia (Figure 3e). Combination of Spp1a and Spp1c with
Mfge8 did not interfere with the induction of inflammatory
mediators; Mfge8 alone did not induce inflammatory signalling.
These data were corroborated by the results of gene expression.
CM from fibroblasts transfected with pSpp1a or/and pSpp1c
increased the expression of the M1 and M2 genes: arg1, smad7,
mt1-mmp, as well as inos and irf7. CM from fibroblasts expressing
pMfge8 induced c-myc and smad7 expression (Figure 3f). Coex-
pression of pSpp1a and pSpp1c with pMfge8 abolished c-myc
upregulation. These results show non-overlapping functions of
two ligands and demonstrate that osteopontin primarily stimu-
lated proinflammatory signalling and gene expression in micro-
glia, and counteracted the induction of c-myc expression by
lactadherin.
Osteopontin contains three functional and binding domains:
a 158GRGDS162 motif, which binds to RGD-containing integrins,
a thrombin cleavage site at Arg168/Ser169 and a CD44-binding
domain in the C terminus.28 The thrombin cleavage site binds to
the heparan sulphate on syndecan-4, which protects osteopontin
from cleavage by thrombin. Thrombin cleavage may alter
osteopontin functions by exposing an integrin-binding site and
releasing a chemotactic C terminus.28,29 To study how osteopontin
interacts with microglia, we generated a series of constructs:
with the CD44-binding domain deleted (pSpp1ΔC-term), the
integrin-binding RGD site mutated (pSpp1-RGD-RAE) and the
thrombin recognition site mutated (pSpp1-RSK-RAH) (Figure 4a).
Microglia exposed to CM from fibroblasts transfected with pSpp1-
RGD-RAE did not effectively induce inos, irf7 and mt1-mmp,
whereas the expression of arg1 and smad7 was barely affected
Oncogene (2016) 6366 – 6377
 Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6368

Figure 1. Glioma-derived osteopontin and lactadherin are found in microglia-activating fractions. (a) GCM was separated by HPLC into
90 fractions and each fraction was separately evaluated for the ability to induce amoeboid transformation of microglia. Morphological
transformation was scored from 1 to 6 independently by two researchers. (b) A list of top proteins identified in activating fractions subjected
to tandem mass spectrometry analysis. (c) Stimulation of microglial phagocytosis by fractions or GCM. Phagocytosis of fluorescent beads was
determined 90 min after incubation with the beads followed by washing and fixation. Total fluorescence was measured using a microplate
reader (mean ± s.e.m., n = 2 in triplicates); P-values calculated with Student’s t-test. (d) Expression of osteopontin and lactadherin in C6 glioma
cells and primary cortical astrocyte cultures was determined by qPCR and, in the case of osteopontin, by ELISA (mean ± s.e.m., n = 3); P-values
calculated with Student’s t-test. (e) Assessment of the SPP1 expression in 23 GBMs, 17 pilocytic astrocytomas (PA) and normal brains (a dashed
line) revealed upregulation of SPP1 levels in GBM. (f) Kaplan–Meier survival plot derived from REMBRANDT database illustrates negative
correlation between SPP1 expression and survival time; upregulated vs downregulated P = 3.24E − 5, upregulated vs intermediate P = 2.13E − 7,
downregulated vs intermediate P = 0.02.

(Figure 4b). Amoeboid transformation and phagocytic activity
were reduced in microglia exposed to CM from fibroblasts
transfected with pSpp1 RGD-4RAE (Figures 4c and d). This shows
that the integrin-binding RGD site of osteopontin is crucial for the
induction of morphological alterations, phagocytosis and inflam-
matory gene expression.
Stimulation of microglia with CM from fibroblasts expressing a
wild type Spp1 significantly induced the production of nitric oxide
and proinflammatory cytokines and chemokines, such as tumour
necrosis factor-α, Ccl2/Mcp1 and Ccl3/Mip1α (macrophage inflam-
matory protein-1α) to a similar extent as lipopolysaccharide. These
effects were reduced when microglia were stimulated with CM

Oncogene (2016) 6366 – 6377 © 2016 Macmillan Publishers Limited, part of Springer Nature.
 Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6369

Figure 2. A synthetic 7-amino-acid (aa) RGD peptide blocks amoeboid transformation, cell motility and phagocytosis induced by glioma
in microglia. (a) Microglia were preincubated with 500 μM of synthetic 7-aa RGD (RGD) or a scrambled sequence (SCR) peptide 30 min before
GCM stimulation. Phalloidin-tetramethylrhodamine B isothiocyanate (TRIC) staining shows that the RGD peptide prevented F-actin
reorganization in microglia (indicated with arrows). (b) RGD, but not the SCR peptide, completely reduced the GCM-induced phagocytosis
determined 24 h post-treatment. Percentages of cells with low ( o2 beads), medium (⩾2 to o10) or high (⩾10) phagocytic activity were
counted in 20 randomly selected fields (mean ± s.e.m., n = 3); Student’s t-test. (c) Preincubation of microglia with 500 μM RGD, but not with the
SCR peptide, reduced the GCM-stimulated motility of microglia. Microglial cells (1.5 × 106) were cultured to confluence and gently scratched.
After washing, cells were exposed for 3 h to GCM alone or with 500 μM 7-aa synthetic RGD peptide or a control scrambled peptide. Migrating
cells were visualized by phase-contrast microscopy. (d) Silencing of αv, β3 or both integrin subunits in microglial cultures with siRNA reduced
GCM-stimulated phagocytosis (less cells with ⩾ 10 beads per cell). Microglial cells (1x105) were transfected with 25 nM ON-TARGET siRNA and
DharmaFECT3. After 48 h, the transfection medium was replaced by GCM, and after 24 h, the phagocytic properties of microglia were
determined. Microglia were incubated with 2-μm fluorescent latex beads for 90 min. The cells were washed, fixed with 2% paraformaldehyde
(PFA) and stained with FITC-conjugated isolectin B4. The percentage of cells with low ( o2 beads per cell), medium (⩾2 to o10) or high (⩾10)
phagocytic activity was determined, mean ± s.e.m., n = 3; one-way ANOVA followed by post hoc Newman–Keuls test. (e) Levels of p-FAK, p-Akt
and p-ERK were upregulated by GCM in microglia, and this effect was prevented by incubation with the RGD peptide. Protein extraction and
western blot were performed as described with specific antibodies recognizing the p-AKT, p-ERK1/2 and p-FAK and t-AKT, t-ERK1/2 and t-FAK.
(f) Pretreatment with the RGD peptide affects GCM-induced gene expression determined by qPCR 6 h following treatment. The results
are presented as fold change vs untreated microglia in log 2 scale (mean ± s.e.m.; n = 5); one-way ANOVA followed by post hoc Newman–Keuls
test. (g) Graphical representation of signalling pathways underlying glioma–microglia communication.

from fibroblasts co-transfected with pSpp1 and pMfge8, and did
not occur when microglia were stimulated with CM from
fibroblasts co-transfected with pSpp1ΔC-term and pMfge8. The
ΔC-terminal Spp1 construct is depleted of C-terminal fragment
and mimics the N-terminal, thrombin-cleaved Spp1. Protumori-
genic effects of microglia exposed to various forms of Spp1 with
Mfge8 were determined using a Matrigel assay. Media collected
from microglia preactivated with CM from fibroblasts producing
Mfge8 with either Spp1 or Spp1ΔC-term significantly stimulated
glioma cell invasion (Supplementary Figure S3). This confirms
the proinflammatory action of a full form of Spp1 produced by
non-transformed cells and proinvasive support by microglia
activated by combination of Mfge8 with tumour-processed Spp1
(ΔC-terminal Spp1).

© 2016 Macmillan Publishers Limited, part of Springer Nature. Oncogene (2016) 6366 – 6377
 Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6370
Osteopontin is differentially processed by thrombin and
metalloproteinases in glioma cells
Distinct action of ostepontin produced by transformed and
non-transformed cells suggested different processing of this
protein in glioma cells. Using antibodies specific for C and N
terminus of the protein, we found the N-terminal Spp1 fragment
(22 kDa) in GCM but not in CM from fibroblasts (Figure 5a),
Oncogene (2016) 6366 – 6377
suggesting that proteolytic cleavage is specific for glioma. The
short, N-terminal SPP1 fragment was also detected in CM from
four human GBM cell lines (including patient-derived WG4 cells)
but not in normal human astrocytes (Figure 5b).
Osteopontin can be cleaved by MMPs (i.e. MMP-3 and MMP-7 in
human prostate and liver cancer cells)30 or thrombin.28,29 The
analysis of the C6 glioma transcriptome indicated mmp-3, mmp-7
© 2016 Macmillan Publishers Limited, part of Springer Nature.

and mmp-10 expression. To verify if MMPs participate in
osteopontin processing, we silenced the expression of mmp-3,
mmp-7 or mmp-10 in C6 glioma cells using specific siRNAs.
Knockdown of mmp-3 or mmp-7 in glioma cells reduced the
N-terminal Spp1 fragment levels (Figure 5c), whereas a control
siRNA or mmp-10 siRNA had no effect. CM from glioma cells with
reduced mmp-3 and mmp-7 expression did not induce the
expression of M2 phenotype genes (arg1, c-myc and smad7) in
microglia (Figure 5d).
To evaluate a role of Spp1 cleavage by thrombin in microglia
activation, we knockdown osteopontin expression in glioma cells and
overexpressed a wild-type Spp1 or a pSpp1-RSK-RAH construct. The
Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6371
N-terminal Spp1 fragment was absent if C6 cells were transfected
with a pSpp1-RSK-RAH construct (Figure 5e), and, consistently, CM
from these cells stimulated the expression of M1 genes (inos and irf7)
in the microglia (Figure 5f). This suggests that cleavage of osteopontin
by thrombin is a prerequisite for subsequent cleavage by MMPs and
ablation of its proinflammatory activity.
Knockdown of osteopontin or lactadherin in glioma cells blocks
tumour growth and M2 reprogramming
We developed cell lines stably expressing control (shNeg) or gene-
specific short hairpin RNAs (shRNAs) (shSpp1, shMfge8). The efficacy

Figure 4. Activation of microglia by osteopontin depends on integrin binding via the RGD motif. (a) Graphical summary of osteopontin
mutants used in the study. (b) Microglia were exposed for 6 h to CM from fibroblasts transfected with pGFP, wild-type or mutated osteopontin
constructs. Gene expression was analysed by qPCR 6 h after treatment; fold changes vs untreated microglia (mean ± s.e.m.; n = 5); one-way
ANOVA followed by post hoc Newman–Keuls test. (c) Representative images of morphological alterations of microglia 24 h after exposure
to conditioned media from control fibroblasts (transfected with an empty vector) or fibroblasts expressing wild-type (Spp1) or mutated (Spp1
RGD-4RAE) osteopontin. Cells were stained with FITC-conjugated isolectin B4, and cell nuclei were counterstained with DAPI.
(d) Quantification of phagocytosis in microglia exposed for 24 h to CM from fibroblasts transfected with pGFP, wild-type (Spp1) or mutated
osteopontin constructs (mean ± s.e.m.; n = 3); one-way ANOVA followed by post hoc Newman–Keuls test.

Figure 3. Osteopontin produced by non-transformed cells induces the proinflammatory activation of microglia and counteracts the action
of lactadherin. (a) NIH/3T3 fibroblasts were transiently transfected with the control GFP plasmid, plasmids coding for lactadherin (pMfge8)
or/and two isoforms of osteopontin (pSpp1a, pSpp1c). (b) Ectopic gene expression in NIH/3T3 cells was confirmed by qPCR 24 h after
transfection (representative of n = 3). (c) Quantification of phagocytosis in untreated microglia (Ctrl) or microglia exposed to CM from glioma
(C6-cm), control fibroblasts (3T3-cm) or fibroblasts transfected with plasmids. Microglia grown at a density of 1 × 106 cells were exposed for
24 h to different CM. Phagocytosis of fluorescent beads was determined 90 min after incubation with the beads followed by washing and
fixation. Total fluorescence was measured using a microplate reader (mean ± s.e.m.; n = 5; one-way ANOVA followed by post hoc Newman–
Keuls test). (d) Representative images of morphological alterations of microglia (amoeboid cells indicated with arrows) 24 h after exposure to
CM from control fibroblasts or fibroblasts transfected with plasmids. Cells were stained with FITC-conjugated isolectin B4 and counterstained
with DAPI. (e) Representative immunoblot shows activation of NF-κB and STAT signalling in microglia treated with CM from fibroblasts
transfected with respective plasmids. A control GFP plasmid (empty vector) or plasmids coding for lactadherin (Mfge8) alone or in
combination with plasmids coding for the two forms of osteopontin (Spp1a and Spp1c) were transfected into murine NIH/3T3 fibroblasts.
Conditioned media (cm) were collected from transfected cells after 24 h and added to primary microglia cultures. Microglial cultures
stimulated with 100 ng/ml lipopolysaccharide (LPS) or GCM serve as additional controls. Total proteins were collected after 90 min as
described. Western blot analyses were performed with specific antibodies recognizing the p-STAT1, p-STAT3, p-STAT5, and the p- and t-IκB;
β-actin was used as a loading control (n = 3). (f) CM from fibroblasts transfected with pSpp1 triggers the upregulation of genes characteristic
for the M2 (arg1, smad7, mt1-mmp) and the M1 (inos, irf7) phenotype in the microglia. Gene expression was measured by qPCR 6 h after the
treatment. The results are presented as fold change vs untreated microglia (mean ± s.e.m.; n = 5). P-values were calculated for microglia
treated with CM from control 3T3-cm vs transfected fibroblasts; one-way ANOVA followed by post hoc Newman–Keuls test.

© 2016 Macmillan Publishers Limited, part of Springer Nature. Oncogene (2016) 6366 – 6377
 Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6372

Figure 5. Osteopontin is differentially processed in glioma cells. (a) The N-terminal fragment of Spp1 (22 kDa) is detected by western blot
analysis in CM from glioma cells (but not fibroblasts) transfected with wild-type Spp1. The upper panel shows the full length of osteopontin
detected in C6 cells and fibroblasts (with the C-terminus-specific antibody). The detection of laminin served as a loading control. Data shown
are from a single representative experiment out of three repeats. (b) A representative immunoblot shows the N-terminal fragment of Spp1
(22 kDa) detected by western blot analysis in CM from human glioma cells (but normal human astrocytes). (c) Knockdown of MMP-3 and
MMP-7 (but not MMP-10) reduces the expression of M2 marker genes upregulated in the microglia exposed to GCM. Glioma cells were
transfected with 100 pmol specific to mmp-3, mmp-7 and mmp-10 siRNA, and after 48 h, the transfection medium was replaced by GCM
and collected after additional 24 h. Real‐time PCR analyses of M1/M2 marker expression in primary microglial cultures were carried out 6 h
after GCM or control MGMG treatment; fold change vs untreated microglia, mean ± s.e.m.; n = 3; one-way ANOVA followed by post hoc
Newman–Keuls test. (d) Knockdown of MMP-3 or MMP-7 (but not MMP-10) with siRNA reduces the levels of the N terminus of osteopontin in
glioma CM. To knockdown selected genes in C6 glioma cells, 100 pmol siRNA specific to mmp-3, mmp-7 and mmp-10 were delivered using
Amaxa electroporation. After 48 h, the transfection medium was replaced by GCM and collected after additional 24 h. The N-terminal
fragment of endogenous Spp1 (22 kDa) was detected by western blot analysis in CM from transfected glioma cells. (e) The N-terminal
fragment of Spp1 is not detected in CM (conditioned for 48 h) from osteopontin-depleted glioma cells transfected with Spp1-RSK-RAH, in
contrast to WT Spp1. (f) Overexpression of pSpp1-RSK-RAH in osteopontin-depleted glioma cells results in the induction of M1 genes in the
microglia. C6 glioma cells were transfected with control or Spp1 targeting siRNA 48 h before subsequent transfection with the plasmids
coding for wild-type (WT) or mutated Spp1; one-way ANOVA followed by post hoc Newman–Keuls test, fold change vs untreated microglia,
mean ± s.e.m.; n = 3.

of gene knockdown was verified by qPCR and western blotting
(Figure 6a). The proliferation of shNeg, shSpp1.1, shSpp1.2 (different
clones) or shMfge8 glioma cells was similar to the proliferation of
parental C6 glioma cells, as demonstrated using carboxyfluorescein
succinimidyl ester labelling followed by flow cytometry and
bromodeoxyuridine incorporation assay (Figure 6b).
Histological images show tumour reduction in the brains of
animals implanted with shMfge8 or shSpp1 glioma cells by 70 and
80–88%, respectively, compared with parental C6 or shNeg glioma
cells (Figures 6c and d). The number of Iba1+ cells in shNeg, shSpp1
and shMgfe8 gliomas was comparable; however, Iba1+ cells
infiltrating shNeg gliomas were more amoeboid than those in
shSpp1 gliomas (Figure 7a). The number of amoeboid Iba1+ cells
was reduced in shSpp1 gliomas in comparison with controls
(Figure 7b). Staining for Arg1 (a M2 phenotype marker) was very
abundant in shNeg brains and showed a significant reduction in the
number of Arg1+ cells in shSpp1 and shMgfe8 gliomas (Figures 7c
and d). We did not detect Arg1+ cells in contralateral hemispheres.
Double staining for Arg1 and Iba1 showed colocalisation of both
markers in microglia/macrophages (Figure 7e). These results show
that glioma-derived osteopontin and lactadherin modulate M2
reprogramming of GAMs and support tumour growth.

Oncogene (2016) 6366 – 6377 © 2016 Macmillan Publishers Limited, part of Springer Nature.
 Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6373

Figure 6. Knockdown of osteopontin or lactadherin in glioma cells inhibits tumour growth. (a) Clones of C6 glioma cells that stably express
control (shNeg), lactadherin (shMfge8) or osteopontin targeting (shSpp1) shRNA were developed. Silencing of gene expression was confirmed
by qPCR (mean ± s.e.m.; n = 2) and immunoblotting. (b) Knockdown of osteopontin or lactadherin in glioma cells does not affect cell
proliferation, as evidenced with CSFE staining, flow cytometry and the bromodeoxyuridine (BrdU) incorporation test (mean ± s.e.m.; n = 3);
one-way ANOVA followed by post hoc Newman–Keuls test. (c) Representative images of toluidine blue-stained sections of rat brains show
a reduction in the size of shMfge8 or shSpp1 gliomas. Parental C6 or modified glioma cells (5 × 104 cells/in 2.5 μl of Dulbecco's modified
Eagle's medium (DMEM)) were implanted into the right striatum of 250 g Wistar rats. At day 15th after implantation, brains were perfused,
sectioned (20 μm), stained with toluidine blue and analysed (original magnification 10 × ). (d) Quantification of tumour volumes reveals that
gliomas depleted of lactadherin (n = 10) or osteopontin (n = 10) formed smaller tumours than parental cells (n = 6) and cells expressing control
shRNA (n = 6). Implanted shNeg cells developed tumours of similar sizes as parental C6 cells. Toluidine blue-stained sections of rat brains were
scanned and images were acquired using a Leica DM4000B microscope. Tumour areas were measured using ImageJ in an every fourth brain
slice, and tumour volumes were calculated; the mean ± s.e.m., P values were calculated using the Mann–Whitney U-test.

DISCUSSION Akt signalling. It results in enhanced cell motility, phagocytosis

Glioma-derived osteopontin and lactadherin induce the M2
reprogramming of microglia
Studies of cytokine cocktails or proteomics of tumour secretome
indicated different molecules as mediators of tumour–macro-
phage interactions.5–7,31 We report the successful identification of
osteopontin and lactadherin as microglia-activating factors, which
induce microglial amoeboid transformation, migration, phagocy-
tosis and M2 gene expression, stimulate integrin signalling and
have a protumorigenic role, as knockdown of either protein blocks
glioma growth and M2 reprogramming in vivo. Knockdown of
osteopontin or lactadherin in glioma cells reduced tumour
volume. Interestingly, it did not affect the accumulation of
microglia/macrophages, but blocked some attributes of the
M2 polarization such as amoeboid transformation and the
expression of Arg1. We have demonstrated that glioma-derived
granulocyte–macrophage colony-stimulating factor is responsible
for the accumulation of GAMs in murine gliomas17 and human
gliomas in nude mice (unpublished).
We show a novel role for tumour-derived integrin ligands in
microglia reprogramming. Osteopontin via interaction with
integrins (i.e., αvβ3, αvβ5, αvβ6 and αvβ1) can modulate cell
adhesion and migration,32 whereas lactadherin interacts with
αvβ3 and αvβ5 integrins.33 We demonstrate that GCM activates
integrin receptors on the microglia, which induce FAK and PI-3 K/
and distinct genomic responses without the induction of
inflammatory gene expression. Blockade of integrin signalling by
the RGD peptide inhibits GCM-induced microglial functions and
induction of M2 gene expression (Figure 2). Osteopontin
cooperates with lactadherin to induce M2 reprogramming of the
microglia, as only lactadherin stimulates in microglia the expres-
sion of the transcription factor c-Myc. c-Myc regulates the cell
cycle, transcription, differentiation and processes that may be
a prerequisite for M2 programming. Cytokines (IL-4, IL-13, IL-10
and transforming growth factor-β) that induce M2 reprogramming
induced c-Myc and several M2 genes in human blood
monocytes.34
Osteopontin is secreted by immune cells under inflammatory
conditions, and acts primarily as a proinflammatory molecule and
chemoattractant.35 It regulates the recruitment of macrophages
and T cells, and the production of inflammatory mediators by
immune cells. The N-terminal, thrombin-cleaved fragment of SPP1,
through its binding to the α9β1 and α4β1 integrins, regulates the
release of hematopoietic stem cells from bone marrow niche.36
Lactadherin (produced by macrophages) stimulates phagocytosis
of apoptotic cells without inducing inflammation.37
We demonstrate that osteopontin produced by normal
fibroblasts induces inflammatory responses and competes with
lactadherin (Figure 3). Osteopontin lacking a CD44-binding site
exerted a similar action on microglia as a wild-type protein. The

© 2016 Macmillan Publishers Limited, part of Springer Nature. Oncogene (2016) 6366 – 6377
 Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6374

Figure 7. Knockdown of osteopontin or lactadherin in C6 cells affects M2 polarization but not the accumulation of microglia/macrophages in
gliomas. (a and b) Iba1 staining and assessment of the number and morphology of Iba1+ cells (note rare, ramified Iba1+ cells in the
contralateral hemisphere and Iba1+ amoeboid cells in glioma-bearing hemisphere). The brains were perfused, sectioned (12 μm) and
incubated with the anti-Iba1 antibody. Representative sections of the glioma-bearing brain and a contralateral hemisphere stained with an
antibody specific for Iba1 are shown. Total and amoeboid Iba1+ cells were counted in five randomly selected fields (n = 5 rats per group),
P-values were calculated using the Mann–Whitney U-test. (c and d) Arginase 1 staining shows accumulation of Arg1+ cells in the shNeg
gliomas and significant reduction of the number of Arg1+ in shSpp1 and shMfge8 gliomas. There is no Arg1+ cells in the contralateral
hemisphere. For the detection of Arg1+ cells (red), sections were incubated with anti-Arg1 antibody and cell nuclei were visualized with DAPI
staining (blue). Arg1+ cells were counted in five randomly selected fields (n = 5 rats per group). (e) Colocalization of Arg1+ and Iba1+ cells in
shNeg gliomas. Representative sections of the glioma-bearing brains stained with antibodies specific for Iba1 and Arg1, and costained with
DAPI. Images were taken using a Leica DM4000B microscope.

mutation in the RGD site (Spp1-RGD-RAE) abolished the ability
of Spp1 to induce proinflammatory genes (inos, irf7) and microglial
phagocytosis, but did not affect the M2 gene expression (arg1 and
smad7) (Figure 4). The failure to inhibit arg1 expression by the 7-
amino-acid RGD peptide indicates that arg1 expression was not a
directly regulated by RGD-integrin (αvβ3 integrin) pathway in
in vitro GCM-stimulated microglia. However, CM from fibroblasts
overexpressing pSpp1a/c induced the arg1 expression along with
several functional processes. Moreover, knockdown of Spp1 in
glioma cells significantly reduced the number of arg1+ microglia
in vivo, suggesting regulation of Arg1 expression by Spp1 but not
necessarily via RGD-integrin (αvβ3 integrin) pathway. Spp1
processing by thrombin and MMPs may generate not only
N-terminal but also short C-terminal fragments with chemotactic
action,28,29 which may act via different integrins.
The mutations in the RGD site of Spp1 produced by fibroblasts
affected the expression of M1 genes and mt1-mmp, phagocytosis and
morphological transformation. The thrombin cleavage site mutations
of Spp1 produced by fibroblasts had no effect on M1/M2 gene
expression. These results show that osteopontin produced by non-
transformed cells activates primarily a proinflammatory response in
the microglia. Glioma cells misuse this microglia-activating signal and
redesign for their advantage to induce M2 reprogramming.
Processing of osteopontin by thrombin and MMPs in glioma cells
generates the short osteopontin fragments that induce M2
reprogramming.
Differential processing of osteopontin in glioma cells provides
a plausible explanation of the contradictory action of tumour- and
non-tumour cell-derived osteopontin. The secreted, N-terminal
Spp1 fragment is produced in glioma cells by the sequential
action of thrombin and MMP-3 and 7. Overexpression of a mutant
with a non-functional thrombin cleavage site (Spp1-RSK-RAH) in
endogenous osteopontin-depleted glioma cells reduced the accu-
mulation of an N-terminal fragment and the expression of M2 genes
(Figure 5e). Thrombin, MMP-3 and MMP-7 can cleave osteopontin at
sites close to or within the integrin-binding sites.38

Oncogene (2016) 6366 – 6377 © 2016 Macmillan Publishers Limited, part of Springer Nature.

The SPP1 mRNA is upregulated in human GBM in comparison
with benign gliomas. SPP1 expression was found highly elevated
in tumour tissues and sera from GBM patients, and inversely
correlated with patient survival.38 We detected the N-terminal
SPP1 fragment in four human GBM cell lines but not in
human astrocytes (Figure 5b). The thrombin-cleaved fragments
of osteopontin have been found in malignant gliomas and
promoted motility of U87MG glioma cells, altered the expression
of cell cycle genes and conferred resistance to apoptosis.28
Pleiotropic effects of osteopontin on tumour progression have
been attributed to direct effects on cell viability, motility and
invasion mediated by binding to CD44 or integrins on cancer or
endothelial cells.32,39 The requirement for the upregulation of
osteopontin and lactadherin by immune cells may be regarded as
a safeguard that allows for tissue repair and remodelling. Glioma
cells mimic those signals to induce tissue remodelling without
inducing inflammation. Combined action of osteopontin and
lactadherin resembles effects of periostin released by glioma stem
cells and recruiting M2 microglia/macrophages.40 However,
periostin seems to attract not brain-resident microglia, but
mainly monocyte-derived macrophages from peripheral
blood,40 which migrate to a glioma tissue as a secondary wave
of myeloid cells.13 Periostin recruits tumour-infiltrating myeloid
cells through the integrin αvβ3 as blocking this signalling by an
RGD peptide-inhibited cell recruitment.40 Such findings high-
light the possibility of improving GBM treatment by targeting
glioma–microglia interactions.
MATERIALS AND METHODS
Cell culture
Microglial primary cultures were prepared from the cerebral cortices of 1-
day-old Wistar rat pups as described.41 Cell viability was determined by
trypan blue exclusion. Cells (2–3x105 cells/cm2) were plated onto culture
dishes for suspension cells and adherent cells (496% positive for isolectin B4)
were incubated for 48 h to quiet microglia. Astrocyte cultures (496%
positive for glial fibrillary acidic protein) were cultured in Dulbecco's modified
Eagle's medium with 10% fetal bovine serum, glutamax and antibiotics. NIH/
3T3 fibroblasts and rat C6 glioma cells obtained from the American Type
Culture Collection were cultured for 24 h after seeding 1x106 cells. Human
LN18, U87 and LN229 glioma cell lines were from American Type Culture
Collection; WG4 was patient-derived GBM cell line. Normal human astrocytes
(NHA; Lonza, Cologne, Germany) were cultured in Clonetics media and
reagents (Lonza, Walkersville, MD, USA). CM from glioma cultures (GCM)
grown in 8 ml of high-glucose Dulbecco's modified Eagle's medium with 10%
fetal bovine serum and glutamax were collected and centrifuged at 300 g for
10 min. All CMs were tested for mycoplasma.
HPLC fractionation, detection of activating fractions, mass
spectrometry and protein identification
GCM was dialyzed overnight against Tris-HCl buffer and lyophilized. The
obtained preparation was subjected to HPLC fractionation using an anion
exchanger Q (Shodex IEC QA-825 PHM gel, Showa Denko K.K., Tokyo,
Japan), and 90 fractions were collected. Each fraction (diluted 1:10 in
culture medium) was tested for the ability to induce the morphological
transformation of microglia. Morphology was evaluated 24 h post-
treatment and scored from 1 to 6 by two researchers.
Peptides from the HPLC fractions were analysed by Nano-Spray liquid
chromatography tandem mass spectrometry. Unprocessed data files
containing tandem mass spectrometry spectra were submitted to the
Mascot search engine (Matrix Science Ltd, London, UK) for database
searching using the Mascot daemon. The sequence of each peptide was
compared against the reference rat and bovine protein sequence
databases (International Protein Index) using the WU-Blast 2.0 software
package (licensed from Washington University, St Louis, MO, USA) run with
the following options: -matrix blosum80 -E 1 -B 1 -topcomboN 1 -W 2 -Q 12
-R 12 -mformat 2. These options guarantee a single, best-fitting protein
match for each peptide, while taking advantage of the precomputed tables
of the blast parameters (blast e-value). For all peptides with hits in the
databases, we calculated the log ratio of the bovine e-value to the
rat e-value. To identify proteins that can be reliably traced to the rat origin,
© 2016 Macmillan Publishers Limited, part of Springer Nature.
Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6375
we chose the proteins with blast hits to two or more unique peptides,
with probability scores that met or exceeded the threshold (Po 0.05) for
statistical significance and with positive log ratio for each sequence.
Functional assays and treatments
To study morphological alterations, cells were fixed with 2% paraformal-
dehyde and incubated with phalloidin-tetramethylrhodamine B isothio-
cyanate for 30 min at room temperature or stained with fluorescein
isothiocyanate (FITC)-conjugated isolectin B4 overnight at 4 °C. Cells were
washed with phosphate-buffered saline, and the cell nuclei were stained
with 10 μg/ml DAPI (4′,6-diamidino-2-phenylindole; Sigma-Aldrich,
St Louis, MO, USA) for 10 min. Morphological alterations were monitored
by fluorescent microscopy with excitation at 450–490 nm. For the scratch
assay, 1.5x106 cells were cultured to confluence and gently scratched. After
washing, cells were exposed to various conditions for 3 h. Migrating cells
were visualized by phase-contrast microscopy. To determine phagocytic
properties, microglia grown at a density of 1x106 cells were exposed
for 24 h to different conditions. Phagocytosis was determined by
incubating cells with 2 μm fluorescent latex beads (Sigma-Aldrich)
for 90 min. The cells were washed two times with phosphate-buffered
saline, fixed with 2% paraformaldehyde and stained with FITC-isolectin B4.
Alternatively, 1.5x105 cells were seeded, incubated with the fluorescent
beads, washed and fixated, after which total fluorescence was measured
using a microplate reader. In some experiments, microglial cultures were
treated with 500 μM 7-amino-acid synthetic RGD peptide, a control
scrambled peptide (GeneCust, Dudelange, Luxemburg) and the broad-
spectrum MMP inhibitor GM6001 (Sigma-Aldrich).
GBM and PA samples
Frozen glioma biopsies were obtained from the Brain Tumour Tissue Bank
(London Health Sciences Centre, London, ON, Canada). Additional pilocytic
astrocytoma samples were obtained under protocol no. 14/KBE/2012,
approved by the Committee of Bioethics at the Children's Memorial Health
Institute (Warsaw, Poland). Informed consent was obtained from
all subjects. The reference brain RNA was a mixture from 23 normal
male/female brains (Ambion FirstChoice Human Brain Reference RNA,
Thermo Fisher Scientific, Waltham, MA, USA).
Gene silencing with siRNA
For integrin knockdown, 1x105 microglial cells were transfected with 25 nM
siRNA (ON-TARGET siRNA; Dharmacon, Lafayette, CO, USA) using
DharmaFECT3 or Viromir blue (Lipocalyx, Halle (Saale), Germany) transfec-
tion reagents. After 48 h, the transfection medium was replaced by GCM,
and the phagocytic properties of microglia were determined. To estimate
the silencing efficiency, cells were collected by trypsinization, and cell
surface integrin expression was analysed using antibodies to αv or β3 (BD
Biosciences, Franklin Lakes, NJ, USA) and anti-mouse-Alexa Fluor 647
antibody (Invitrogen, Carlsbad, CA, USA) with a FACSCalibur flow
cytometer (Becton Dickinson, Franklin Lakes, NJ, USA) or by qPCR with
specific primers. To knockdown selected genes in C6 glioma cells,
100 pmol siRNA specific to mmp-3, mmp-7, mmp-10 or Spp1 (Flexitube;
Qiagen, Hilden, Germany) were delivered using Amaxa (Lonza)
electroporation.
Protein isolation, electrophoresis and detection
CM from fibroblasts and C6 glioma cells (150 μl) were precipitated using
methanol/chloroform and resuspended in buffer containing phosphatase
and protease inhibitors. CM from human glioma cells or non-transformed
astrocytes were centrifuged at 200 g for 5 min, and then 10 times
concentrated using centrifugation at 14000 g for 5 min in Vivaspin 500
concentrators with 10000 Da MWCO (Sigma-Aldrich) and mixed with
4x Laemmli buffer for sodium dodecyl sulphate–polyacrylamide gel
electrophoresis. Protein extraction and western blot were performed
as described12 with specific antibodies (Cell Signalling, Beverly, MA, USA, all
1:1000); recognizing the phosphorylated (p) and total (t) proteins: p-ERK1/2
(no. 9101), t-ERK1/2 (no. 9102), p-STAT1 Tyr701 (no. 9171, p-STAT3 TYR705
(no. 9131), p-AKT SER473 (no. 9271), p-FAK Tyr925 (no. 3284), t-FAK
(no. 3285), p-IκBα (no. 2859) and t-IκBα (no. 9242). Other antibodies
included an antibody recognizing the first 0–17 amino-acid of osteopontin
(1:100; no. ABIN18083; Sino Biological, Beijing, China) and MAB153P
antibody specific to the N-terminal region of osteopontin (1:75; Maine
Biotechnology, Portland, ME, USA); anti-laminin antibody (no. Ab11575;
Oncogene (2016) 6366 – 6377
 Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6376
Abcam, Cambridge, UK) and horseradish peroxidase-conjugated anti-
rabbit IgG (1:2000; Cell Signalling). Immunocomplexes were visualized
using ECL (GE Healthcare, Little Chalfont, UK). The membranes were
stripped and reprobed with horseradish peroxidase-conjugated anti-β-
actin antibody (1:10 000; no. A3854; Sigma-Aldrich). The molecular weight
of proteins was estimated with prestained protein markers (Sigma-Aldrich).
Generation of shRNA-expressing vectors and stably transfected C6
glioma clones
To silence the expression of osteopontin and lactadherin, shRNA-
expressing vectors were constructed using pSilencer2.0-U6hygro (Ambion).
Two pairs of complementary oligonucleotides encoding spp1shRNA and
mfge8 shRNA with BamH1 and HindIII overhangs were designed
(Supplementary Information). Annealed oligonucleotides were ligated
with pSilencer2.0-U6hygro (Ambion). The sequences of resulting plasmids
were verified by sequencing. The pSilencer-spp1, pSilencer-mfg8 and
pSilencer2.0-U6 Negative Control (Ambion) were purified using the Qiagen
Plasmid Maxi Kit (Qiagen), and 3 μg was transfected into C6 cells using
Amaxa nucleofection. Hygromycin B (500 μg/ml) was added 24 h after
transfection. Resistant clones were picked after 2 weeks, expanded and
analysed for the expression of appropriate mRNA by quantitative real-
time–PCR. Osteopontin production was determined by ELISA (Enzo Life
Sciences, Farmingdale, NY, USA).
Cloning of wild-type and mutated spp1 and mfge8
The coding sequences of spp1a, spp1c and mfge8 were amplified from a rat
glioma cDNA template using the specific primers (Supplementary Table S2)
with Taq DNA polymerase (Applied Biosystems, Foster City, CA, USA)
and subcloned into the HindIII and NotI sites of the pEGFP-N1
plasmid (Clontech, Mountain View, CA, USA). Mutations in spp1 ORF
were introduced using two-step PCR. For Spp1-RGD-RAE mutant
following primer sets carrying mutations and restriction sites for cloning
were used: RnSpp1_SalI_F (5′-GCAGTCGACGGATGAGACTGGCAGTGGT
T-3′)+Spp1_RGD-RAE_R (5′-GCTTTCAGCTCGGCCGTCAGGGACATCG-3′) and
Spp1_RGD-RAE_F (5′-CGGCCGAGCTGAAAGCTTGGCTTACG-3′)+Spp1 STOP
NotIR (5′-GTGGCGGCCGCCCTTAATTGACCTCAGA-3′). PCR products were
excised from the agarose gel (1%), purified and used as templates for
second PCR with flanking primers: RnSpp1_SalI_F+Spp1 STOP NotIR.
Product was purified and cloned into pEGFP-N1 plasmid using SalI/NotI
restriction enzymes. For Spp1-RSK-RAH mutant following primer sets
carrying mutations and restriction sites for cloning were used: RnSpp1_Sa-
lI_F+Spp1_RSK-RAH_R (5′-TGGAATGTGCCCTCAGTCCGTAAGCC-3′) and
Spp1_RSK-RAH_F (5′-CTGAGGGCACATTCCAGGAGTTTCCC-3′)+Spp1 STOP
NotIR. PCR products were processed as described for Spp1-RGD-RAE
mutant. The resulting plasmids were verified by sequencing.
Real-time PCR
Total RNA was isolated from various cells and after exposures using
an RNeasy Kit (Qiagen) and 2 μg cDNA was used as a template.
Amplifications were performed in duplicates in a 20- μl reaction volume
containing SYBR PCR MasterMix (Life Technologies, Carlsbad, CA, USA) and
a set of primers (Supplementary Table 1). The amount of target mRNA was
first normalized to the expression level of the 18 S rRNA amplified from the
same sample and then to untreated controls. Human SPP1 expression was
measured using TaqMan PCR mix (Life Technologies), specific primers and
FAM-labelled probe sets (Hs00983892 for isoform 1 and Hs00986534 for
isoform 2) and normalized to Gapdh expression (Hs99999905). Data were
analysed with the relative quantification (ΔΔCt) method using 7500 System
SDS Software (Life Technologies).
Glioma implantation and quantification of tumour size
This study was approved by the Local Ethics Committee. Animal
experiments were performed in accordance with national guidelines
and EU regulations (Directive 2010/63/EU) by authorized investigators.
Parental C6 or modified glioma cells (5 × 104 cells in 2.5 μl of Dulbecco's
modified Eagle's medium) were implanted into the right striatum of 250 g
male wistar rats. Animals were randomly allocated to the study groups.
Rats were housed in a temperature- and humidity-controlled environment
under a 12-h light/dark cycle and had free access to water and foodstuff. At
day 15th after implantation, animals were killed and intracardially perfused
with 4% paraformaldehyde. Brains were removed, postfixed in 4%
paraformaldehyde for 48 h, followed by immersion in 30% saccharose
Oncogene (2016) 6366 – 6377
and then frozen with dry CO2. Serial coronal sections were collected. To
quantify the tumour size, 20- μm-thick sections were stained with toluidine
blue, and images were acquired using a Leica DM4000B microscope (Leica
Microsystems, Wetzlar, Germany). Tumour areas were measured using
ImageJ software (NIH, Bethesda, MD, USA) in an every fourth brain slice,
and tumour volumes were calculated as described.13
Immunohistochemistry and immunofluorescence
To detect microglia/macrophages, 12 μm sections were incubated with the
anti-Iba1 antibody (Wako, Osaka, Japan; no. 019-19741; 1:1000) for 16 h at
4 °C, followed by incubation with secondary antibody and visualization
with 3,3′-diaminobenzidine. For the detection of Arg1-positive cells,
sections were incubated with anti-Arg-1 (Santa Cruz Biotechnology, Dallas,
TX, USA; sc-20150, 1: 100) followed by the donkey anti-goat Alexa Fluor
555-conjugated antibody (1:1000; Invitrogen). For double staining follow-
ing staining with anti-Iba1 and anti-Arg1, sections were incubated with
donkey anti-rabbit Alexa Fluor 488 and donkey anti-goat Alexa Fluor 555
antibodies (1:1000; Invitrogen).
Functional tests
Microglia cells were treated for 24 h with CM from 3T3 fibroblasts
co-transfected with lactadherin (Mfge8) and wild-type (Spp1) or mutated
(Spp1 ΔC-term) osteopontin constructs; C6 glioma CM or 100 mg/ml
lipopolysaccharide were used as respective controls. After activation,
the media were exchanged to standard culture media (including
non-treated cells) and microglia culture supernatants were collected
for additional 24 h, centrifuged at 500 g for 10 min and used for
determination of nitric oxide and cytokine production. Released
cytokines/chemokines were measured using the Milliplex Kit (EMD
Millipore, Billerica, MA, USA) on Magpix Instrument (Luminex, Austin, TX,
USA). Nitric oxide production was assessed using Griess reagent (Active
Motif, Rixensart, Belgium) as described (Sliwa et al.12). Free active
transforming growth factor-β was measured using Legend Max ELISA
(BioLegend, San Diego, CA, USA) according to the manufacturer's
instructions. Invasion of C6 glioma cells in the presence of media from
differentially stimulated microglia (microglia culture supernatant) was
evaluated using BioCoat Fluoroblock Invasion System (Corning, Corning,
NY, USA). C6 cells were seeded at 2.5x104 cells per Matrigel Matrix-
coated insert in serum-free medium and microglia culture supernatants
diluted 1:3 in serum-free medium were added to the bottom chambers.
After 20 -h incubation, the cells were fixed onthe insert membranes,
stained with DAPI (1 μg/ml) and microphotographs of the invading cells
from five randomly selected fields per insert were taken. A total number
of cells was counted using ImageJ.
Statistical analysis
Most experiments were performed on 3–5 independently derived primary
microglial cultures. Statistical analyses were performed using Student’s
t-test or, for multiple comparisons, one-way analysis of variance (ANOVA)
followed by post hoc Newman–Keuls test with Statistica software (StatSoft,
Tulsa, OK, USA). The Mann–Whitney U-test was used for the analysis of
in vivo experiments. The n, P‐values and statistical analysis methods
are indicated in the figure legends.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
ACKNOWLEDGEMENTS
We thank Beata Kaza and Marcin Sliwa for technical assistance. This work was
supported by the Grant 2012/04/A/NZ3/00630 (to BK) from the National Science
Center (Poland). PG was supported by GA no. 264173 Bio-Imagine and IP2011013171
from the Ministry of Science and Higher Education.
REFERENCES
1 Mantovani A, Sica A. Macrophages, innate immunity and cancer: balance,
tolerance, and diversity. Curr Opin Immunol 2010; 22: 231–237.
2 Noy R, Pollard JW. Tumor-associated macrophages: from mechanisms to therapy.
Immunity 2014; 41: 49–61.
© 2016 Macmillan Publishers Limited, part of Springer Nature.

3 Murray PJ, Allen JE, Biswas SK, Fisher EA, Gilroy DW, Goerdt S et al. Macrophage
activation and polarization: nomenclature and experimental guidelines. Immunity
2014; 41: 14–20.
4 Heusinkveld M, de Vos van Steenwijk PJ, Goedemans R, Ramwadhdoebe TH,
Gorter A, Welters MJ et al. M2 macrophages induced by prostaglandin E2 and IL-6
from cervical carcinoma are switched to activated M1 macrophages by CD4+
Th1 cells. J Immunol 2011; 187: 1157–1165.
5 Roca H, Varsos ZS, Sud S, Craig MJ, Ying C, Pienta KJ. CCL2 and interleukin-6
promote survival of human CD11b+ peripheral blood mononuclear cells and
induce M2-type macrophage polarization. J Biol Chem 2009; 284: 34342–34354.
6 Solinas G, Schiarea S, Liguori M, Fabbri M, Pesce S, Zammataro L et al.
Tumor-conditioned macrophages secrete migration-stimulating factor: a new
marker for M2-polarization, influencing tumor cell motility. J Immunol 2010; 185:
642–652.
7 Kim S, Takahashi H, Lin WW, Descargues P, Grivennikov S, Kim Y et al. Carcinoma-
produced factors activate myeloid cells through TLR2 to stimulate metastasis.
Nature 2009; 457: 102–106.
8 Badie B, Schartner JM. Flow cytometric characterization of tumor-
associated macrophages in experimental gliomas. Neurosurgery 2000; 46:
957–961; discussion 961–952.
9 Komohara Y, Ohnishi K, Kuratsu J, Takeya M. Possible involvement of the M2
anti-inflammatory macrophage phenotype in growth of human gliomas. J Pathol
2008; 216: 15–24.
10 Wei J, Gabrusiewicz K, Heimberger A. The controversial role of microglia in
malignant gliomas. Clin Dev Immunol 2013; 2013: 285246.
11 Markovic DS, Vinnakota K, Chirasani S, Synowitz M, Raguet H, Stock K et al.
Gliomas induce and exploit microglial MT1-MMP expression for tumor expansion.
Proc Natl Acad Sci USA 2009; 106: 12530–12535.
12 Sliwa M, Markovic D, Gabrusiewicz K, Synowitz M, Glass R, Zawadzka M et al.
The invasion promoting effect of microglia on glioblastoma cells is inhibited by
cyclosporin A. Brain 2007; 130: 476–489.
13 Gabrusiewicz K, Ellert-Miklaszewska A, Lipko M, Sielska M, Frankowska M,
Kaminska B. Characteristics of the alternative phenotype of microglia/macro-
phages and its modulation in experimental gliomas. PLoS One 2011; 6: e23902.
14 Hussain SF, Heimberger AB. Immunotherapy for human glioma: innovative
approaches and recent results. Expert Rev Anticancer Ther 2005; 5: 777–790.
15 Hussain SF, Yang D, Suki D, Aldape K, Grimm E, Heimberger AB. The role of human
glioma-infiltrating microglia/macrophages in mediating antitumor immune
responses. Neuro-Oncology 2006; 8: 261–279.
16 Hussain SF, Yang D, Suki D, Grimm E, Heimberger AB. Innate immune functions
of microglia isolated from human glioma patients. J Transl Med 2006; 4: 15.
17 Sielska M, Przanowski P, Wylot B, Gabrusiewicz K, Maleszewska M, Kijewska M
et al. Distinct roles of CSF family cytokines in macrophage infiltration and
activation in glioma progression and injury response. J Pathol 2013; 230: 310–321.
18 Wei J, Barr J, Kong LY, Wang Y, Wu A, Sharma AK et al. Glioma-associated cancer-
initiating cells induce immunosuppression. Clin Cancer Res 2010; 16: 461–473.
19 Wesolowska A, Kwiatkowska A, Slomnicki L, Dembinski M, Master A, Sliwa M et al.
Microglia-derived TGF-beta as an important regulator of glioblastoma invasion—
an inhibition of TGF-beta-dependent effects by shRNA against human TGF-beta
type II receptor. Oncogene 2008; 27: 918–930.
20 Ye XZ, Xu SL, Xin YH, Yu SC, Ping YF, Chen L et al. Tumor-associated microglia/
macrophages enhance the invasion of glioma stem-like cells via TGF-beta1
signaling pathway. J Immunol 2012; 189: 444–453.
21 Ellert-Miklaszewska A, Dabrowski M, Lipko M, Sliwa M, Maleszewska M, Kaminska
B. Molecular definition of the pro-tumorigenic phenotype of glioma-activated
microglia. Glia 2013; 61: 1178–1190.
22 Platten M, Kretz A, Naumann U, Aulwurm S, Egashira K, Isenmann S et al.
Monocyte chemoattractant protein-1 increases microglial infiltration and
aggressiveness of gliomas. Ann Neurol 2003; 54: 388–392.
Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)
© 2016 Macmillan Publishers Limited, part of Springer Nature.
Microglia reprogramming by glioma-derived signals
A Ellert-Miklaszewska et al
6377
23 Okada M, Saio M, Kito Y, Ohe N, Yano H, Yoshimura S et al. Tumor-associated
macrophage/microglia infiltration in human gliomas is correlated with MCP-3,
but not MCP-1. Int J Oncol 2009; 34: 1621–1627.
24 Held-Feindt J, Hattermann K, Muerkoster SS, Wedderkopp H, Knerlich-Lukoschus
F, Ungefroren H et al. CX3CR1 promotes recruitment of human glioma-infiltrating
microglia/macrophages (GIMs). Exp Cell Res 2010; 316: 1553–1566.
25 Ku MC, Wolf SA, Respondek D, Matyash V, Pohlmann A, Waiczies S et al. GDNF
mediates glioblastoma-induced microglia attraction but not astrogliosis.
Acta Neuropathol 2013; 125: 609–620.
26 Pyonteck SM, Akkari L, Schuhmacher AJ, Bowman RL, Sevenich L, Quail DF et al.
CSF-1 R inhibition alters macrophage polarization and blocks glioma progression.
Nat Med 2013; 19: 1264–1272.
27 Raymond A, Ensslin MA, Shur BD. SED1/MFG-E8: a bi-motif protein that
orchestrates diverse cellular interactions. J Cell Biochem 2009; 106: 957–966.
28 Yamaguchi Y, Shao Z, Sharif S, Du XY, Myles T, Merchant M et al. Thrombin-
cleaved fragments of osteopontin are overexpressed in malignant glial tumors
and provide a molecular niche with survival advantage. J Biol Chem 2013;
288: 3097–3111.
29 Shao Z, Morser J, Leung LL. Thrombin cleavage of osteopontin disrupts
a pro-chemotactic sequence for dendritic cells, which is compensated by the
release of its pro-chemotactic C-terminal fragment. J Biol Chem 2014; 289:
27146–27158.
30 Agnihotri R, Crawford HC, Haro H, Matrisian LM, Havrda MC, Liaw L. Osteopontin,
a novel substrate for matrix metalloproteinase-3 (stromelysin-1) and matrix
metalloproteinase-7 (matrilysin). J Biol Chem 2001; 276: 28261–28267.
31 Martinez FO, Helming L, Milde R, Varin A, Melgert BN, Draijer C et al. Genetic
programs expressed in resting and IL-4 alternatively activated mouse and human
macrophages: similarities and differences. Blood 2013; 121: e57–e69.
32 Anborgh PH, Mutrie JC, Tuck AB, Chambers AF. Role of the metastasis-promoting
protein osteopontin in the tumour microenvironment. J Cell Mol Med 2010; 14:
2037–2044.
33 Silvestre JS, Thery C, Hamard G, Boddaert J, Aguilar B, Delcayre A et al. Lactadherin
promotes VEGF-dependent neovascularization. Nat Med 2005; 11: 499–506.
34 Pello OM, De Pizzol M, Mirolo M, Soucek L, Zammataro L, Amabile A et al. Role
of c-Myc in alternative activation of human macrophages and tumor-associated
macrophage biology. Blood 2011; 119: 411–421.
35 Denhardt DT, Noda M, O'Regan AW, Pavlin D, Berman JS. Osteopontin as a means
to cope with environmental insults: regulation of inflammation, tissue remodel-
ing, and cell survival. J Clin Invest 2001; 107: 1055–1061.
36 Smith LL, Cheung HK, Ling LE, Chen J, Sheppard D, Pytela R et al. Osteopontin
N-terminal domain contains a cryptic adhesive sequence recognized by
alpha9beta1 integrin. J Biol Chem 1996; 271: 28485–28491.
37 Hanayama R, Tanaka M, Miwa K, Shinohara A, Iwamatsu A, Nagata S. Identification
of a factor that links apoptotic cells to phagocytes. Nature 2002; 417: 182–187.
38 Grassinger J, Haylock DN, Storan MJ, Haines GO, Williams B, Whitty GA et al.
Thrombin-cleaved osteopontin regulates hemopoietic stem and progenitor
cell functions through interactions with alpha9beta1 and alpha4beta1 integrins.
Blood 2009; 114: 49–59.
39 Cook AC, Tuck AB, McCarthy S, Turner JG, Irby RB, Bloom GC et al. Osteopontin
induces multiple changes in gene expression that reflect the six 'hallmarks
of cancer' in a model of breast cancer progression. Mol Carcinogen 2005; 43:
225–236.
40 Zhou W, Ke SQ, Huang Z, Flavahan W, Fang X, Paul J et al. Periostin secreted
by glioblastoma stem cells recruits M2 tumour-associated macrophages and
promotes malignant growth. Nat Cell Biol 2015; 17: 170–182.
41 Przanowski P, Dabrowski M, Ellert-Miklaszewska A, Kloss M, Mieczkowski J, Kaza B
et al. The signal transducers Stat1 and Stat3 and their novel target Jmjd3 drive
the expression of inflammatory genes in microglia. J Mol Med (Berl) 2014; 92:
239–254.
Oncogene (2016) 6366 – 6377