Content uploaded by Guido Serini
Author content
All content in this area was uploaded by Guido Serini
Content may be subject to copyright.
Content uploaded by Guido Serini
Author content
All content in this area was uploaded by Guido Serini
Content may be subject to copyright.
Research article
The Journal of Clinical Investigation http://www.jci.org
Semaphorin 3A overcomes cancer hypoxia
and metastatic dissemination induced
by antiangiogenic treatment in mice
Federica Maione,1,2,3 Stefania Capano,1,2,3 Donatella Regano,1,2,3 Lorena Zentilin,4 Mauro Giacca,4
Oriol Casanovas,5 Federico Bussolino,2,3,6,7 Guido Serini,2,3,7,8 and Enrico Giraudo1,2,3,7
1Laboratory of Transgenic Mouse Models, 2IRCC, and 3Department of Oncological Sciences, University of Torino School of Medicine, Candiolo, Italy.
4International Centre for Genetic Engineering and Biotechnology (ICGEB), Molecular Medicine Laboratory, Trieste, Italy. 5Tumor Angiogenesis Group,
Translational Research Laboratory, Catalan Institute of Oncology — IDIBELL, L’Hospitalet de Llobregat, Spain. 6Laboratory of Vascular Oncology,
University of Torino School of Medicine, Candiolo, Italy. 7Center for Complex Systems in Molecular Biology and Medicine (SysBioM), University of Torino,
Torino, Italy. 8Laboratory of Cell Adhesion Dynamics, University of Torino School of Medicine, Candiolo, Italy.
Cancer development, progression, and metastasis are highly dependent on angiogenesis. The use of antiangio-
genic drugs has been proposed as a novel strategy to interfere with tumor growth, but cancer cells respond by
developing strategies to escape these treatments. In particular, animal models show that antiangiogenic drugs
currently used in clinical settings reduce tumor tissue oxygenation and trigger molecular events that foster can-
cer resistance to therapy. Here, we show that semaphorin 3A (Sema3A) expression overcomes the proinvasive
and prometastatic resistance observed upon angiogenesis reduction by the small-molecule tyrosine inhibitor
sunitinib in both pancreatic neuroendocrine tumors (PNETs) in RIP-Tag2 mice and cervical carcinomas in
HPV16/E2 mice. By improving cancer tissue oxygenation and extending the normalization window, Sema3A
counteracted sunitinib-induced activation of HIF-1α, Met tyrosine kinase receptor, epithelial-mesenchymal
transition (EMT), and other hypoxia-dependent signaling pathways. Sema3A also reduced tumor hypoxia
and halted cancer dissemination induced by DC101, a specific inhibitor of the VEGF pathway. As a result,
reexpressing Sema3A in cancer cells converts metastatic PNETs and cervical carcinomas into benign lesions.
We therefore suggest that this strategy could be developed to safely harnesses the therapeutic potential of the
antiangiogenic treatment.
Introduction
Angiogenesis is required for invasive tumor growth and meta-
static dissemination, providing the rationale for the development
of antiangiogenic therapies (1). Despite the generation of innova-
tive antiangiogenic strategies, such as inhibitors of the VEGF-A
pathway, resistance to anti-VEGF therapy has been recently
observed in both preclinical and clinical trials (2, 3). For instance,
preclinical studies provided evidence for anti-VEGF drug evasion
by activation of alternate proangiogenic pathways, likely induced
by a significant increase of tumor tissue hypoxia (4). Therefore, to
extend the optimal therapeutic windows and design more effective
antiangiogenic combinatory regimens that could prevent or block
tumor invasion and metastasis formation, it is critical to identify
new angiogenic modulators and uncover their molecular and cel-
lular mechanisms of action in vivo.
It is well documented that, as a result of architectural and bio-
logical abnormalities such as tortuosity, leakiness, and lack of
pericytes, tumor blood vessels are structurally and functionally
aberrant (5), causing cancer tissue hypoxia (6). Notably, abnormal
vascular permeability and chronic oxygen shortage promote tumor
invasiveness, for example, by upregulating HIF-1α expression (3, 7),
downregulating E-cadherin expression (8), and hyperactivating
hepatocyte growth factor/Met (HGF/Met) signaling (9). Further-
more, several independent preclinical studies (10, 11), which have
not yet been paralleled by analogous clinical trials, revealed that
although impairing cancer angiogenesis with different therapeu-
tic approaches initially causes remarkable shrinkage of the tumor
mass, this approach eventually causes dramatic enhancement of
tumor invasiveness and increased distal metastasis formation.
These data, together with the formal demonstration that improv-
ing oxygenation can suppress metastatization of cancer cells and
promote their differentiation (12), further support the hypoth-
esis that vascular normalization could represent a remarkably
advantageous anticancer strategy, as it is also able to favor chemo-
therapy delivery and response to radiotherapy (6). We previously
showed that endothelial semaphorin 3A (Sema3A) is an endog-
enous antiangiogenic agent that, when reexpressed in cancers that
lost it, is able to normalize the vasculature and to block tumor
growth, finally inducing a stable disease (13). In the present study,
we investigated the molecular and cellular mechanisms by which
Sema3A, alone or in combination with different antiangiogenic
drugs, is able to impair tumor cell dissemination and overcome
evasive resistance to angiogenesis inhibition (14).
Results
Sema3A halts tumor invasion and metastasis formation caused by antian-
giogenic treatment. We previously demonstrated that reexpressing
Sema3A in tumors of a spontaneous mouse model of pancreatic
neuroendocrine cancer (RIP-Tag2) by somatic gene transfer using
adeno-associated virus–8 (AAV8) resulted in reduced vascular den-
sity, inhibition of tumor growth, significant survival extension,
normalization of tumor vasculature, and decreased tumor hypoxia
(13). Stemming from these data, we sought to investigate whether
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest doi:10.1172/JCI58976.
research article
The Journal of Clinical Investigation http://www.jci.org
Sema3A also impairs tumor invasion and metastasis formation
to overcome the evasive resistance observed in RIP-Tag2 mice and
other mouse models in response to antiangiogenic therapies (10,
11). We first compared the effect of AAV8-Sema3A (referred to here-
in as Sema3A) and sunitinib, a prototypical small-molecule tyro-
sine kinase (TK) inhibitor and antiangiogenic drug (15), on tumor
dissemination in RIP-Tag2 mice by performing a 4-week regression
trial between 12 and 16 weeks of age. Treatment of tumor-bearing
RIP-Tag2 mice with sunitinib induced primary tumor shrinkage
and strongly inhibited angiogenesis (see below), but at the same
time promoted local invasiveness and distant metastasis formation
compared with controls (Figure 1, Figure 2, A–G, and Supplemental
Figure 1; supplemental material available online with this article;
doi:10.1172/JCI58976DS1), consistent with previous findings (10).
On the contrary, treatment with Sema3A alone not only reduced
tumor burden and vascularization (ref. 13 and see below), but also
dramatically decreased cancer invasion in the surrounding tissues
as well as the incidence and volume of peripancreatic LN metas-
tases and the incidence, number, and volume of liver metastases
compared with controls (Figure 1 and Figure 2, A–G). Hence, we
showed Sema3A to be an angiogenesis inhibitor that, differently
from sunitinib, also displayed a powerful antimetastatic activity.
The recent observation that systemic delivery of Sema3A inhibits
angiogenesis and metastatization in xenograft tumor models as
well (16) further supports the role of Sema3A as an effective phar-
macological inhibitor of cancer progression. To start investigating
whether Sema3A antagonizes the previously described proinvasive
effect of sunitinib in RIP-Tag2 mice (10), we set up a combinatory
therapeutic regimen, treating RIP-Tag2 mice simultaneously with
Sema3A and sunitinib for 1 month, after which we assessed the
frequency of invasive lesions and metastasis formation. Notably,
the combination of Sema3A with sunitinib strongly reduced the
incidence of fully invasive (IC2) tumors and the extent of both LN
and liver metastases in sunitinib-treated animals (Figure 1, B and C,
and Figure 2, A–G). Together, these data demonstrated that
Sema3A not only impaired metastasis formation during spontane-
ous tumorigenesis, but also curbed the increased cancer aggressive-
ness stimulated by sunitinib treatment.
Sema3A and sunitinib synergize to enhance survival. Based on our
observation that the combination of Sema3A with sunitinib effec-
tively hampered the evasive resistance elicited by sunitinib treat-
ment alone in RIP-Tag2 mice, we next investigated whether these
2 drugs could synergistically impair tumor progression and hence
extend RIP-Tag2 survival as well. We performed a longer survival
trial in which RIP-Tag2 tumor-bearing mice were treated beginning
at 12 weeks of age with AAV8-LacZ (referred to herein as LacZ)
plus vehicle (control), Sema3A, sunitinib, or combined Sema3A
and sunitinib. The median survival of control mice was 2.5 weeks.
Similar to our previous observations (13), Sema3A significantly
prolonged the survival of RIP-Tag2 mice by 9.0 weeks compared
with control-treated animals (P < 0.001), 2.3 weeks longer than
that observed with sunitinib treatment alone (P < 0.01). Treatment
Figure
Sema3A blocks tumor invasion caused by antiangiogenic treatment. (A) Total tumor volume in 4-week treatment regression trial (from 12
to 16 weeks of age) showed that sunitinib (SUN), AAV8-Sema3A in combination with sunitinib, and AAV8-Sema3A reduced tumor burden
64%, 63%, and 66%, respectively, compared with controls (AAV8-LacZ–injected RIP-Tag2 treated with vehicle; see Methods). (B) Percent
encapsulated (IT), microinvasive (IC1), and fully invasive (IC2) carcinomas. Combined AAV8-Sema3A and sunitinib decreased IC2 carcinoma
incidence 62% compared with sunitinib alone. Sema3A-treated tumors showed a statistically significant 64% decrease of IC2 carcinomas
compared with controls. (C) Analysis of tumor invasiveness by means of SV40 T-antigen immunostaining. Sunitinib-treated tumors displayed
an invasive front extensively intercalated into the surrounding tissue (arrows). T, tumor; Ac, acinar tissue. *P < 0.05, **P < 0.01, unpaired
Mann-Whitney U test. Scale bars: 50 μm.
research article
The Journal of Clinical Investigation http://www.jci.org
Figure
Sema3A blocks distal metastasis formation caused by antiangiogenic treatment. (A and B) SV40 T-antigen immunostaining (arrows) of peri-
pancreatic LN and liver metastases (Met). Representative images were derived from serial section analysis in each animal (n = 30 per treat-
ment group). (C and D) Metastasis incidence per animal. Combined Sema3A and sunitinib decreased LN and liver metastases 57% and 54%,
respectively, compared with sunitinib alone. Sema3A reduced LN and liver metastases 62% and 50%, respectively, compared with controls.
(E) LN metastasis volume. Combined Sema3A and sunitinib decreased LN volume 89% compared with sunitinib alone; Sema3A decreased
LN volume 79% compared with controls. (F and G) Number (F) and volume (G) of liver metastases (n = 30 per treatment group). Combined
Sema3A and sunitinib diminished the number and volume of liver metastases 88% and 91%, respectively, compared with sunitinib alone;
Sema3A reduced liver metastasis number and volume 80% and 85%, respectively, compared with controls. (H) Survival trial in tumor-bearing
RIP-Tag2 mice treated continuously with 40 mg/kg/d sunitinib, Sema3A, combined Sema3A and sunitinib, or LacZ beginning at 12 weeks
(n = 20 per group). *P < 0.05, **P < 0.01, unpaired Mann-Whitney U test or log-rank test (for survival trials). Scale bars: 50 μm.
research article
The Journal of Clinical Investigation http://www.jci.org
with sunitinib increased survival 6.7 weeks compared with con-
trols (P < 0.001) (Figure 2H), as previously shown (10). Interest-
ingly, this survival trial clearly demonstrated that the combination
of Sema3A with sunitinib significantly enhanced the survival of
RIP-Tag2 mice by 16.2, 7.2, and 9.5 weeks compared with control,
Sema3A, and sunitinib, respectively (P < 0.001 for all comparisons;
Figure 2H), suggestive of effective synergism of Sema3A and suni-
tinib regarding survival and tumor progression. Of note, 18 weeks
after the initial treatment with combined Sema3A and sunitinib,
6 of 20 mice of the survival trial were still alive, and 2 of these
were tumor free. Interestingly, similarly to what we observed in the
4-week regression trial, this combinatorial treatment resulted in very
small and round tumors (Supplemental Figure 4A) and strongly
halted tumor invasiveness in the RIP-Tag2 mice that survived until
the end of the trial. Importantly, none of the 6 surviving mice had
liver or peripancreatic LN metastases. Together, these data indicate
that the combination of Sema3A with sunitinib results in a syner-
gistic effect by prolonging animal survival and inducing smaller,
less invasive, and less frequent metastatic cancers.
Sema3A counteracts basal and sunitinib-elicited tumor hypoxia. Both
primary tumors and metastases of mice treated with antiangiogen-
ic drugs are highly hypoxic (10), and preclinical studies suggest that
evasion to antiangiogenic therapies could depend on the hypoxia-
driven induction of alternative proangiogenic pathways in tumor
cells (4, 14). Based on the vascular normalizing effect of Sema3A we
previously observed in RIP-Tag2 tumors (13), we hypothesized that
this molecule could overcome the evasive resistance to angiogen-
esis inhibition by hampering tumor hypo-oxygenation. We there-
fore measured tissue hypoxia in RIP-Tag2 insulinomas treated with
sunitinib, Sema3A, or both in combination. The strong reduction
of vessel area induced by sunitinib was accompanied by an increase
in intratumoral hypoxia, as assessed by pimonidazole staining
(Figure 3, A and B). As previously shown, treating RIP-Tag2 mice
with Sema3A for 1 month proportionally restrained the amount of
blood vessels and normalized the remaining vasculature, abrogat-
ing the tumor hypoxia observed in control mice at both the begin-
ning and the end of the therapeutic trial (ref. 13 and Supplemental
Figure 2). Remarkably, combinatory treatment with Sema3A com-
pletely reversed the sizeable hypoxia observed in sunitinib-treated
RIP-Tag2 insulinomas (Figure 3, A and B).
To further characterize the extent of tumor hypoxia associated
with the different therapeutic regimens, we assessed the expres-
sion of HIF-1α, a master regulator of cellular adaptation to oxygen
deprivation that acts as a survival factor for hypoxic cancer cells,
being expressed in many human cancers and associated with poor
prognosis and treatment failure (3, 7). Remarkably, Western blot
analysis revealed a strong increase of HIF-1α protein in sunitinib-
treated tumors that was dramatically reduced by simultaneous
treatment with Sema3A (Figure 3, C and D). Of note, adminis-
tering sunitinib, alone or in combination with Sema3A, resulted
Figure
Sema3A impairs basal and sunitinib-elicited tumor hypoxia and HIF-1α expression. (A) Tumor hypoxia was assessed by means of pimoni-
dazole adduct immunostaining (arrows) in serial sections of tumors of RIP-Tag2 mice treated for 4 weeks with sunitinib, Sema3A, and the
combination compared with control. Representative images were derived from serial section analysis in each animal (n = 15 per treatment
group). (B) Quantification of hypoxic tumors, expressed as percent of pimonidazole density area per tumor, per animal. Combined Sema3A
and sunitinib reduced hypoxic area 84% compared with sunitinib alone; Sema3A decreased hypoxic area 90% compared with controls.
(C) Western blot analysis of HIF-1α, CA9, and vinculin (as protein loading control) in tumors isolated from RIP-Tag2 mice treated as indicated.
(D) Relative levels of HIF-1α and CA9 protein expression, normalized to vinculin (n = 12 per treatment group). Combined Sema3A and sunitinib
diminished HIF-1α expression and CA9 amount 91% and 60%, respectively, compared with sunitinib alone; Sema3A decreased CA9 expres-
sion 55% compared with control. *P < 0.05, **P < 0.01, unpaired Mann-Whitney U test. Scale bars: 50 μm.
research article
The Journal of Clinical Investigation http://www.jci.org
Figure
Sema3A alone or in combination with sunitinib enhances pericyte coverage, reduces blood vessel leakage, improves tissue perfusion, and
increases doxorubicin delivery to tumors. (A and B) As assessed by confocal analysis of Meca32 immunostaining, vessel density in RIP-Tag2
tumors after a 4-week regression trial was reduced 89% in sunitinib-treated tumors, 60% in Sema3A and sunitinib–treated tumors, and 47%
Sema3A-treated tumors compared with controls. (A and C–F) Pericyte coverage, as evaluated by confocal colocalization analysis of Meca32
with NG2 (A, arrows, and C) or with α-SMA (D), PDGFR-β (E), or desmin (F). Combined Sema3A and sunitinib enhanced pericyte coverage
62% (NG2), 72% (α-SMA), 84% (PDGFR-β), and 69% (desmin) compared with sunitinib alone; Sema3A increased pericyte coverage 45%
(NG2), 47% (α-SMA), 40% (PDGFR-β), and 43% (desmin) compared with controls. (G) In tumors treated with Sema3A alone or in combination
with sunitinib, FITC-lectin perfused vessels (arrows) were increased compared with sunitinib-treated and control insulinomas. (H) FITC-dextran
extravasation (arrows) decreased in tumors treated with Sema3A alone or in combination with sunitinib compared with sunitinib-treated and con-
trol insulinomas. Results are from 5 fields per mouse (n = 12 per treatment group). (I) Amount of doxorubicin (DXR) present in tumors, expressed
as μg equivalent/g tumor. Combined Sema3A and sunitinib enhanced doxorubicin 87% compared with sunitinib alone; Sema3A enhanced
doxorubicin 83% compared with controls. *P < 0.05, **P < 0.01, unpaired Mann-Whitney U test. Scale bars: 50 μm.
research article
The Journal of Clinical Investigation http://www.jci.org
in similar modulation of the HIF-1α target gene carbonic anhy-
drase 9 (CA9) (Figure 3, C and D), which is also upregulated in
several human cancers (3, 17). Moreover, in agreement with the
normoxic tumor environment induced by Sema3A (ref. 13 and
Supplemental Figure 2), we also observed a significant reduction
of CA9 in animals treated with Sema3A alone compared with con-
trols (Figure 3, C and D). Thus, by virtue of its ability to normalize
tumor blood vessels and to reestablish tissue normoxia, Sema3A
efficiently overcame the invasive phenotype elicited by sunitinib
in RIP-Tag2 mice.
The combination of Sema3A and sunitinib increases pericyte coverage,
reduces blood vessel leakage, enhances tumor tissue perfusion, and pro-
longs the vascular normalization window. Increased pericyte coverage
and reduction in vascular density and branching are hallmarks
of tumor blood vessel normalization, a process that occurs in
response to some antiangiogenic agents and allows for more effi-
cient delivery of oxygen and chemotherapeutic drugs (6, 18, 19). As
expected (10, 15, 20), in sunitinib-treated tumors, in addition to a
strong reduction of blood vessel area, we observed remarkable inhi-
bition of pericyte coverage, as revealed by confocal analysis of NG2
staining (Figure 4, A–C). On the contrary, as previously described
(13, 20), Sema3A treatment increased the number of perivascular
NG2+ cells. Of note, simultaneous treatment with Sema3A and
sunitinib significantly increased pericyte coverage (Figure 4, A
and C). Similar observations with respect to pericyte coverage in
the different treatment groups were obtained using other markers,
such as α-SMA, PDGFR-β, and desmin (Figure 4, D–F). To better
characterize Sema3A-elicited tumor blood vessel normalization,
we further studied the perfusion and permeability of the tumor
vasculature of RIP-Tag2 mice undergoing different therapeutic
Figure
Sema3A reduces both basal and sunitinib-induced EMT and inhibits NF-κB expression. (A–D) RIP-Tag2 tumors were treated for 4 weeks
with sunitinib, AVV8-Sema3A, and the combination; shown are (A and C) representative confocal microscopy and (B and D) protein expres-
sion, quantified by MFI per animal. (A and B) Low E-cadherin expression levels in controls and sunitinib-treated insulinomas were significantly
upregulated by Sema3A treatment alone or in combination with sunitinib. (C and D) Increased vimentin expression in sunitinib-treated tumors
compared with controls was inhibited by treatment with Sema3A alone or in combination with sunitinib. (E) Western blot analysis of protein
lysates derived from tumors of the different treatment groups. Sema3A inhibited sunitinib-induced Snail1 and vimentin expression. E-cadherin
was highly expressed with Sema3A treatment, alone or in combination with sunitinib, compared with the control and sunitinib alone. Sema3A
decreased both basal and sunitinib-induced NF-κB expression. (F) Relative protein expression levels of EMT markers and NF-κB expression,
normalized to vinculin (n = 6 per treatment group). *P < 0.05, **P < 0.01, unpaired Mann-Whitney U test. Scale bars: 50 μm.
research article
The Journal of Clinical Investigation http://www.jci.org
regimens. At the end of each scheme of treat-
ment, blood vessel perfusion and permeabil-
ity were respectively assessed by tail-injecting
animals with FITC-lectin and 70-kDa FITC-
dextran. Interestingly, Sema3A (both alone
and in combination with sunitinib) strongly
increased the amount of FITC-lectin in the
vasculature of insulinomas (Figure 4G and
Supplemental Figure 3, A and C), demon-
strating a clear improvement of tumor blood
vessel perfusion. Moreover, the therapy with
Sema3A significantly reduced both basal and
sunitinib-elicited leakiness of tumor blood
vessels, as measured by confocal analysis of
70-kDa FITC-dextran extravasation (Figure
4H and Supplemental Figure 3, B and D).
To ass ess wh ether Sema3 A, al one o r in
combination with sunitinib, increases the
efficacy of tumor blood vessels in delivering a
chemotherapeutic drug, we injected doxoru-
bicin in RIP-Tag2 mice that were previously
treated for 4 weeks with LacZ plus vehicle
(control) or with Sema3A, sunitinib, or both
in combination, then assessed the amount
of drug present within tumor tissues. We
detected decreased levels of doxorubicin in
insulinomas of sunitinib-treated mice com-
pared with controls (Figure 4I). In contrast,
Figure
Sema3A inhibits basal and sunitinib-induced
expression and activation of Met TK receptor.
(A) Protein analysis showed greater total Met
and phospho-Met in RIP-Tag2 tumors treated
with sunitinib compared with controls; notably,
both were reduced by Sema3A, either alone or
in combination with sunitinib. Vinculin protein
was used as loading control. (B) Protein lev-
els of phospho-Met, normalized to total Met.
(C–F) Presence and localization of Met and
phospho-Met in RIP-Tag2 tumors treated with
sunitinib, Sema3A, and the combination com-
pared with controls, as evaluated by colocaliza-
tion of anti-Met or anti–phospho-Met Abs with
Meca32. Shown are (C and E) representative
confocal microscopy, representative of 5 fields
per mouse, and (D and F) quantification by MFI
(n = 8 per treatment group). (C and D) Total
Met was present in a subset of tumor blood
vessels and in cancer cells in basal conditions
and was highly expressed in both vessels and
tumor cells after sunitinib treatment (arrows);
combined Sema3A and sunitinib reduced Met
both in vessels and tumor cells. (E and F) Phos-
pho-Met was present in some tumor vessels
(arrows) and in cancer cells in basal conditions,
but increased mainly in tumor cells in sunitinib-
treated animals. Notably, Sema3A alone or
combined with sunitinib inhibited both total and
phospho-Met. *P < 0.05, **P < 0.01, unpaired
Mann-Whitney U test. Scale bars: 50 μm.
research article
The Journal of Clinical Investigation http://www.jci.org
greater amounts of doxorubicin were present in Sema3A-treated
tumors (alone or in combination with sunitinib) compared with
control and sunitinib-treated insulinomas (Figure 4I). Based on
these findings, we next investigated whether Sema3A is capable of
increasing and sustaining the ability of the cancer vasculature to
deliver a drug in tumors over a longer period of time, in the hopes
of verifying the presence of an extended normalization window.
We evaluated the degree of pericyte coverage, tissue hypoxia, and
doxorubicin delivery efficiency to tumors of RIP-Tag2 mice treated
with combined Sema3A and sunitinib and found to be alive at the
end of the survival trial. Remarkably, similarly to what we observed
in the 4-week regression trial (Figure 4, A–I), the insulinomas of
the surviving mice treated with combined Sema3A and sunitinib
were not hypoxic and displayed a functional, nonleaky vasculature,
covered by pericytes (Supplemental Figure 4, B–E), that was able
to efficiently deliver doxorubicin to tumors (24.6 ± 2.3 μg equiva-
lents/g tumor). Taken together, these data suggest that Sema3A,
alone or in combination with sunitinib, substantially extends the
normalization window of tumor blood vessel and improves the
delivery efficiency of chemotherapeutic drugs to cancers.
Sema3A impairs sunitinib-induced epithelial-mesenchymal transition
and NF-κB expression. To better investigate the molecular mecha-
nisms by which Sema3A effectively counteracts the evasive resis-
tance induced by angiogenesis inhibition in RIP-Tag2 mice, we
first evaluated the expression of proteins that are involved in epi-
thelial-mesenchymal transition (EMT) and support cancer cell
Figure
Sema3A halts sunitinib-elicited tumor invasion and metastasis formation in HPV16/E2 mice. (A) Total tumor volume of HPV16/E2 mice in a
4-week regression trial (from 5 to 6 months of age) revealed that sunitinib, Sema3A combined with sunitinib, and Sema3A alone reduced tumor
burden 58%, 76%, and 53%, respectively, compared with LacZ-injected, vehicle-treated HPV16/E2 controls. (B and C) Liver and lung metas-
tases (arrows), as assessed by oncogene E7 immunostaining. (D and E) Combined Sema3A and sunitinib decreased lung (D) and liver (E)
metastasis incidence 67% and 70%, respectively, compared with sunitinib; Sema3A reduced lung and liver metastasis incidence 84% and 74%,
respectively, compared with controls. (F and G) Combined Sema3A and sunitinib diminished the number (F) and volume (G) of liver metastases
82% and 94%, respectively, compared with sunitinib; Sema3A reduced liver metastasis number and volume 60% and 92%, respectively, com-
pared with controls. Representative images were derived from serial section analysis of each animal ( n = 12 per treatment group). *P < 0.05,
**P < 0.01, unpaired Mann-Whitney U test. Scale bars: 50 μm.
research article
The Journal of Clinical Investigation http://www.jci.org
research article
0 The Journal of Clinical Investigation http://www.jci.org
metastatic dissemination (21). In EMT, tumor cells typically lose
the epithelial marker E-cadherin and gain mesenchymal markers,
such as vimentin and N-cadherin (21, 22). In addition, E-cadherin
transcriptional repressors, such as Snail1, are also upregulated
during EMT (23). Sunitinib-treated tumors demonstrated high
expression of Snail1 and the mesenchymal markers vimentin and,
to a lesser extent, N-cadherin; in contrast, the Snail1 target E-cad-
herin was strongly inhibited (Figure 5). Hence, sunitinib treatment
promoted invasiveness by activating an EMT program. Remark-
ably, addition of Sema3A completely reverted the effects of suni-
tinib, dramatically inhibiting Snail1 and vimentin and enhancing
E-cadherin expression (Figure 5). In addition, treating animals with
Sema3A alone similarly inhibited the synthesis of mesenchymal
markers and promoted E-cadherin expression as well.
NF-κB is involved in both physiological and pathological pro-
cesses and plays pivotal roles in promoting the EMT-dependent
invasive phenotype of several cancers (24, 25). NF-κB induces
HIF-1α (26), is activated by hypoxia (27), and is a critical component
of the molecular machinery that senses low oxygen levels (28). In
agreement with the above data, we observed that NF-κB protein lev-
els were high in tumors treated with sunitinib and that cotreatment
with Sema3A returned NF-κB expression levels to those observed
with control or Sema3A treatment alone (Figure 5, E and F).
Sema3A inhibits both basal and sunitinib-induced expression and activa-
tion of the Met TK receptor. Based on the known inductive effects of
hypoxia on the expression and activation of the proinvasive TK
receptor Met (9), we assessed total protein and tyrosine phosphor-
ylation levels of Met in treated RIP-Tag2 mice. Western blot analy-
sis revealed that sunitinib treatment caused a significant increase
of both total Met and phospho-Met in tumors (Figure 6, A and B).
However, whereas total Met immunoreactivity was observed in
both blood vessels and tumor cells, phospho-Met was mainly
detected in cancer cells (Figure 6, C–F). Interestingly, concomitant
Sema3A administration fully inhibited the induction of both total
Met and phospho-Met observed with sunitinib treatment alone.
Tumors receiving Sema3A alone displayed a similar reduction of
Met activation (Figure 6, A–C and E). The clear inhibition of Met
TK receptor phosphorylation we observed identified a potential
mechanism through which Sema3A might inhibit metastatization,
namely the inhibition of Met receptor signaling in tumor cells as
result of the reduced tumor hypoxia induced by Sema3A itself.
Sema3A overcomes metastasis formation caused by sunitinib treatment in
a mouse model of spontaneous cervical cancer. To evaluate whether the
effects of Sema3A on tumor progression during angiogenesis inhi-
bition in RIP-Tag2 mice are recapitulated in another tumor model
and histotype, we used the 17β-estradiol–treated K14-HPV16 trans-
genic mouse model of spontaneous cervical carcinogenesis (referred
to herein as HPV16/E2 mice). In our previous work (13), we iden-
tified a similar expression pattern of Sema3A in both HPV16/E2
and RIP-Tag2 transgenic mouse models; in fact, Sema3A was sig-
nificantly expressed in normal and angiogenic premalignant stages,
but was lost in invasive cervical carcinomas. This cervical cancer
model, which recapitulates the progressive development of human
cancers, has high and remarkable similarity in terms of both his-
tological and molecular cross-correlation with its corresponding
human counterpart (29) and has been widely used to test the effect
of antiangiogenic drugs on tumor progression (30).
First, we established an AAV8-based strategy that we believe to
be novel to deliver Sema3A in vivo to cervical tumors of HPV16/E2
mice. To achieve specific gene delivery to the cervix, we established
a route of administration by injecting the recombinant AAV8 virus
expressing either LacZ (control) or Sema3A-myc in the distal por-
tion of the abdominal aorta, just before its bifurcation into the
2 common iliac arteries. This procedure allowed us to efficiently
target and express within the cervix these 2 exogenous gene con-
structs. We observed that the transformation zone — the area at
the border between endo- and ectocervix that undergoes physi-
ological metaplasia, where cervical carcinomas most frequently
develop (30, 31) — was efficiently transduced by AAV8 (see Meth-
ods and Supplemental Figure 5). This Sema3A delivery strategy
allowed us to perform a regression trial with HPV16/E2 mice, as
previously described (30). Tumor-bearing animals were treated
with LacZ control or with Sema3A, sunitinib, or the combination
for 4 weeks, from the fifth to the sixth month of age. Compared
with controls, treatment with either sunitinib or Sema3A alone
strongly reduced tumor volume. Interestingly, the combination
resulted in much more efficient tumor growth inhibition com-
pared with single drug treatments (Figure 7A).
To assess the effect of the different therapeutic regimens on
the formation of distant metastases in HPV16/E2 mice, we evalu-
ated incidence, volume, and number of micrometastases in vital
organs. We evaluated tumor dissemination by means of immuno-
fluorescence and immunohistochemistry using an anti-E7 antibody
that specifically detects the HPV16 viral oncogene E7-positive can-
cer cells in HPV16/E2 mice organs (Figure 7, B and C). Metastasis
incidence in 6-month-old LacZ-treated (control) HPV16/E2 mouse
liver and lungs was 58% and 50%, respectively (Figure 7, D and E).
Interestingly, as observed in RIP-Tag2 mice, 1 month of treatment
with sunitinib increased liver and lung metastasis incidence to 83%
and 75%, respectively. Remarkably, Sema3A, alone and in combina-
tion with sunitinib, strongly reduced the percentage of metastases
compared with sunitinib- or control-treated mice.
Analysis of metastasis number and volume demonstrated that
sunitinib induced a greater number of liver metastases, with
Figure
Sema3A decreases basal and sunitinib-induced hypoxia in HPV16/
E2 mice by normalizing the tumor vasculature. (A) Tumor hypoxia,
assessed by pimonidazole adduct immunostaining in serial sections
of tumors from HPV16/E2 mice treated as indicated for 4 weeks. (B)
Quantification of hypoxic tumors, expressed as percent pimonidazole
density area per tumor, per animal. Sema3A and sunitinib combined
decreased hypoxia 92% compared with sunitinib alone; Sema3A
decreased hypoxia 88% compared with controls. (C) Vessel density,
as assessed by confocal analysis of Meca32 immunostaining, was
reduced 67%, 62%, and 52% by treatment with sunitinib, sunitinib plus
Sema3A, and Sema3A, respectively, compared with controls. (D–H)
Pericyte coverage, as evaluated by confocal colocalization (arrows) of
Meca32 with α-SMA (D and F) or with NG2 (E), PDGFR-β (G), or des-
min (H) and expressed as percent colocalization of pericyte markers on
tumor ECs. Combined Sema3A and sunitinib enhanced pericyte cover-
age 71% (NG2), 68% (α-SMA), 82% (PDGFR-β), and 69% (desmin)
compared with sunitinib alone. Sema3A increased pericyte coverage
40% (NG2), 48% (α-SMA), 39% (PDGFR-β), and 32% (desmin) com-
pared with controls. (I) Increased incidence of FITC-lectin–perfused
vessels (arrows) in Sema3A-treated (alone or in combination with
sunitinib) tumors compared with sunitinib-treated and control insulino-
mas. (J) Decreased FITC-dextran extravasation (arrows) in Sema3A-
treated (alone or in combination with sunitinib) tumors compared with
sunitinib-treated and control insulinomas. Results are from 5 fields per
mouse (n = 8 per treatment group). *P < 0.05, **P < 0.01, unpaired
Mann-Whitney U test. Scale bars: 50 μm.
research article
The Journal of Clinical Investigation http://www.jci.org
larger volumes, compared with controls (Figure 7, F and G).
These data further corroborated the sunitinib-induced evasive
resistance previously observed in RIP-Tag2 and other models
(10) in a different spontaneous mouse model of carcinogenesis.
Notably, Sema3A alone or combined with sunitinib dramatically
decreased the degree of liver and lung metastasis formation (Fig-
ure 7, B–G). These findings bolster our results obtained with RIP-
Tag2 insulinomas, demonstrating that in a very different tumor
Figure
Sema3A blocks DC101-induced cancer invasion and metastatic dissemination. (A) Total tumor volume of RIP-Tag2 mice in a 4-week regres-
sion trial (from 12 to 16 weeks of age). Treatment with DC101, DC101 plus Sema3A, and Sema3A reduced tumor burden 60%, 79%, and 68%,
respectively, compared with LacZ-injected, purified rat IgG–treated RIP-Tag2 controls. (B) Percent encapsulated IT, IC1, and IC2 carcinomas.
Combined treatment with Sema3A and DC101 decreased IC2 carcinoma incidence 67% compared with DC101 alone. (C) DC101-treated tumors
displayed highly invasive tumors (arrows), as shown by SV40 T-antigen immunostaining. (D–G) Analysis of peripancreatic LN (D and E) and liver
(F and G) metastasis formation. Shown is (D and F) representative SV40 T-antigen immunostaining, derived from serial sections analysis of each
animal, and (E and G) quantification of metastasis incidence per animal (n = 15 per group). Combined Sema3A and DC101 decreased LN and
liver metastasis incidence 64% and 75%, respectively, compared with DC101. Sema3A treatment reduced LN and liver metastasis incidence
62% and 50%, respectively, compared with controls. *P < 0.05, **P < 0.01, unpaired Mann-Whitney U test. Scale bars: 50 μm.
research article
The Journal of Clinical Investigation http://www.jci.org
type (i.e., cervical cancer), treatment with Sema3A alone ham-
pered tumor invasiveness and dissemination. Moreover, these
data strengthen the notion that Sema3A can be conceived as a
drug able to overcome the proinvasive and prometastatic effect
of sunitinib in different cancer types.
Sema3A reduces basal and sunitinib-induced hypoxia in cervical cancer
by normalizing the vasculature. We next sought to determine whether
Sema3A counteracts the prometastatic effect of sunitinib in the
HPV16/E2 model by reducing tumor hypoxia, as was observed
in RIP-Tag2 mice. We analyzed tumor hypoxia by pimonidazole
immunostaining. Control HPV16/E2 mice displayed a substantial
amount of hypoxia both in CIN3 premalignant lesions and in cer-
vical tumors (Figure 8, A and B). As observed in RIP-Tag2 mice,
treating HPV16/E2 mice with sunitinib significantly enhanced
hypoxic levels in tumors, in CIN3 lesions, and in the transforma-
tion zone (Figure 8A). Sema3A, alone and in combination with
sunitinib, strongly reduced both basal and sunitinib-induced
tumor hypoxia (Figure 8, A and B). Interestingly, the increase in
tissue hypoxia is a critical factor that promotes cervical carcino-
genesis and has been associated with shorter progression-free and
overall survival and with treatment failure in clinic (32). Therefore,
the reduced hypoxia may represent a major mechanism by which
Sema3A reduces cervical cancer progression and the metastatic
spreading induced by sunitinib.
To confirm that Sema3A decreases tumor hypoxia by promoting
blood vessel normalization in the HPV16/E2 model, we character-
ized pericyte coverage of tumor blood vessels by confocal micros-
copy analysis of the pericyte markers NG2, α-SMA, PDGFR-β, and
desmin. Similar to our findings in RIP-Tag2 mice (Figure 4, A–F),
we observed that 1 month of sunitinib treatment in tumor-bear-
ing HPV16/E2 mice significantly decreased blood vessel area and
greatly reduced pericyte coverage compared with controls (Figure
8, C–H). As expected, Sema3A lessened blood vessel area, but at
the same time induced blood vessel normalization by reducing
vascular branching and increasing the extent of pericyte coverage
of the vasculature compared with sunitinib or control treatments.
Notably, combined Sema3A and sunitinib treatment restored
pericyte coverage compared with sunitinib alone and induced a
vascular phenotype similar to that observed in tumors treated with
Sema3A alone. Next, we assessed blood vessel perfusion and per-
meability in order to determine the functionality of the tumor vas-
culature. In line with the pericyte coverage analysis, we found very
poorly perfused and highly permeable blood vessels in sunitinib-
treated carcinomas, whereas treatment with Sema3A (alone or in
combination with sunitinib) promoted the formation of a highly
perfused and less leaky tumor vasculature (Figure 8, I and J, and
Supplemental Figure 6, A–D). Together, these data in a transgenic
mouse model of cancer other than RIP-Tag2 compellingly suggest
that, by normalizing tumor vasculature and consequently reduc-
ing hypoxia, Sema3A is capable of halting cancer invasiveness and
metastatic spreading while inhibiting tumor angiogenesis.
Sema3A overcomes the evasive resistance induced by an inhibitor of
the VEGF pathway. In order to evaluate whether Sema3A is able to
overcome the resistance to antiangiogenic therapies that specifi-
cally and selectively interfere with the VEGF signaling pathway, we
used DC101, a function-blocking rat monoclonal antibody raised
against VEGFR-2 and previously used in the RIP-Tag2 mouse
model to assess the evasive resistance to angiogenesis inhibition
(4, 10). Similar to the trials performed with sunitinib, we treated
RIP-Tag2 mice for 4 weeks with DC101 alone or in combination
with Sema3A and compared them with mice treated with Sema3A
alone or LacZ plus purified IgG control. Treatment with DC101
exerted effects similar to those we obtained with sunitinib in RIP-
Tag2 mice as well as previously described findings (10). Indeed,
DC101 inhibited tumor angiogenesis and growth, but at the same
time it increased cancer invasiveness as well as the incidence, vol-
ume, and number of LN and liver metastases (Figure 9 and Supple-
mental Figure 7, A–C). Interestingly, when Sema3A treatment was
combined with DC101, we observed a strong reduction of tumor
invasiveness and metastasis formation compared with DC101-
treated mice and controls (Figure 9, B–G).
In addition, whereas insulinomas treated with DC101 were highly
hypoxic and displayed a less pericyte-covered, leakier, and poorly
perfused vasculature, the combination of DC101 with Sema3A
strongly reduced tumor hypoxia, increased blood vessel coverage,
and restored the functionality of the tumor vasculature (Figure
10). Additionally, DC101 exerted a milder effect on the tumor
vasculature than did sunitinib. In fact, although it strongly inhib-
ited the blood vessel area, DC101 reduced the pericyte coverage of
blood vessels less severely than sunitinib did (compare Figure 10,
C–H, and Figure 4, A–F). Of note, in tumors treated with DC101,
we observed a decreased number of NG2+, PDGFR-β+, and desmin+
pericytes, but a significant increase of α-SMA+ perivascular cells,
compared with controls (Figure 10, D–H). These observations
corroborate recent data showing that treatment of RIP-Tag2 mice
with DC101 specifically increased the content of α-SMA+ pericytes,
but not perivascular cells identified by other markers (33).
Compared with the control, DC101 significantly impaired the
perfusion and increased the permeability of tumor blood vessels,
(Figure 10, I and J, and Supplemental Figure 8, A–D), which sug-
gests that the DC101-induced rise of α-SMA+ pericytes was not
sufficient to maintain the function of blood vessels and indicates
that the other pericyte subpopulations would indeed be neces-
sary to fully normalize and improve the function of the tumor
vasculature. Accordingly, simultaneous treatment with DC101
and Sema3A strongly increased all the subpopulation of pericytes
(NG2+, PDGFR-β+, desmin+, and α-SMA+) and simultaneously
improved the perfusion and reduced the vascular leakage, simi-
lar to Sema3A treatment alone (Figure 10, D–J, and Supplemental
Figure 8, A–D). Together, these findings indicate that Sema3A is
able to counteract the evasive resistance induced by the specific
inhibition of VEGF signal pathways.
Discussion
Using 2 transgenic mouse models of spontaneous tumorigenesis,
RIP-Tag2 and HPV16/E2, we here demonstrate what we believe to
be a novel role for Sema3A in overcoming the evasive resistance
previously observed in preclinical mouse models upon angiogen-
esis inhibition (10, 11). When used as single therapeutic agent,
Sema3A strongly inhibited tumor growth, similar to the effects of
sunitinib and DC101; however, different from these latter drugs,
Sema3A also impaired tumor invasion and dissemination to dis-
tal organs. Moreover, thanks to its vascular normalizing activity,
Sema3A ameliorated blood vessel function, improved cancer tis-
sue oxygenation, and lessened several hypoxia-regulated signaling
pathways that support tumor progression and invasion. Conse-
quently, Sema3A efficiently drove sunitinib- or DC101-treated
tumors back from a prometastatic to a benign phenotype.
Recently, several reports on acquired resistance to antiangiogenic
therapies highlighted the need to revisit the current therapies and
research article
The Journal of Clinical Investigation http://www.jci.org
research article
The Journal of Clinical Investigation http://www.jci.org
investigate the possibility of combining tumor shrinkage (induced
by cancer starvation from oxygen and nutrients) with blood vessel
normalization to effectively counteract the metastatic dissemina-
tion of cancer cells, favored, for example, by a hypoxic microenvi-
ronment (34). Here, we showed that the combination of Sema3A
with sunitinib synergistically enhanced RIP-Tag2 mouse survival
and reduced HPV16/E2 mouse tumor burden, finally inducing less
invasive and less frequent metastatic cancers in both transgenic
mouse models. Thus, administration of Sema3A in combination
with sunitinib may represent an innovative and more efficient
therapeutic strategy, thanks to the coupling of sunitinib’s robust
antitumorigenic and antiangiogenic activities (observed in several
human tumors; refs. 35, 36) with Sema3A’s pronormalizing, anti-
invasive, and antimetastatic activities. The main mechanism by
which Sema3A overcame the evasive resistance both to sunitinib
and to DC101 was the ability of this repulsive guidance cue to
restore tumor tissue oxygenation as a result of its powerful blood
vessel normalizing activity. It is known that the pharmacological
targeting of pericytes may disrupt the integrity of the tumor vas-
culature, thus enabling cancer cells to transit into the circulation
system and metastasize (37). Therefore, our data suggest that the
vascular normalizing effect of Sema3A could reduce the proinva-
sive effects of sunitinib by simultaneously inducing tumor tissue
normoxia and blocking cancer cell extravasation.
It has been proposed that the duration of the vascular normal-
ization window is critical to the achievement of long-lasting and
successful therapeutic synergy between antiangiogenic and che-
motherapeutic drugs (6, 18, 19, 38). Notably, all our different
trials demonstrated that Sema3A, alone or in combination with
sunitinib, significantly extended the normalization window of
tumor blood vessels and improved the delivering efficiency of che-
motherapeutic drugs to cancer tissues, by proportionally restrain-
ing the number of blood vessels and simultaneously favoring their
coverage, maturation, and function. It is therefore conceivable to
hypothesize new treatment strategies in which Sema3A could be
combined with other clinically approved chemotherapeutic and/
or antiangiogenic compounds.
The strong inhibition of HIF-1α protein expression we observed,
due to the restoration of tumor tissue oxygenation upon combined
treatment with Sema3A and sunitinib, highlights the crucial role
played by Sema3A in overcoming the resistance to antiangiogenic
therapies. Several HIF-1α inhibitors identified thus far strongly
impair tumor progression in xenograft tumor models and are either
in the early stages of clinical trials or FDA approved for anticancer
therapy (3, 39). Notably, it has previously been shown that the com-
bination of bevacizumab and irinotecan, a topoisomerase I inhibitor
that also inhibits HIF-1α, induced clinical benefit in glioblastoma
patients (40). By correlating the anti-invasive and antimetastatic
effects of Sema3A on hypoxia-stressed cancers with the inhibition
of expression of HIF-1α and its target genes, such as the TK receptor
Met, our data further corroborate the main concept that combining
HIF-1α inhibitors with antiangiogenic drugs can increase the thera-
peutic efficacy and avoid the described side effects (3).
In recent years, several mechanisms of intrinsic and acquired
resistance to antiangiogenic agents have been described (14, 41).
For instance, preclinical studies provided evidence of anti-VEGF
drug evasion by activation of alternative pathways of angiogen-
esis and tumor progression (2). RIP-Tag2 mice have shown rapid
adaptation to anti-VEGF agents, followed by tumor regrowth as
a result of FGF signaling activation (4). An additional possibility
could therefore be that activation of proangiogenic pathways, such
as those triggered by FGFs, can be involved in developing suni-
tinib resistance in RIP-Tag2 mice and that the addition of Sema3A
can inhibit the activation of these compensatory signal path-
ways. Interestingly, the data we obtained by combining Sema3A
with DC101 further corroborated and strengthened the results
obtained with sunitinib and demonstrated that by normalizing
the vasculature and reducing tumor hypoxia, Sema3A overcame
the evasive resistance induced by inhibition of both VEGF (via
DC101) and multiple TK receptor–dependent signaling pathways
(via sunitinib). Notably, our observation of an enhanced number
of α-SMA+ mural cells, paralleled by a simultaneous reduction
of the other pericyte subpopulations (i.e., NG2+, PDGFR-β+, and
desmin+ cells), corroborates recent data showing that treatment
of RIP-Tag2 mice with DC101 specifically increases the content
of α-SMA+ pericytes (33). As those authors suggested, DC101 is
likely to induce a subpopulation of tumor blood vessels covered
by α-SMA+ pericytes deriving from co-opted blood vessels. One
could hence speculate that the increased amount of tumor blood
vessels surrounded by α-SMA+ pericytes may be attributable to the
milder effect that DC101 exerts on blood vessel perfusion and per-
meability compared with sunitinib. However, since we observed
that DC101 significantly impaired perfusion and increased the
permeability of tumor blood vessels compared with controls, such
a DC101-induced rise in α-SMA+ pericytes does not appear suf-
ficient to support the reconstitution of physiologically function-
ing blood vessels. Therefore, the other pericyte subpopulations
seem to be necessary to warrant the appearance of an efficiently
normalized tumor vasculature. Accordingly, simultaneous treat-
ment with DC101 and Sema3A, similar to what we observed with
Sema3A alone, strongly increased all the pericyte subpopulations
and simultaneously improved the perfusion and reduced the
vascular leakage. These observations indicate that sunitinib and
DC101 exert different effects on the tumor vasculature, suggesting
Figure 0
Sema3A impairs DC101-induced hypoxia and improves tumor blood
vessel function. (A) Tumor hypoxia, assessed by pimonidazole adduct
immunostaining (arrows) in tumors from mice treated as indicated
for 4 weeks. (B) Quantification of hypoxic tumors, expressed as per-
cent pimonidazole density area per tumor, per animal. Sema3A and
DC101 combined decreased hypoxia 88% compared with DC101
alone; Sema3A decreased hypoxia 90% compared with controls.
(C) Vessel density, as assessed by confocal microscopy of Meca32
immunostaining, was reduced 75%, 73%, and 50% by treatment with
DC101, DC101 plus Sema3A, and Sema3A, respectively, compared
with controls. (D–H) Pericyte coverage, as evaluated by confocal colo-
calization (arrows) of Meca32 with NG2 (D and E) or with α-SMA (F),
PDGFR-β (G), or desmin (H) and expressed as percent colocalization
of pericyte markers on tumor ECs. Combined Sema3A and DC101
enhanced pericyte coverage 56% (NG2), 59% (PDGFR-β), and 48%
(desmin) compared with DC101. Sema3A increased pericyte coverage
45% (NG2), 47% (PDGFR-β), and 40% (desmin) compared with con-
trols. DC101 increased α-SMA+ pericyte coverage 46% versus con-
trols. (I) Increased incidence of FITC-lectin–perfused vessels (arrows)
in Sema3A-treated (alone or in combination with DC101) tumors com-
pared with DC101-treated and control insulinomas. (J) Decreased
FITC-dextran extravasation (arrows) in Sema3A-treated (alone or
in combination with DC101) tumors compared with DC101-treated
and control insulinomas. Results are from 5 fields per mouse (n = 8
per treatment group). *P < 0.05, **P < 0.01, unpaired Mann-Whitney
U test. Scale bars: 50 μm.
research article
The Journal of Clinical Investigation http://www.jci.org
how these 2 drugs may induce evasive resistance to angiogenesis
inhibition through different molecular mechanisms. Indeed, simi-
lar to sunitinib, DC101 triggered tumor hypoxia, but, differently
from sunitinib, also co-opted blood vessels, a phenomenon that
has previously been correlated with the development of acquired
resistance to antiangiogenic therapies in RIP-Tag2 mice (4).
In this study, we showed that treating RIP-Tag2 tumors with
sunitinib highly increased NF-κB expression. Since NF-κB acti-
vates HIF-1α and promotes EMT, cancer invasion, and tumor
angiogenesis in several tumor types (25, 26, 42), our data suggest
that NF-κB plays an important role in the development of eva-
sive resistance in response to standard antiangiogenic therapies
and that inhibition of NF-κB expression may represent a further
mechanism by which Sema3A can overcome the side effects caused
by angiogenesis inhibition. It has also been observed that during
progression, tumors recruit proangiogenic myeloid cells that can
contribute to the intrinsic resistance to antiangiogenic therapies
(43). Of note, Gr1+MMP9+ cells, which increase the bioavailability
of VEGF for its receptors (44), and tumor-associated macrophages
(TAMs) expressing cathepsins B and S are critical promoters of
tumor growth, angiogenesis, and invasion in RIP-Tag2 mice (45).
Because NF-κB orchestrates the tissue inflammatory response
induced by hypoxia, including leukocyte infiltration (25, 46, 47),
it is conceivable that, by upregulating NF-κB expression, sunitinib
could induce the recruitment and activation of neutrophils, TAMs,
and other protumoral myeloid cells. Simultaneous administration
of Sema3A together with sunitinib restored tumor tissue normoxia
and reduced NF-κB, and could therefore inhibit the appearance of
these inflammatory cell populations. Further studies are required
to clarify the effect of Sema3A on bone marrow–derived proangio-
genic cells during angiogenesis inhibition.
Inhibition of tumor angiogenesis by sunitinib strongly increased
the expression and tyrosine phosphorylation of Met receptor in
RIP-Tag2 tumors. Met and phospho-Met were present in cancer
cells and, to a lesser extent, in vessels of untreated mice, highlight-
ing the key role played by the Met receptor in tumor angiogenesis
and progression (48–50). Notably, 1 month of sunitinib treatment
strongly increased Met phosphorylation in tumor cells and not in
ECs, suggestive of specific activation of the proinvasive HGF/Met
pathway in cancer cells, but not in the tumor vasculature. The dra-
matic inhibition of Met expression and phosphorylation induced
by Sema3A, alone or in combination with sunitinib, along with
the substantial reduction of tumor spreading and metastatization
indicated that HGF/Met signaling inhibition is an additional key
mechanism by which Sema3A can overcome the evasive resistance
to antiangiogenic therapies.
It is worth noting that treatment of tumor-bearing RIP-Tag2
mice with AVV8-Sema3A as a single agent reduced invasiveness
and metastasis formation by increasing E-cadherin expression
and inhibiting Met TK receptor activation in cancer cells com-
pared with control insulinomas. Even though untreated tumors
displayed a milder hypoxia than did sunitinib-treated insulino-
mas, 2 hypoxia-induced genes, CA9 and NF-κB, were significantly
reduced in Sema3A-treated mice compared with controls. We spec-
ulate that Sema3A, by completely restoring tumor oxygenation
and by inhibiting hypoxia-induced signal pathways in end-stage
RIP-Tag2 tumors, may be responsible for the observed increase of
E-cadherin levels, inhibition of Met activation, and consequent
reduction of tumor invasion and metastatization. However, given
the complexity of the tumor microenvironment (14, 41), comple-
mentary mechanisms may also mediate the effects of Sema3A on
tumor angiogenesis and cancer progression. Further investigation
is hence needed to clarify these aspects.
In conclusion, our studies indicate that Sema3A administration
may represent a new therapeutic approach to inhibit angiogenesis,
while promoting the maturation of the surviving vasculature and
hence avoiding the commonly observed long-lasting tumor hypox-
ia that, if not hindered, can support the lethal dissemination of
cancer cells throughout the body. This pharmacological strategy
may help to better and safely harness the therapeutic potential of
antiangiogenic drugs for the final benefit of oncologic patients.
Methods
Further information can be found in Supplemental Methods.
Transgenic tumor model. The RIP-Tag2 transgenic mouse model has been
previously described (10, 13, 51). RIP-Tag2 mice were generated and main-
tained in the C57BL/6 background (Jackson Laboratory). From 12 weeks
of age, all RIP-Tag2 mice received 50% sugar food (Harlan Teklad) and 5%
sugar water to relieve hypoglycemia induced by the insulin-secreting tumors.
Generation of K14-HPV16 transgenic mice (52) and E2 treatment for cervi-
cal carcinogenesis has been previously reported (30, 31). Briefly, 1-month-
old virgin female transgenic (heterozygous K14-HPV16) were anesthetized,
and continuous-release pellets that deliver E2 at 0.05-mg doses over 60 days
(Innovative Research of America Inc.) were implanted s.c. in the dorsal back
skin. Subsequent pellets were implanted at 3 and 5 months of age. The result-
ing HPV16/E2 mice were maintained in the FVB/n background (The Jackson
Laboratory). Mice were monitored throughout the experiments for compli-
cations caused by the dysplastic nature of their skin or by E2 treatment.
Therapeutic treatments. Tumor- bearing RIP-Tag2 or HPV16/E2 mice were
treated for 4 weeks, from 12 until 16 weeks or from 5 until 6 months of
age, respectively (regression trial). Different regression trials were designed:
(a) 40 mg/kg/d sunitinib l-malate (Axon Medchem BV) was administered
daily by oral gavage (n = 30); (b) 1 mg/mouse rat monoclonal function-
blocking antibodies against VEGFR-2 (DC101), obtained in bulk by affin-
ity purification from the supernatant of a hybridoma culture (DC101)
(Translational Research Laboratory, Catalan Institute of Oncology), was
administered twice weekly i.p. (n = 15), as previously reported (4); (c) 100 μl
Sema3A was injected slowly through the abdominal aorta of RIP-Tag2
mice using a 30-gauge needle (Roboz Surgical Instrument) (n = 30), as
previously described (13), or through the distal portion of the abdomi-
nal aorta just before its bifurcation into the 2 common iliac arteries of
HPV16/E2 mice (n = 12 per group) (see In vivo AAV8 administration); and (d)
Sema3A-injected mice were treated daily by oral gavage with 40 mg/kg/d
sunitinib l-malate or twice weekly with 1 mg/mouse DC101 (n = 15). Con-
trol mice were injected with LacZ (13) and treated with methylcellulose
vehicle daily by oral gavage (n = 30) or with 1 mg/mouse purified rat IgG
(Jackson Immunoresearch) i.p. (n = 15). For the survival trial, 12-week-old
Rip-Tag2 mice were treated with 40 mg/kg/d sunitinib, Sema3A, combined
Sema3A and sunitinib, or LacZ plus vehicle (n = 20 per group), and their
survival was monitored over time.
In vivo AAV8 administration. AAV8-Sema3A was administered in RIP-Tag2
mice as previously described (13). For AAV8-LacZ or AAV8-Sema3A delivery
in HPV16/E2 mice, animals were anesthetized by 1.5% isoflurane anesthesia.
The distal portion of the abdominal aorta just before its bifurcation into
the 2 common iliac arteries was exposed following a displacement of intes-
tine and urinary bladder and isolated from the surrounding fat tissue. 50 μl
recombinant AAV8-Sema3A or AAV8-LacZ virus was injected slowly through
the abdominal aorta, by means of a 31-gauge needle of an insulin syringe.
After injection, homeostasis was performed. The abdomen was then closed
layer-to-layer with 5-0 chromic gut sutures. Animals were subsequently mon-
research article
The Journal of Clinical Investigation http://www.jci.org
itored and allowed to recover 1–2 hours after surgery. Postsurgical analgesia
was achieved by buprenorphine (0.1 mg/kg s.c. every 12 hours for 1 day) and
antibiotic prophylaxis with ampicillin (150 mg/kg s.c.).
Invasion index and metastasis analysis. Paraffin sections of pancreas, liver,
and LN from each group of treatment were serially cut (10 μm) and rehy-
drated through 100% xylene and 100%, 95%, and 70% ethanol before
immersion in 1× PBS. Sections were stained with H&E as previously
described (10). Tissues were visualized with a BX-60 microscope (Olym-
pus) equipped with a color Qicam Fast 1394-digital CCD camera (12 bits;
QImaging Corp.). The invasion index of tumors was determined using
5 H&E-stained sections per animal. Based on gross morphological and
detailed histopathological characteristics, tumor islets were subdivided
into insulinomas with well-defined margins and frequent fibrous capsules
(IT), carcinomas with focal regions of invasion (IC1), or fully invasive car-
cinomas (IC2) (10). The presence or absence of tumor cell dissemination
was first evaluated by H&E staining and confirmed by rabbit anti-SV40
T-antigen (sc-20800, 1:50; Santa Cruz) or HPV16-E7 (250629, 1:100; Abbio-
tec) Abs and by immunof luorescence, according to previously described
protocols. LN and liver metastasis incidence in RIP-Tag2 mice and lung
and liver metastasis incidence in HPV16/E2 transgenic mice was evaluated
by scoring for presence or absence in each animal using 10 sections per
animal. The number of liver metastases was measured as SV40 T-antigen
or HPV16-E7 protein-positive regions in 10 images per mouse per treat-
ment group. To quantify the metastatic volume, we used ImageJ software
to compare the metastatic mass with a spheroid. In each image, we drew
a line corresponding to the width (w) of the metastatic region and a line
corresponding to the length (l), then calculated the LN metastatic volume
as 0.52 × w2 × l, as for a spheroid.
Tumor hypoxia analysis. The amount of tumor hypoxia was determined
2 hours after injection of 60 mg/kg pimonidazole hydrochloride (HP2-
100 Hypoxyprobe Kit-Plus; Natural Pharmacia International Inc.) into
RIP-Tag2 mice (13). The formation of pimonidazole adducts was detected
by immunostaining with Hypoxyprobe-1-Mab1 FITC Ab according to the
manufacturer’s instructions. Immunofluorescence images were captured
and analyzed using a Leica TCS SP2 AOBS confocal laser-scanning micro-
scope (Leica Microsystems) and then evaluated by Image-ProPlus 6.2 soft-
ware (Media Cybernetics). Quantification was done by analyzing at least
5 sections and 5 fields per tumor.
Tumor vessel perfusion and vascular permeability. To detect tumor vessel
perfusion and vascular permeability, mice were injected i.v. with 0.05 mg
FITC-labeled tomato lectin (Lycopersicon esculentum; Vector Laborato-
ries) or 0.05 mg 70-kDa FITC-conjugated dextran (Molecular Probes). After
2 minutes, the animals were euthanized, and the heart was perfused with
saline solution followed by 2% PFA. Lectin and dextran distribution was
visualized by fluorescent confocal microscopy z-sectioning that allowed
for 3D reconstruction of the vascular network.
Confocal microscopy quantifications. To quantify pericyte coverage (NG2,
α-SMA, PDGFR-β, or desmin, shown by red channels) in each image, we
drew a region of interest (ROI) close to each blood vessel (Meca32, shown
by green channels) and then quantified the MFI of red and green channels
using the Leica Confocal Software Histogram Quantification Tool. We
then calculated the ratio between red and green channel MFI; values are
expressed as percentage of red-green costaining. A similar procedure was
followed to quantify vascular perfusion by FITC-labeled lectin. Blood vessel
permeability was analyzed by measuring the area of dextran extravasation.
To determine the expression levels of E-cadherin (green channel), vimentin
(red channel), total Met, and phospho-Met (red channel) in each analyzed
image, we considered 5 random ROIs of the same size. Then we measured
the ratio between the red or green channel and the blue (DAPI) channel MFI;
data are presented as percent positive cells relative to total cell number.
Measurement of in vivo tumor bioavailability of doxorubicin. A group of RIP-
Tag2 mice, previously treated for 4 weeks with LacZ plus vehicle (con-
trol) or with Sema3A, sunitinib, or Sema3A and sunitinib combined, was
injected with 10 mg/kg doxorubicin hydrochloride (Sigma-Aldrich) via the
lateral tail vein 4 hours before sacrifice. Pancreatic tumors and kidneys
as controls were collected from each mouse and weighed. Samples were
resuspended in a lysis buffer (0.25 M sucrose, 5 mM TrisHCl pH 7.6, 1 mM
MgSO4, 1 mM CaCl2) and homogenized in an ice-cold Potter homogenizer.
200 μl of each homogenate was placed into a new microcentrifuge tube,
and 100 μl 10% Triton X-100, 200 μl water, and 1.5 ml acidified isopro-
panol was added. The mixture was vortexed and kept at –20°C overnight.
The next day, samples were warmed to room temperature ad centrifuged at
15,000 g for 20 minutes. Doxorubicin was quantified by spectrophotomet-
ric analysis (λ ex, 470 nm; λ em, 590 nm) using the Synergy HT (BioTek)
plate reader. These values were compared with a standard curve of known
amounts of doxorubicin and normalized based on the weight of the each
organ and on the fluorescence emission of the control tissue, calculated
as the fluorescence/weight ratio of the tumor divided by the fluorescence/
weight ratio of the kidney. Data are mean ± SD of triplicate aliquots from
tumor homogenates expressed as μ equivalents/g tissue of doxorubicin.
Statistics. All values are expressed as mean ± SD. For all statistical analyses,
a 2-tailed, unpaired Mann-Whitney U test was used. Statistical analysis for
the survival trial was performed using the log-rank test. A P value less than
0.05 was considered significant.
Study approval. All animal procedures were approved by the Ethical Com-
mission of the University of Torino and by the Italian Ministry of Health
in compliance with the international laws and policies.
Acknowledgments
We thank Marta Paez-Ribes for helping to evaluate tumor invasive-
ness and metastasis formation and Doug Hanahan for discussion
and insightful suggestions. This work was supported by Associazione
Italiana per la Ricerca sul Cancro (AIRC) investigator grants (5837 to
E. Giraudo, 9211 to G. Serini, and 10133 to F. Bussolino); by AIRC
2010 Special Program in Molecular Clinical Oncology 5% Project no.
9970 (to E. Giraudo and F. Bussolino); by Regione Piemonte Ricerca
Sanitaria Finalizzata 2007, 2008, and 2008 bis (to E. Giraudo, G.
Serini, and F. Bussolino); by Ricerca industriale e sviluppo precom-
petitivo 2006 grants PRESTO and SPLASERBA (to F. Bussolino); by
Associazione Augusto per la Vita (to G. Serini); by Compagnia di San
Paolo — Neuroscience Program Multicentre Projects (to G. Serini); by
Ministero della Salute Programma di Ricerca Finalizzata 2006 and
Programma Straordinario di Ricerca Oncologica 2006 (to E. Giraudo
and F. Bussolino); by Converging Technologies grant PHOENICS (to
E. Giraudo and F. Bussolino); by Fondazione Guido Berlucchi (to E.
Giraudo and G. Serini); by Fondazione Cassa di Risparmio Torino
(CRT) (to E. Giraudo and F. Bussolino); by Fondazione Piemontese
per la Ricerca sul Cancro-ONLUS (Intramural Grant 5% 2008) (to
E. Giraudo and G. Serini); and by Telethon Italy (to G. Serini). F.
Maione was supported by fellowship “26 fellowship — FIRC” granted
by Fondazione Italiana per la Ricerca sul Cancro (FIRC).
Received for publication May 12, 2011, and accepted in revised
form February 22, 2012.
Address correspondence to: Enrico Giraudo, Laboratory of Trans-
genic Mouse Models, IRCC, and Department of Oncological Scienc-
es, University of Torino School of Medicine, Strada Provinciale 142,
Km 3.95, I-10060 Candiolo, Turin, Italy. Phone: 39.011.9933505;
Fax: 39.011.9933524; E-mail: enrico.giraudo@ircc.it.
research article
The Journal of Clinical Investigation http://www.jci.org
1. Folkman J. Angiogenesis. Annu Rev Med. 200 6;
57:1–18.
2. Duda DG, Batchelor TT, Willett CG, Jain RK.
VEGF-targeted cancer therapy strategies: current
progress, hurdles and future prospects. Trends Mol
Med. 2007;13(6):223–230.
3. Rapisarda A, Melillo G. Role of the hypoxic tumor
microenvironment in the resistance to anti-angio-
genic therapies. Drug Resist Updat. 2009;12(3):74–80.
4. Casanovas O, Hicklin DJ, Bergers G, Hanahan D.
Drug resistance by evasion of antiangiogenic tar-
geting of VEGF signaling in late-stage pancreatic
islet tumors. Cancer Cell. 2005;8(4):299–309.
5. Baluk P, Hashizume H, McDonald DM. Cellular
abnormalities of blood vessels as targets in cancer.
Curr Opin Genet Dev. 2005;15(1):102–111.
6. Jain RK. Normalization of tumor vasculature: an
emerging concept in antiangiogenic therapy. Sci-
ence. 2005;307(5706):58–62.
7. Semenza GL. Hypoxia and cancer. Cancer Metastasis
Rev. 2007;26(2):223–224.
8. Sabbah M, et al. Molecular signature and therapeu-
tic perspective of the epithelial–to–mesenchymal
transitions in epithelial cancers. Drug Resist Updat.
2008;11(4–5):123–151.
9. Pennacchietti S, Michieli P, Galluzzo M, Mazzone
M, Giordano S, Comoglio PM. Hypoxia promotes
invasive growth by transcriptional activation of the
met protooncogene. Cancer Cell. 2003;3(4):347–361.
10. Paez-Ribes M, et al. Antiangiogenic therapy elic-
its malignant progression of tumors to increased
local invasion and distant metastasis. Cancer Cell.
2009;15(3):220–231.
11. Ebos JM, Lee CR, Cruz-Munoz W, Bjarnason GA,
Christensen JG, Kerbel RS. Accelerated metastasis
after short-term treatment with a potent inhibitor of
tumor angiogenesis. Cancer Cell. 2009;15(3):232–239.
12. Galluzzo M, Pennacchietti S, Rosano S, Comoglio
PM, Michieli P. Prevention of hypoxia by myoglo-
bin expression in human tumor cells promotes dif-
ferentiation and inhibits metastasis. J Clin Invest.
2009;119(4):865–875.
13. Maione F, et al. Semaphorin 3A is an endogenous
angiogenesis inhibitor that blocks tumor growth and
normalizes tumor vasculature in transgenic mouse
models. J Clin Invest. 2009;119(11):3356–3372.
14. Bergers G, Hanahan D. Modes of resistance to
anti-angiogenic therapy. Nat Rev Cancer. 2008;
8(8):592–603.
15. Christensen JG. A preclinical review of sunitinib,
a multitargeted receptor tyrosine kinase inhibitor
with anti-angiogenic and antitumour activities.
Ann Oncol. 2007;18 suppl 10:x3–x10.
16. Casazza A, et al. Systemic and targeted delivery of
semaphorin 3A inhibits tumor angiogenesis and
progression in mouse tumor models. Arterioscler
Thromb Vasc Biol. 2011;31(4):741–749.
17. Sathornsumetee S, et al. Tumor angiogenic and
hypoxic profiles predict radiographic response and
survival in malignant astrocytoma patients treated
with bevacizumab and irinotecan. J Clin Oncol.
2008;26(2):271–278.
18. Winkler F, et al. Kinetics of vascular normaliza-
tion by VEGFR2 blockade governs brain tumor
response to radiation: role of oxygenation, angio-
poietin-1, and matrix metalloproteinases. Cancer
Cell. 2004;6(6):553–563.
19. Goel S, et al. Normalization of the vasculature for
treatment of cancer and other diseases. Physiol Rev.
2011;91(3):1071–1121.
20. Bergers G, Song S, Meyer-Morse N, Bergsland E,
Hanahan D. Benefits of targeting both pericytes and
endothelial cells in the tumor vasculature with kinase
inhibitors. J Clin Invest. 2003;111(9):1287–1295.
21. Yilmaz M, Christofor i G. EM T, the cy toskeleton,
and cancer cell invasion. Cancer Metastasis Rev.
2009;28(1–2):15–33.
22. Perl AK, Wilgenbus P, Dahl U, Semb H, Christo-
fori G. A causal role for E-cadherin in the transi-
tion from adenoma to carcinoma. Nature. 1998;
392(6672):190–193.
23. McConkey DJ, et al. Role of epithelial–to–mesen-
chymal transition (EMT) in drug sensitivity and
metastasis in bladder cancer. Cancer Metastasis Rev.
2009;28(3–4):335–344.
24. Min C, Eddy SF, Sherr DH, Sonenshein GE. NF-
kappaB and epithelial to mesenchymal transition
of cancer. J Cell Biochem. 2008;104(3):733–744.
25. Sark ar FH, Li Y, Wang Z, Kong D. NF-kappaB
signaling pathway and its therapeutic implica-
tions in human diseases. Int Rev Immunol. 2008;
27(5):293–319.
26. van Uden P, Kenneth NS, Rocha S. R egulation of
hypoxia-inducible factor-1alpha by NF-kappaB.
Biochem J. 2008;412(3):477–484.
27. Cummins EP, et al. Prolyl hydroxylase-1 negatively
regulates IkappaB kinase-beta, giving insight into
hypoxia-induced NFkappaB activity. Proc Natl Acad
Sci U S A. 2006;103(48):18154–18159.
28. Culver C, Sundqvist A, Mudie S, Melvin A, Xirodi-
mas D, Rocha S. Mechanism of hypoxia-induced
NF-kappaB. Mol Cell Biol. 2010;30(20):4901–4921.
29. Smith-McCune K, Zhu YH, Hanahan D, Arbeit J.
Cross-species comparison of angiogenesis during
the premalignant stages of squamous carcinogene-
sis in the human cervix and K14-HPV16 transgenic
mice. Cancer Res. 1997;57(7):1294–1300.
30. Giraudo E, Inoue M, Hanahan D. An amino-
bisphosphonate targets MMP-9-expressing macro-
phages and angiogenesis to impair cervical carcino-
genesis. J Clin Invest. 2004;114(5):623–633.
31. Arbeit JM, Howley PM, Hanahan D. Chronic
estrogen-induced cervical and vaginal squamous
carcinogenesis in human papillomavirus type 16
transgenic mice. Proc Natl Acad Sci U S A. 1996;
93(7):2930–2935.
32. Seeber LM, et al. The role of hypoxia inducible fac-
tor-1alpha in gynecological cancer. Crit Rev Oncol
Hematol. 2011;78(3):173–184.
33. Franco M, Paez-Ribes M, Cor tez E, Casanovas O,
Pietras K. Use of a mouse model of pancreatic neu-
roendocrine tumors to find pericyte biomarkers of
resistance to anti-angiogenic therapy. Horm Metab
Res. 2011;43(12):884–889.
34. Carmeliet P, Jain RK. Principles and mechanisms of
vessel normalization for cancer and other angiogen-
ic diseases. Nat Rev Drug Discov. 2011;10(6):417–427.
35. Goodman VL, et al. Approval summar y: suni-
tinib for the treatment of imatinib refractory or
intolerant gastrointestinal stromal tumors and
advanced renal cell carcinoma. Clin Cancer Res.
2007;13(5):1367–1373.
36. Raymond E, et al . Sunitin ib malat e for t he treat -
ment of pancreatic neuroendocrine tumors. N Engl
J Med. 2011;364(6):501–513.
37. Xian X, et al. Pericytes limit tumor cell metastasis.
J Clin Invest. 2006;116(3):642–651.
38. Sorensen AG, et al. A “vascular normalization
index” as potential mechanistic biomarker to
predict survival after a single dose of cediranib in
recurrent glioblastoma patients. Cancer Res. 2009;
69(13):5296–5300.
39. Rapisarda A, et al. Schedule-dependent inhibi-
tion of hypoxia-inducible factor-1alpha protein
accumulation, angiogenesis, and tumor growth by
topotecan in U251-HRE glioblastoma xenografts.
Cancer Res. 2004;64(19):6845–6848.
40. Vredenburgh JJ, et al. Bevacizumab plus irinotecan
in recurrent glioblastoma multiforme. J Clin Oncol.
2007;25(30):4722–4729.
41. Carmeliet P, Jain RK. Molecular mechanisms
and clinical applications of angiogenesis. Nature.
2011;473(7347):298–307.
42. Nam SY, et al. A hypoxia-dependent upregulation
of hypoxia-inducible factor-1 by nuclear factor-
kappaB promotes gastric tumour growth and
angiogenesis. Br J Cancer. 2011;104(1):166–174.
43. Shojaei F, et al. Bv8 regulates myeloid-cell-
dependent tumour angiogenesis. Nature. 2007;
450(7171):825–831.
44. Nozawa H, Chiu C, Hanahan D. Infiltrating neu-
trophils mediate the initial angiogenic switch in
a mouse model of multistage carcinogenesis. Proc
Natl Acad Sci U S A. 2006;103(33):12493–12498.
45. Gocheva V, et al. IL-4 induces cathepsin protease
activity in tumor-associated macrophages to pro-
mote cancer growth and invasion. Genes Dev. 2010;
24(3):241–255.
46. Fitzpatrick SF, et al. An intact canonical NF-kap-
paB pathway is required for inf lammatory gene
expression in response to hypoxia. J Immunol. 2011;
186(2):1091–1096.
47. Porta C, et al. Cellular and molecular pathways
linking inflammation and cancer. Immunobiology.
2009;214(9–10):761–777.
48. Christensen J, Anderes K. Beyond VEGF: target-
ing tumor growth and angiogenesis via alternative
mechanisms. Adv Exp Med Biol. 2008;610:43–53.
49. Abounader R, Laterra J. Scatter factor/hepatocyte
growth factor in brain tumor growth and angio-
genesis. Neuro Oncol. 2005;7(4):436–451.
50. Bussolino F, et al. Hepatocyte growth factor is a potent
angiogenic factor which stimulates endothelial cell
motility and growth. J Cell Biol. 1992;119(3):629–641.
51. Hanahan D. Heritable formation of pancreatic
beta-cell tumours in transgenic mice expressing
recombinant insulin/simian virus 40 oncogenes.
Nature. 1985;315(6015):115–122.
52. Coussens LM, Hanahan D, Arbeit JM. Genetic pre-
disposition and parameters of malignant progres-
sion in K14-HPV16 transgenic mice. Am J Pathol.
1996;149(6):1899–1917.