Human blood late outgrowth endothelial cells for gene
therapy of cancer: determinants of efficacy
J Wei1, G Jarmy1, J Genuneit, K-M Debatin and C Beltinger
University Children’s Hospital, Ulm, Germany
Human adult blood late outgrowth endothelial cells (BOECs)
are potential yet untested cellular vehicles to target tumor-
cytotoxic effectors to tumors. We show that, following
intravenous injection into irradiated mice, BOECs home to
Lewis lung carcinoma (LLC) lung metastases, but less so to
liver or kidney metastases. BOECs targeted most but not all
of the lung metastases, to a different degree. While most of
the homed BOECs took up an extravascular position, some
integrated into tumor vessels. Sequestration into normal
tissue was low. Placental growth factor mediated both
migration and invasion of BOECs into LLC spheroid masses
in vitro, as did VEGF. When armed with a suicide gene,
BOECs exerted a bystander effect on LLC cells in vitro and
in vivo. Surprisingly, i.v. administration of armed BOECs into
mice bearing multi-organ LLC metastases did not prolong
survival. In addition to homing efficacy other parameters
impacted upon the efficacy of BOECs. These include the
ultimate susceptibility of BOECs to suicide gene-induced cell
death, their paracrine proliferative effect on LLC cells and
their low proliferation rate compared to LLC cells. Addressing
these determinants may make BOECs a useful addition to
the arsenal of tumor-targeting moieties.
Gene Therapy (2007) 14, 344–356. doi:10.1038/sj.gt.3302860;
published online 5 October 2006
Keywords: endothelial progenitor cells; cellular gene therapy; cancer
There is evidence, both experimental and clinical, that
bone marrow-derived cells play an important role in the
vascularization of tumors. The extent to which these cells
contribute to the endothelial, pericytic and proangio-
genic non-structural cells of the tumor vasculature is a
matter of debate.1–22As most tumors depend on building
new vessels for growth, tumor vessel formation may
provide a common therapeutic target shared by different
tumor entities. Therefore, efforts have been made to
exploit the innate tumor tropism of bone marrow-
derived cells, including circulating endothelial progeni-
tor cells, to deliver therapeutic payloads specifically to
tumor sites. Initial studies transplanted genetically
modified bone marrow-derived cells into irradiated
mice, followed by post-transplant tumor challenges.5,7,23
These pioneering studies proved the principle that
genetically engineered cells can home from the bone
marrow to the tumor, where they elicit an anti-tumor
response. Similarly, modified CD34+cells homed to sites
of neovascularization in primates and elicited a bystan-
der effect in vitro.24However, future clinical application
of bone marrow-derived cells for cancer therapy will
most likely consist of systemic injection of ex vivo
expanded and modified cells, including endothelial
progenitors, without bone marrow transplantation. Thus,
emerging data showing that expanded or modified
mouse and, more important, human EPCs home to
experimental tumors, including brain tumors, are en-
couraging.2,17,25–28We investigated the parameters deter-
mining the efficacy of human adult blood outgrowth
endothelial cells (BOECs) for systemic cancer gene
therapy in mice. These cells are attractive candidate
cellular vehicles since they can be readily procured from
autologous donors and are easily expanded and geneti-
cally modified ex vivo.29,30
Homing of EPCs to tumors may constitute a major
parameter determining their therapeutic efficacy. How-
ever, little is known about tumor homing of both
endogenous and ex vivo expanded EPCs. VEGF –
overexpressed in many tumors – which acts on VEGFR2
expressed by tumor endothelial cells and also on
VEGFR1, has been established as a pivotal angiogenic
mechanism in tumors.31,32VEGF administration into
mice and VEGF secreted by patients with vascular
trauma mobilizes endogenous EPCs from the bone
marrow.1,33VEGF-triggered mobilization of EPCs from
the bone marrow is mediated by matrix metalloprotei-
nase-9 induced in bone marrow cells, possibly via
releasing soluble Kit-ligand.34VEGF may also attract
EPCs to tumors since endogenous and early outgrowth
EPCs migrate towards VEGF in vitro.1,23
Placental growth factor (PlGF) acting on VEGFR1 is
emerging as an enhancer of
VEGFR1 is expressed by endothelial cells, monocytes/
macrophages, neutrophils and their precursors.35A
common feature of VEGFR1 expressing cells is homing
to normal and diseased tissue. Thus, chemotaxis of
Received 13 March 2006; revised 1 August 2006; accepted 2 August
2006; published online 5 October 2006
Correspondence: Dr C Beltinger, University Children’s Hospital,
Eythstr. 24, 89075 Ulm, Germany.
1These authors contributed equally to this work
Gene Therapy (2007) 14, 344–356
& 2007 Nature Publishing Group All rights reserved 0969-7128/07 $30.00
monocytes is mediated by VEGFR1 in response to both
PlGF and VEGF.36Furthermore, following transplanta-
tion of wild type bone marrow cells into Id1+/?Id3?/?
mice deficient in tumor angiogenesis, growth and
angiogenesis of tumors was restored and VEGFR1+EPCs
were found within the tumor vessels, a process abro-
gated by blocking VEGFR1.6Tumor angiogenesis was
impaired in PlGF
expression of PlGF.37PlGF overexpressed in Lewis lung
carcinoma cells enhanced tumor growth and angio-
genesis.38PlGF is also known to increase survival of
tumor endothelial cells.39Many tumors, including brain
tumors, overexpress PlGF associated with increased
tumor angiogenesis.40–48Taken together, while these
studies implicate PlGF in tumor angiogenesis, its role
in recruiting EPCs and BOECs is unknown.
In this paper, we elucidate important parameters
mediated cancer gene therapy.
?/?mice and could be restored by
of experimental BOEC-
Growth and phenotype of BOECs
We compared growth and phenotype of BOECs in a total
of 42 cultures derived from PBMCs of 16 healthy donors.
In all cultures cells outgrew after 15–20 days, proliferated
rapidly and could be passaged for up to 25 times. BOECs
expressed VEGFR2, VEGFR1, vWF, CD31,VE-cadherin,
avb3, avb5and CD105. They did not or only marginally
express CD133, c-kit, CD45, CXCR4, ESL (E-selectin
ligand), CD162, CLA, CD49d, CD11a and CD11b
(Supplementary Figure 1) and decreased with culture
time. BOECs took up acetylated LDL and built vessel-
like structures on Matrigel (data not shown).
Taken together, these data confirm that BOECs have
a mature endothelial phenotype and furthermore show
that they express few progenitor- and homing-associated
CD34 expression varied
BOECs target lung metastases and brain tumors
Disseminated metastases cannot be effectively treated
by intratumoral injections. We therefore investigated,
whether systemically administered BOECs home effi-
ciently to disseminated metastases. An early time point
after BOEC injection was chosen for analysis of homing
to coincide with the time point at which the prodrug was
administered. Tail vein injected DiI-labeled BOECs were
found in large numbers in the margins and, to a lesser
degree, in the center of lung metastases, but rarely
homed to liver or kidney metastases (Figure 1a and b
and data not shown). In contrast, i.v. injected human DiI-
labeled fibroblasts did not home to metastases, but were
sequestered in the lung (Figure 1c). Assessment of
homing in irradiated mice showed that 76% of the 97
lung metastases examined recruited BOECs (Figure 1d).
While 17 and 18% of the metastases showed strong
(Figure 1a) or marked (Figure 1e and f) BOEC recruit-
ment, respectively, the majority of recruiting metastases
showed moderate recruitment. Recruitment did not
depend on the size or localization of the lung metastases.
Staining of the tumor vessels showed that the margins
of the metastases, areas that heavily recruited BOECs
(Figure 1a and e), were especially well vascularized
(Figure 1e). Of note, a minority of BOECs was found to
be associated with tumor vessels while the majority was
found extravascularly (Figure 1e and f).
We also wanted to know, whether i.v. injected BOECs
target orthotopically implanted human brain tumors.
BOECs were labeled with BrdU, since the autofluores-
cence of brain tumors precluded the efficient use of DiI-
labeled BOECs. Following systemic administration of
BrdU-labeled BOECs numerous BOECs were seen in U87
gliomas, few BOECs were found in LNT-229 and SF126
glioma, and no BOECs were detected in normal brain
tissue (Supplementary Figure 2).
Taken together, after intravenous injection BOECs
target the extravascular region of most but not all LLC
lung metastases with moderate efficacy. Kidney or liver
metastases are not targeted efficiently and brain tumors
are targeted differentially.
BOECs are sequestered only transiently and at low
density in lung and spleen
Sparing normal tissue is a prerequisite for a candidate
cellular vehicle for cancer gene therapy. We therefore
assesssed the distribution and fate of BOECs injected into
the tail vein of tumor-free mice. To this end, BOECs were
similiar to111In oxine loading described previously.49The
label was stably retained in the BOECs without leakage
into the culture medium during the observation period
of 48 h and neither morphology nor proliferation was
influenced by the radioactive label (data not shown).
Except for the liver, where cell-free111In-DTPA Ac-LDL
111In-DTPA fragments, radioactive label recovered in
organs following injection of111In-DTPA Ac-LDL loaded
BOECs represents homed BOECs.
Sequestration of BOECs occurred in the lung and the
spleen (Figure 2a). No BOECs were detected in heart,
gut, brain and blood, and only few in the bone marrow.
The radioactivity observed in the liver and bladder most
likely represents intermediates of
metabolism and excreted111In-DTPA fragments, respec-
tively. Organ-bound radioactivity rapidly decreased far
beyond the degree attributable to radioactive decay,
consistent with rapid destruction of the sequestered
BOECs and subsequent excretion of the cell-free label.
We also determined sequestration of BOECs by
immunohistochemistry. Few BOECs were detected in
the lung (Figure 2b) and spleen, while none were found
in liver, kidneys and brain.
In summary, sequestration of human BOECs injected
into the tail vein of non-irradiated mice is confined to the
lung and spleen, where it occurs at low degree and is
111In-DTPA labeled Ac-LDL, an approach
PlGF and VEGF increase migration and invasiveness
The molecular mechanisms attracting EPCs or BOECs to
tumors are unknown. As VEGF and, less so, PlGF have
been suggested to guide recruitment of endogenous bone
marrow-derived EPCs to tumors, we investigated the
response of BOECs to these growth factors. Indeed,
BOECs were attracted in a dose-dependent fashion by
PlGF (Figure 3a) and migrated along a VEGF gradient
BOEC cancer gene therapy
J Wei et al
n = 97
injected into the tail vein of immunodeficient Rag2?/?cgc?/?mice. After metastases had formed, the mice received either no irradiation
(a–c) or 6 Gy of total body g-irradiation (d–e). The next day, DiI-labeled BOECs were injected into the tail vein, this was repeated 2 days later.
As a control, the same sequence was employed with DiI-labeled fibroblasts. (a, b) Tail vein injected BOECs home to lung but not to liver
metastases. DiI-labeled BOECs were injected i.v. twice spaced 2 days apart, organs were procured 4 days later, cryosectioned and
counterstained with Hoechst 33258 (blue). Numerous BOECs (red) are seen around and within a lung metastasis (a) while only very few
BOECs are associated with a liver metastasis (b). Scale bars equal 200 mm. (c) Fibroblasts are sequestered in the lung but do not home to
metastases. At 2 days after the last fibroblast injection lungs were procured, cryosectioned and stained with Hoechst 33258 (blue). Fibroblasts
(red) are seen in the lung but not in the metastasis. Scale bar equals 100 mm. (d) Most lung metastases are targeted by tail vein injected BOECs.
Homing efficacy of i.v. injected BOECs to established lung metastases was analyzed in five irradiated mice. The presence of less than five
BOECs in a metastasis was considered as lack of homing and graded ‘?’. The presence of at least five BOECs was defined to constitute
homing and graded as moderate (‘+’), marked (‘++’, as seen in e and f) and strong (‘+++’, as seen in a). Shown are the percentages of
metastases (n¼97) with homing of a given grade. (e, f) Some BOECs colocalize with tumor vessels, while the majority of BOECs are found
extravascularly. Cryosections were stained with isolectin B4-FITC to decorate tumor vessels (green). Nuclei were stained with Hoechst 33258
(blue). The majority of BOECs (red) were found within the tumor proper while few BOECs (indicated by arrows) colocalized with tumor
vessels. (f) Is a higher magnification of the rectangle in (e). Scale bars equal 200 mm (e) and 50 mm (f).
Intravenously administered BOECs target lung but not liver metastases, with limited contribution to tumor vessels. LLC cells were
BOEC cancer gene therapy
J Wei et al
(Figure 3b). Interestingly, very high concentrations of
VEGF did not increase migration.
BOECs have to invade the tumor mass, if they are to
exert a cytotoxic effect on tumor cells in addition to
tumor endothelial cells. We therefore investigated inva-
sion of BOECs using spheroids of LLC cells, which are
known to express VEGF38and of LLC cells forced to
express PlGF. In Figure 3c (upper panel) the space
occupied by a single spheroid is surrounded by DiI-
labeled BOECs that were added to the medium 1 h prior.
At 48 h the BOECs have migrated to the spheroid with
some of the BOECs invading it (visible as a ring-like
structure; Figure 3c, lower panel). Invasion was much
more pronounced in LLC cells overexpressing PlGF.
Next we determined, whether a chimeric protein of
VEGFR1 and IgG1 Fc blocks invasion of BOECs into
spheroids of LLC cells overexpressing PlGF. In medium
controls and in the presence of IgG1Fc, BOECs migrated
to and invaded the spheroid (Figure 3d, left and
middle panels, respectively). In contrast, VEGFR1/IgG1
Fc chimeric protein attenuated invasion (Figure 3d,
As VEGFR1 of the chimeric protein binds to PlGF and
VEGF, this shows that PlGF and, possibly, VEGF enhance
invasion of BOECs.
BOECs armed with yeast cytosine deaminase exert
a strong bystander effect on LLC cells in vitro
We opted for a suicide gene to arm the BOECs, since this
allows to delay cytotoxicity until the BOECs have
incorporated into the tumor. We chose yeast cytosine
deaminase fused to uracil phosphoribosyl transferase,
because this fusion gene (FCU1) efficiently converts
the non-toxic prodrug 5-fluorocytosin (5-FC) to 5-fluoro-
uracil (5-FU),51which exerts a strong bystander effect.52
As differential susceptibility to 5-FU impacts upon the
bystander effect, we first determined the sensitivity of
BOECs and tumor cells to 5-FU. BOECs were less
sensitive to 5-FU in clinically relevant concentrations
than LLC cells and a random selection of glioblastoma,
neuroblastoma, Ewing sarcoma and rhabdomyosarcoma
cells (Figure 4a). This contributed to a pronounced
bystander effect of BOEC/FCU1 on LLC cells following
5-FC administration: a fraction of o6.25% BOEC/FCU1
cells sufficed to elicit near-complete killing of surround-
ing LLC cells (Figure 4b).
Systemic BOEC/FCU1 do not increase survival
of mice with multi-organ metastases
To investigate the efficacy of BOEC/FCU1 cells on
metastases disseminated in multiple organs, we injected
LLC cells into the tail vein of Rag2?/?cgc?/?mice. Lung
metastases became visible on day 8 after tumor cell
injection (data not shown). To decrease competition
between endogenous EPCs from the bone marrow and
therapeutic BOECs, mice received 3 Gy of total body
g-irradiation. The sequence of therapy and its results are
shown in Figure 5a. No unspecific toxicity of BOEC/
FCU1 cells, 5-FC or irradiation was noted since no deaths
occurred in the tumor-free toxicity control groups
Irr-BOEC/FCU1-5FC and Irr-BOEC/FCU1-PBS. Surpris-
ingly, mice treated with BOEC/FCU1 cells and 5-FC
(with or without irradiation, groups LLC-Irr-BOEC/
FCU1-5FC or LLC-BOEC/FCU1-5FC, respectively) did
not survive longer than control mice (groups LLC-Irr-
BOEC/FCU1-PBS, LLC-Irr-5FC and LLC-Irr-PBS). Upon
necropsy, metastases were found in lung, liver and, less
pronounced, in the kidneys of both the therapy and the
control groups (Figure 5b).
Proliferation, cell death and paracrine effect of
BOEC/FCU1 cells codetermine their efficacy
BOEC/FCU1 cells and 5-FC did not decrease the number
or size of the lung metastases (Figure 5b). We therefore
sought to define the mechanisms determining this
outcome. Following i.v. injection, BOEC/FCU1 cells need
time to home and incorporate into the tumor before 5-FC
can be given. However, an interim period of just one day
sufficed to markedly decrease the proportion of BOEC/
EGFP cells compared to LLC cells (Figure 6a). Thus,
BOECs proliferate less than LLC cells. This places them
at a disadvantage to exert a bystander effect as the
proportion of BOEC/FCU1 cells within the tumor will
rapidly diminish with time.
Next, we investigated the duration of the bystander
effect. Supernatant of BOEC/FCU1 cells and control
BOECs treated with 5-FC was collected daily, applied to
LLC cells and the viability of LLC was determined. A
bystander effect was evident with supernatant from the
first 24 h of treatment with 250 mM 5-FC, decreased with
of BOECs injected i.v. into tumor-free non-irradiated mice.
(a) Sequestration of BOECs is transient. BOECs were loaded with
111In-DTPA labeled Ac-LDL. Radiolabeled cells (4?105) were injec-
ted into the tail vein of 8- to 10-week-old male Rag2?/?cgc?/?mice.
Organs were procured 2 and 5 days later, weighted and analyzed in
a g-counter. Results are expressed as counts per minute per mg
tissue of each organ to correct for the different size of the organs.
(b) Sequestration of BOECs is low-level. BOECs (4?105) were
injected into the tail vein of 8- to 10-week-old male Rag2?/?cgc?/?
mice. At 2 days after BOEC injection solid organs were procured,
cryosectioned and analyzed by immunohistochemistry for human
vWF to detect BOECs. Arrows point to BOECs (the bar represents
200 mm). The insert shows an area of the same lung at higher
Transient and low-level sequestration into normal tissue
BOEC cancer gene therapy
J Wei et al
supernatant collected on days 2 and 3 and dropped
markedly with supernatant from day 4 (Figure 6b). Thus,
the duration of significant 5-FU production by BOEC/
FCU1 was limited to 3 day either by metabolic impair-
ment of the BOEC/FCU1 cells or their eventual death.
Indeed, BOEC/FCU1 cells started to die in response to
high concentrations of 5-FC on 3 day (Figure 6c) and to
lower concentrations thereafter (data not shown).
IgG1FcVEGFR1/IgG1Fc Medium only
BOEC cancer gene therapy
J Wei et al
Toaddress, whetherBOECsexerta paracrine
effect, supernatant of BOECs was added to LLC
target cells. Indeed, the target cells responded with a
significant increase in cell proliferation, which was
not seen when supernatant from LLC cells was used
To evaluate the bystander effect of BOEC/FCU1
cells on LLC tumors in vivo, mixtures of 25% BOEC/
FCU1 and 75% LLC cells were implanted subcutanou-
sly into Rag2?/?cgc?/?mice and 5-FC was adminis-
tered once tumors were palpable. Tumor growth was
significantly retarded initially (Figure 6e). However,
tumor growth resumed.
tumor growth inhibition suggests that the combined
effects of low proliferation, short duration of the
bystander effect and the presence of a paracrine effect
compromise the ability of BOEC/FCU1 cells to control
Targeting of experimental cancers by EPCs is just
beginning to emerge. PBMCs are a ubiquitous source of
EPCs. Unlike early outgrowth EPCs, late outgrowth
BOECs have a large replicative capacity and are easy
to modify, thus making them good candidates for
cellular cancer gene therapy. In contrast to their strong
bystander effect in vitro and their initial inhibitory
effect on admixed tumor cells in vivo, armed BOECs
were not effective when administered systemically in
the presence of disseminated metastases. We have
delineated important determinants of the efficacy of late
BOECs for systemic gene therapy of disseminated
Phenotype of BOECs and progenitor status
Analysis of BOECs derived from a large number of
donors showed the cells to be CD34+CD133?c-kit?
mature endothelial phenotype of BOECs. Of note, the
BOECs did neither express CD133 nor c-kit and expres-
sion of CD34 showed marked interdonor variation. Thus,
BOECs lack or variably express markers associated with
progenitors. Conversely, their large proliferative capacity
characterizes them as progenitor cells. It remains to be
+CD105+lin?. This confirmed the
Ratio of BOEC/FCU1 to LLC cells (%)
Viable cells (× 105, per well)
Viable cells (% of control)
leads to pronounced bystander effect of BOECs armed with yeast
cytosine deaminase fused to uracil phosphoribosyl transferase
(FCU1). (a) Cancer cells are more susceptible to 5-FU than BOECs.
Cells (2?104) per well in six-well plates were treated for 3 days
with 5-FU in concentrations indicated. Viable (propidium iodide
negative) cells were determined by flow cytometry and cell
counting and are expressed as the percentage of untreated controls.
Shown are the means and s.d. of triplicates. Similar results were
obtained in two independent experiments. (b) BOECs armed with
FCU1 (BOEC/FCU1) exert a strong bystander effect on LLC cells.
BOEC/FCU1 cells were mixed with LLC cells in proportions
indicated. Cells (3?103) per well were plated in 24-well plates in
ECBM2 medium containing 2% FBS and treated for 3 days with
250 mM 5-FC or were left untreated. Viable cells were determined by
flow cytometry and cell counting and are expressed as the number
of viable cells per well. Shown are the means and standard
deviations of triplicates. Similar results were obtained in two
Differential sensitivity of BOECs and cancer cells to 5-FU
into the upper well of a modified Boyden chamber in ECBM2 medium with 2% BSA. PlGF was added to the lower well in increasing
concentrations. Migration to the underside of the transwell inserts was quantified after 3 h. Results are expressed as the increase of migration
compared to medium control (i.e., PlGF 0 ng/ml). Shown are the means and s.d. of triplicates. Asterisks denote Po0.01 when compared to
medium control. Similar results were obtained in three independent experiments. (b) VEGF attracts BOECs. VEGF in increasing
concentrations was added to the lower compartment of a modified Boyden chamber and migration was determined and expressed as in (a).
Asterisks denote Po0.01 when compared to medium control. Similar results were obtained in three independent experiments. (c) Forced
expression of PlGF in LLC spheroids increases invasiveness of BOECs. LLC cells overexpressing PlGF (LLC-PlGF) or the parental cells were
placed into 1% agar-coated 96-well plates to form spheroids. DiI-labeled BOECs in single cell suspension were added to the wells containing a
tumor spheroid and invasion was documented after 1 and 48 h. Note the accumulation of BOECs in the vicinity of the spheroids denoting
migration and the ring-like structures within the spheroids associated with invasion. Similar results were obtained in three independent
experiments. (d) VEGFR1/IgG1Fc chimera attenuate invasion of BOECs into LLC spheroids overexpressing PlGF. Single LLC-PlGF spheroids
were generated in agar-coated wells and incubated in medium without additives, with IgG1Fc fragments (as control) or with recombinant
human VEGFR1/IgG1Fc chimera (upper row, light microscopy). DiI-labeled BOECs were then added in single cell suspension and invasion
investigated by fluorescence microscopy after 1 h (second row) and 24 h (third row: light microscopy; fourth row: fluorescence microscopy).
Note that with VEGFR1/IgG1Fc BOECs remain dispersed around the spheroid and invade it less compared to medium only and to IgG1Fc
controls. In an independent experiment cryosections of spheroids stained with Hoechst 33258 (blue) also show less invasion by BOECs (red)
after 48 h when VEGFR1/IgG1FC is present. Bars denote 200 mm. Similar results were obtained in two independent experiments.
BOECs respond to PlGF and VEGF by migration and invasion. (a) BOECs migrate towards a source of PlGF. BOECs were placed
BOEC cancer gene therapy
J Wei et al
determined, whether the lack of progenitor status, as
defined by lack of surface markers, contributes to
the limited capability of BOECs to participate in the
formation of new tumor vessels.
Regional homing of BOECs
Homing is an important determinant of the anti-tumor
efficacy of BOECs. However, little is known about the
homing mechanisms of BOECs. We utilized a multi-
organ, multiple-metastases mouse model to compare the
efficiency of EPC homing to various metastases within
one organ and to metastases in different organs.
We show that metastases or tumors within a given
organ specifically albeit differentially recruit i.v. injected
BOECs. Homing presumably sufficient for a significant
bystander effect occurred in just 35% of the lung
metastases. While U87 gliomas avidly recruited i.v.
injected BOECs, recruitment by LNT-229 and SF126
gliomas residing in the same brain area was modest
compared to U87 gliomas. The causes of this differential
recruitment of BOECs are yet to be uncovered. These
may include competition of large numbers of metastases
for a limited number of BOECs, differential secretion of
PlGF, VEGF or other chemotactic stimuli, differences
in the degree of vascularization and of hypoxia or
divergence of vascular addresses. Irrespective of its
cause, differential and therefore insufficient homing
most likely presents a major reason why i.v. injected
BOEC/FCU1 failed to prolong survival of metastases-
bearing mice. Increasing dose and frequency of BOEC/
FCU1 administration and altering its timing may
increase the efficacy of homing.
We further show that i.v. injected BOECs home to lung
metastases but infrequently to metastases in the liver and
kidneys. This contributed to the failure of i.v. injected
BOEC/FCU1 cells to prolong the life of mice bearing
metastases in these organs. A first pass effect in the lung
of i.v. injected BOECs, organ-specific differences of tumor
vessel architecture and vascular addresses (reviewed by
Pasqualini et al.53) and of chemotactic factors most likely
Survival time (days)
0 1020 30 40
vein injected BOEC/FCU1 cells. (a) BOEC/FCU1 cells injected into the tail vein of mice bearing multi-organ metastases do not prolong life.
Lewis lung carcinoma cells were injected on day 0 into the tail vein of Rag2?/?cgc?/?mice to generate the treatment group LLC-Irr-BOEC/
FCU1-5FC. On day 7, after metastases were established, mice received 3 Gy of total body g-irradiation (jagged arrow) followed by i.v. injection
of 4?105BOEC/FCU1 cells (arrow head). After 2 days, mice received 6.25 mg 5-FC i.p. twice daily for 5 days (arrows). A second cycle of
BOEC/FCU1 cells and 5-FC followed. The LLC-BOEC/FCU1-5FC treatment group was not irradiated. The control group LLC-Irr-BOEC/
FCU1-PBS had LLC metastases, received total body irradiation and BOEC/FCU1 cells, followed by i.p. PBS injections. Additional control
groups bearing LLC metastases received irradiation and either 5-FC (LLC-Irr-5FC) or PBS (LLC-Irr-PBS). Tumor-free mice that received total
body irradiation and BOEC/FCU1 cells followed by either 5-FC (Irr-BOEC/FCU1-5FC) or PBS injections (Irr-BOEC/FCU1-PBS, n¼10) were
used as toxicity controls. Unless stated otherwise, groups consisted of 12 mice. Survival data are depicted as Kaplan–Meier estimator analysis.
Kruskal–Wallis and Mann–Whitney tests showed P40.05 for all comparisons. (b) Liver and kidney metastases, which are not well targeted
by tail vein injected BOECs, contribute to mortality. Lung, liver and kidneys of mice from the same experiment described above were
procured and inspected for visible metastases. Note massive metastases in liver and smaller metastases in the kidneys.
Systemically administered BOEC/FCU1 cells do not impact on survival of mice bearing metastases that are beyond the reach of tail
BOEC cancer gene therapy
J Wei et al
0 5001000 1500 2000
Viable cells (% of control)
At time of
After 1 d
Ratio BOEC/EGFP to
Time after start of 5-FC (d)
Viable cells (% of control)
Viable cells (fold increase)
Tumor volume (mm3)
Treatment duration (d)
decline of the ratio of BOEC/EGFP cells to LLC cells. Mixtures of BOEC/EGFP cells and LLC cells in ratios indicated were plated in 24-well
plates at a density of 3?103per well. After 1 day of culture the ratio of BOEC/EGFP cells to LLC cells was determined by flow cytometry.
(b) The duration of the bystander effect mediated by BOEC/FCU1 cells is short. BOEC/FCU1 cells (3?103) or BOECs were plated into
24-well plates. For the following 4 days medium was replaced daily by fresh medium with 250 mM 5-FC. The collected medium was
centrifuged and transferred to 24-well plates containing 1?104LLC cells per well. After 3 days, LLC cells were harvested and the number of
viable (propidium iodide negative) cells determined by flow cytometry and cell counting. Results are expressed as percentage of viable cells
compared to control, that is, LLC cells incubated with supernatant from BOECs. Shown are the means and s.d. of triplicates. *Po0.01 when
compared to day 1. The experiment was repeated once with similar results. (c) BOEC/FCU1 cells are killed after administration of 5-FC.
3?103BOEC/FCU1 cells or BOEC/Mock cells, that is, BOECs transfected with empty vector, were plated into 24-well plates and treated with
5-FC in doses indicated for 3 days. Viable (propidium iodide negative) cells were determined by flow cytometry and cell counting and are
expressed as the percentage of untreated controls. Shown are the means and s.d. of triplicates. Similar results were obtained in two
independent experiments. (d) BOEC/FCU1 cells exert a paracrine proliferative effect on LLC cells. BOECs or LLC cells (3?103) were plated
into 24-well plates. The next day medium was replaced by supplemented DMEM. After 24 h the supernatant was transferred to 24-well plates
containing 1?104LLC cells per well as targets. Fresh DMEM was used for controls. After 3 days, the number of viable LLC cells was
determined by counting propidium iodide negative cells. Shown is the fold increase of viable LLC cells in relation to LLC cells incubated with
fresh DMEM. Means and s.d. of triplicates are depicted. *Po0.05 and **P40.05 compared to cells incubated with fresh DMEM. (e) Transient
bystander effect of BOEC/FCU1 cells in vivo. LLC cells (1?106) or a mixture of 1?106LLC cells with 3.5?105BOEC/FCU1 cells were
resuspended in Matrigel and injected s.c. into Rag2?/?cgc?/?mice. After 3 days, when tumors were palpable, mice (n¼7 or 8 per group)
received i.p. injections twice daily for 6 days of either 6.25 mg 5-FC (LLC-BOEC/FCU1-5FC group and LLC-5FC group) or PBS (LLC-BOEC/
FCU1-PBS group and LLC-PBS group). Means of tumor volumes are shown. s.d. overlap and are not shown for clarity. Asterisks denote
Parameters impacting on the efficacy of BOEC/FCU1. (a) BOEC/EGFP cells proliferate less than LLC cells, leading to a rapid
BOEC cancer gene therapy
J Wei et al
constitute organ-specific determinants of BOEC homing.
Increasing the number of BOEC/FCU1 cells injected and
administering them by the intraarterial or portal vein
routes may enhance homing to kidney and hepatic
tumors, respectively. Also, sorting the heterogenous
(Supplementary Figure 1) BOECs to achieve homoge-
nous populations may generate BOECs with preferential
homing to tumors in specific organs.
Only few BOECs injected into the tail vein of tumor-
free non-irradiated mice were transiently sequestered in
lung and spleen. Importantly, BOECs were not seques-
tered significantly in the bone marrow, the liver or the
kidneys. Thus, no side effects of armed BOECs should
occur in these organs. Irradiation may increase seques-
tration into normal tissue. However, no side effects
occurred in irradiated mice receiving 5-FC following
systemic administration of BOEC/FCU1: neither survi-
val, nor weight loss or appearance of the treated mice
was different from control mice.
Molecular mechanisms of BOEC homing
BOECs administered into the blood stream have first to
be attracted to the tumor. We have shown that BOECs
migrate along a PlGF gradient and are attracted by tumor
spheroids overexpressing PlGF. PlGF binds to VEGFR1,
which is strongly expressed by BOECs. As many tumors
overexpress PlGF, interaction of PlGF with VEGFR1 may
be important for homing of BOECs in vivo. In addition,
we have shown that BOECs, which strongly express
VEGFR2, are attracted by VEGF in vitro. As many tumors
overexpress VEGF, the VEGF/VEGFR2 axis may be a
prominent mediator of BOEC homing in vivo. In contrast,
BOECs do not express CXCR4 (our data and Rumpold
et al.54) and thus cannot respond to SDF-1, a potent attractor
of hematopoietic stem cells. The role of hepatocyte
growth factor/scatter factor, which is secreted by many
tumors, in homing of BOECs warrants further study,
since it has been implicated as a chemoattractant for
endothelial progenitor cells.55
Few BOECs were incorporated into tumor vessels
when analyzed 2 days after BOEC injection. Longer
follow-up is required to accurately assess the degree of
definitive incorporation. Furthermore, since BOECs may
not have an endogenous counterpart, this finding cannot
be extrapolated to endogenous EPCs. Nevertheless, it is
interesting to note that most bone marrow transplant
studies7,9–11,14,19–23,56and studies that systemically in-
fused endothelial precursors2,23,25,27report a low degree
of EPC incorportation. Thus, notwithstanding tumor
type, tumor site and mouse strain-dependent variations
of BOEC recruitment,22amplification of tumor cell death
by obliteration of tumor vessels may not be prominent
when using therapeutic BOECs. However, once resident
in the tumor tissue proper, BOECs will impact on tumor
cells via the bystander effect, similar to what has been
described for therapeutic cells recruited from trans-
planted bone marrow.7,23
BOEC-FCU1 cells were generated by transduction and
positive selection for 5 days. No change in morphology,
proliferation and expression of VEGFR2, VE-cadherin or
CD34 was noted after this short period of selection (data
not shown). Thus, it is unlikely that transduction and
selection impacted on homing and thus on therapeutic
efficacy of the BOECs.
We employed myelosuppressive irradiation in our
studies under the assumption, that it decreases competi-
tion of therapeutic BOECs with endogenous EPCs. This
may have had an effect on homing yet to be character-
ized, since irradiation has been shown to increase
angiogenesis by induction of VEGF,57eNOS,58MMP-959
and CXCL12 (SDF-1).26In addition, irradiation-induced
tumor necrosis may mechanically trap BOECS. However,
significant homing was also observed in the absence
of irradiation. Furthermore, BOECs just trapped in
an irradiated tumor may nevertheless contribute to
The bystander effect of BOECs
BOECs are outpaced by the highly proliferative LLC cells
leading to a substantial decrease in the ratio of BOECs to
LLCs in vitro within a short period of time. We assume
that this also took place in vivo, thus decreasing the
bystander effect and contributing to the failure of BOEC/
FCU1 to control the disseminated LLC metastases.
Human tumors grow slower than rodent tumors. Thus,
an increased bystander effect can be expected when
human tumors are targeted.
BOECs appear to be promising vehicles for tumor-
cytotoxic genes by virtue of their relative resistance to
cell death. This resistance is demonstrated by the
survival of a substantial minority of BOEC/FCU1 cells
following administration of 5-FC, due to BOECs being
less susceptible to 5-FU than tumor cells. Despite this
resistance, the duration of maximal action by suicide
genes on bystander cells is limited, as we have shown
for FCU1. While generating longer-living BOECs, for
example by transferring antiapoptotic genes, would
increase the bystander effect, this would also increase
the duration of the paracrine proliferative effect on
tumor cells and the – remote – possibility of their
malignant transformation. Thus, cytotoxic effectors other
than suicide genes may constitute better therapeutic
payloads for BOECs. A very intriguing possibility to
increase the bystander effect while rendering payload-
induced cell death of BOECs irrelevant is to employ them
as vehicles for replicating oncolytic viruses.
We have delineated important determinants of the
efficacy of BOEC-mediated tumor gene therapy. First,
tumors overexpressing PlGF or VEGF are candidate
targets for BOECs. Second, the degree and homogeneity
of homing leaves room for improvement. Third, cyto-
toxic effectors not affected by BOEC survival, prolifera-
tion arrest and paracrine tumor proliferative effect
BOECs. Addressing these determinants may allow for
the improvement of BOECs to efficiently target tumors.
Materials and methods
BOECs were isolated and grown as previously de-
gradient centrifugation with Biocoll separating solution
(Biochrom, Berlin, Germany) from healthy volunteers.
BOEC cancer gene therapy
J Wei et al
Heidelberg, Germany), supplemented with Supplement-
Pack (PromoCell) and with 10% FBS (Biochrom) and
seeded into collagen I-coated culture vessels (BD
Biosciences, Heidelberg, Germany). Medium was chan-
ged every day until BOECs appeared between days 15
and 20. Lewis lung carcinoma (LLC) cells, LLC-PlGF
cells and U87, LNT-229 and SF126 human glioma cells
were grown in DMEM with sodium pyruvate (Life
Technologies, Karlsruhe, Germany) supplemented with
10% FBS (Biochrom). LAN-5 human neuroblastoma, U87
human glioblastoma, TC-71 human Ewing Sarcoma and
RD human rhabdomyosarcoma cell lines were grown in
RPMI 1640 (Life Technologies) supplemented with 10%
FBS (Biochrom). Human foreskin fibroblasts were main-
tained in Ham’s F-12 medium (PAA Laboratories,
Pasching, Germany). DMEM, RPMI and Ham’s F-12
media were supplemented with 2 mM glutamine (Bio-
chrom), 100 U/ml penicillin and 100 mg/ml streptomycin
(both from Life Technologies). All cells were cultured in
21% O2, 5% CO2and at 371C.
Characterization of BOECs
Cells were incubated with phycoerythrin (PE)-conju-
gated monoclonal antibodies against CD34, CD133,
CD117 (c-kit) and integrin avb3(all from BD Biosciences
Pharmingen), with FITC-conjugated monoclonal antibo-
dies against CD14, CD45, CD105 (all from Immunotech,
Hamburg, Germany), CLA (cutaneous lymphocyte-asso-
ciated Ag; BD Biosciences Pharmingen), VE-cadherin
(Bender MedSystems, Vienna, Austria) and with uncon-
jugated antibodies against VEGFR2 (Sigma, Munich,
Germany) and integrin avb5 (BD Biosciences Pharmin-
gen) followed by FITC- or PE-conjugated secondary
antibodies (BD Biosciences). All incubations were per-
formed for 30 min at 41C. Isotype-matched antibodies
(BD Biosciences Pharmingen) served as controls. Ana-
lyses were performed using a FACScan flow cytometer
and CellQuest software (BD Biosciences). For immuno-
fluorescence microscopy, cells were fixed with zinc,
incubated at room temperature with anti-VEGFR1
(10 mg/ml; ReliaTech, Braunschweig, Germany) or anti-
vWF (1:20; Dako, Glostrup, Denmark) or CD31 (1:20;
Dako) followed by Cy3-conjugated goat anti-mouse IgG
(Jackson ImmunoResearch Laboratories, West Grove,
USA) at a dilution of 1:700 for 30 min. Nuclei were
stained by Hoechst 33258 (10 ng/ml; Molecular Probes,
Modified Boyden chambers were used consisting of
Transwell membrane filter inserts (Corning Costar Corp.,
Cambridge, MA, USA) with 8.0 mm pore size and coated
with 10 mg/ml fibronectin (Sigma). BOECs (1?105) in
100 ml ECBM2 containing 2% BSA (Serva, Heidelberg,
Germany) were added to the upper inserts, which were
set into wells containing 600 ml ECBM2/2% BSA and
recombinant murine VEGF or recombinant human
PlGF-2 (both from ReliaTech) in concentrations of 0, 0.1,
1, 10, 50 and 100 ng/ml. After 3 h the inserts were
removed and fixed with 2% paraformaldehyde (PFA).
The upper side of the membrane was scraped to remove
cells. The transmigrated cells on the lower side of the
transwell were stained by Hoechst 33258 and counted in
five random high-power-fields using a fluorescence
microscope (AX 70, Olympus, Hamburg, Germany).
Spheroid invasion assay
Ninety-six-well plates were coated with 1% agar (Sigma)
in supplemented DMEM. 1?103LLC cells or LLC-PlGF
cells were seeded in each well of a 96-well agar-coated
plate in 100 ml supplemented DMEM for two days to
form spheroids. These single spheroids were incubated
for 18 h with 37 mg/ml recombinant human VEGFR1/
IgG1Fc chimera or 37 mg/ml human IgG1Fc (both from
R&D Systems, Minneapolis, MN, USA) or supplemented
DMEM. BOECs (2?103) labeled with DiI (1,10-dioctade-
Molecular Probes) were added per well. Invasion of
BOECs into the spheroids was documented at 1, 24
and 48 h. For some experiments tumor spheroids were
picked out, fixed with 2% PFA, embedded in OCT
compound (Sakura Finetek, Loeterwoude, The Nether-
lands) at ?801C, cut into 5 mm thick sections and stained
with Hoechst 33258. Invading DiI-labeled BOECs were
analyzed by fluorescence microscopy (CK40, Olympus).
Gene transfer to BOECs
The retroviral construct pL(FCU1)IN was generated by
inserting the cytosine deaminase/uracil phosphoribosyl-
transferase fusion gene (FCU1) into the pLXIN vector
(BD Biosciences Clontech). The FCU1 cDNA was excised
Strasbourg, France) as a XhoI/SmaI fragment and cloned
into the multiple cloning site of the pLXIN vector. The
vectors pLXIN, pL(FCU1)IN and pLEIN (which contains
the enhanced green fluorescent protein, EGFP) were
stably transfected into GP2-293 packaging cells (BD
Biosciences Clontech), using the calcium phosphate
precipitation method. These cells were then transiently
transfected with 7.5 mg of pVSV-G (BD Biosciences
Clontech), employing the same transfection method.
Virus-containing supernatants were harvested 48 h later,
filtered through a 0.45 mm filter (Nalgene, Neerijse,
Belgium) and used to transduce BOECs in the presence
of 8 mg/ml Polybrene (Sigma). BOECs (5?104) were
seeded 24 h before transduction in supplemented ECBM2
culture medium without heparin into collagen-coated six-
well plates. Cells were exposed to virus-containing
supernatants for 8 h, thereafter, medium was changed.
After 48 h, BOECs transduced with the pLEIN vector
were used to determine transduction efficiency by FACS
analysis. Transduction efficiency routinely exceeded 95%.
Cells were selected with 300 mg/ml G418 (Invitrogen,
Karlsruhe, Germany) for 5 days. Thus, BOECs containing
pLXIN (BOEC/Mock), pL(FCU1)IN (BOEC/FCU1) and
pLEIN (BOEC/EGFP) were generated.
Determination of viable cells
Cells were counted using the CASY cell counter (Scha ¨rfe
System, Reutlingen, Germany). Viability of cells was
assessed by flow cytometry after incubation with 1 mg/
ml of propidium iodide. Cells not taking up propidium
iodide were considered to be viable.
Bystander effect assay
BOECs transduced with pL(FCU1)IN were mixed with
LLC cells to achieve proportions of transduced BOECs of
0, 1.6, 3.2, 6.3, 12.5, 25, 50, 75 and 100%. The mixtures
were plated in triplicates in 24-well plates at a density of
3?103cells per well and allowed to attach overnight.
BOEC cancer gene therapy
J Wei et al
Cells were then incubated with medium (ECBM2, 2%
FBS) alone as control or with 250 mM 5-FC for 3 days and
the number of viable cells were determined.
Assessment of biodistribution
The chelating ligand p-isothiocyanatobenzyl-diethylene-
coupled to acetylated low-density lipoprotein (Ac-LDL)
in Tris-buffer (pH 8.3). After 2 h of incubation at room
temperature and 10 h at 41C excess mx-DTPA was
removed by gel filtration (Sephadex; Fluka, Buchs,
Switzerland). Labeling was performed in citrate/acetate
buffer (pH 5.5) with 80 MBq of
111In-DTP-Ac-LDL was isolated using size exclusion
HPLC (purity 492%). BOECs were incubated with
111In-DTPA-Ac-LDL for 4 h at 371C. Cells were washed
and resuspended in culture medium containing 2% FBS.
Radiolabeled cells (4?105,
injected into 8- to 10-week-old male Rag2?/?cgc?/?mice
via the tail vein. Mice were killed on day 2 (n¼4) and on
day 5 (n¼3) after injection. Blood was aspirated by
intracardiac puncture and femurs and solid organs
were procured. All organs were weighted. Samples were
analyzed using a g-counter and results expressed as
counts per minute per mg tissue of each organ to correct
for the different size of the organs. Samples obtained
from noninjected mice served as background control.
To further assess the biodistribution of BOECs to solid
BOECs were tail vein injected into 8- to 10-week-old
male Rag2?/?cgc?/?mice. Solid organs were procured
2 days after BOEC injection and cryosections were
analyzed for the presence of BOECs by staining for
human vWF (1:25).
111In. The product
Assessment of homing to LLC metastases
LLC cells (5?104) were injected into the tail vein of 8- to
12-week-old male Rag2?/?cgc?/?mice. On day 7, when
metastases were established, mice were subjected to 6 Gy
of total body g-irradiation (137Cs; Nordion, Gammacell
1000, Kanata, Canada) or left unirradiated. Starting 24 h
later, mice received two tail vein injections of 4?105
BOECs labeled with DiI (100 mg/ml) in 400 ml PBS spaced
2 days apart (n¼5 per group). Human foreskin-derived
fibroblasts cells (4?105) injected in same numbers were
used as controls (n¼3). Mice were killed 2 days after the
last injection. Organs were procured, fixed in 4% PFA,
cryosectioned, stained with Hoechst 33258 and micro-
scopically examined for DiI-fluorescence.
Assessment of homing to orthotopic U87 gliomas
Human U87, LNT-229 and SF126 glioma cells (5?105)
were stereotactically implanted unilaterally into the right
striatum of 8- to 12-week-old male Rag2?/?cgc?/?mice.
After 4 days (SF126) or 14 days (U87 and LNT-229), when
the gliomas were established, mice received 6 Gy of
cranial g-irradiation. After 3 days, 1?105BOECs pulsed
with BrdU at a concentration of 10 mM for 5 days were
injected into the tail vein once (n¼3). Control mice did
not receive BOECs (n¼3 for each glioma entity). After 1
day, the brains were procured, fixed in 4% PFA,
cryosectioned and examined by immunohistochemistry
for BrdU containing cells.
In vivo survival
Lewis lung carcinoma cells were injected on day 0 into
the tail vein of 8- to 10-week-old male Rag2?/?cgc?/?
mice. On day 7, after metastases were established, mice
received 3 Gy of total body g-irradiation followed by i.v.
injection of 4?105BOEC/FCU1 cells in 400 ml serum-
free ECBM2 medium. After 2 days, mice received
6.25 mg 5-FC (Sigma) in 300 ml PBS i.p. twice daily for
5 days. A second cycle of BOEC/FCU1 cells followed by
5-FC was then given. A second group was treated as
above except for radiation and was termed LLC-BOEC/
FCU1-5-FC. One control group – LLC-Irr-BOEC/FCU1-
PBS – had LLC metastases, received total body irradia-
tion and BOEC/FCU1 cells, followed by i.p. injections of
300 ml PBS. Additional control groups bearing LLC
metastases received irradiation and either 5-FC (LLC-
Irr-5FC) or PBS (LLC-Irr-PBS) i.p. Tumor-free mice that
received total body irradiation and BOEC/FCU1 cells
followed by either 5-FC (Irr-BOEC/FCU1-5FC) or PBS
i.p. injections (Irr-BOEC/FCU1-PBS, n¼10) were used as
toxicity controls. Unless stated otherwise, groups con-
sisted of 12 mice. Tumor bearing mice were monitored
daily and killed when appearing moribund as deter-
mined by a blinded observer. Mice that had received
only BOEC/FCU1 cells were killed 3 months after the
last inoculation of BOEC/FCU1 cells. Organs were
removed and processed for histological analysis. All
animal work was performed in accordance with state
guidelines for animal care and use.
In vivo tumor inhibition
LLC cells (1?106) or a mixture of 1?106LLC cells with
3.5?105BOEC/FCU1 cells were resuspended in Matri-
gel (BD Biosciences; 1 mg/ml in DMEM) and inoculated
s.c. into 8- to 10-week-old male Rag2?/?cgc?/?mice.
After 3 days, when tumors were palpable, mice (n¼7 or
8 per group) received i.p. injections twice daily for 6 days
of either 6.25 mg 5-FC in 300 ml PBS (LLC-BOEC/FCU1-
5FC group and LLC-5FC group) or 300 ml PBS (LLC-
BOEC/FCU1-PBS group and LLC-PBS group). Tumor
size was determined every two days and tumor volume
calculated using the formula volume¼width2?length
?0.4. The experiment was terminated when the first
tumor started to bleed.
Immunohistochemistry and fluorescence microscopy
Organs were fixed with 4% PFA, cryoprotected in 20%
sucrose/PBS, embedded in OCT compound at ?801C
and serially cut into 5 m thick sections. Slides were
blocked with biotin/avidin blocking system (DAKO), 1%
bovine serum albumin (Serva, Heidelberg, Germany)
and 10% goat serum (DAKO). Slides were incubated at
41C with anti-human vWF (1:25 dilution in Tris-HCl
buffer containing 0.1% Triton/0.1% BSA) followed by
biotinylated goat anti-mouse secondary antibody (1:100;
DAKO) and alkaline phosphatase-conjugated avidin/
biotin complexes (DAKO). Cells were visualized by the
fast-red substrate system (DAKO) and counterstained
with hematoxylin. To detect BrdU containing cells slides
were microwaved in 0.01 M citric acid, pH 6.0 and then
stained with anti-BrdU monoclonal antibody (Roche)
To detect mature endothelial cells serially cut frozen
BOEC cancer gene therapy
J Wei et al
sections were stained with 150 mg/ml FITC-labeled
GSL I – isolectin B4 (Vector Laboratories, Burlingame,
Student’s t-test was used to analyze statistical differences
among groups in the assays on migration, paracrine
proliferative effect, duration of bystander effect and s.c.
tumor growth. Survival data were analyzed by Kaplan–
Meier estimator analysis and compared using the
Kruskal–Wallis test for multiple comparisons and the
Mann–Whitney test for single comparisons. Values of
Po0.05 were considered to be statistically significant.
We thank H Ribbert from the Department of Nuclear
Medicine for help in preparing and detecting111In-DTP-
Ac-LDL-loaded BOECs, M Shibuya for the LLC and
LLC-PLGF cell lines, P Vajkoczy for SF126 cells, W Wick
for LNT-229 cells and for introducing us to stereotactic
manipulations, S Zhou for providing human fibroblasts,
S Miebach for introduction to spheroid assays and
I Gastrock-Balitsch, U Kirchner and H Knauss for
excellent technical help. Supported by a grant from the
Wilhelm Sander-Stiftung (to CB).
1 Asahara T, Takahashi T, Masuda H, Kalka C, Chen D, Iwaguro H
et al. VEGF contributes to postnatal neovascularization by
mobilizing bone marrow-derived endothelial progenitor cells.
EMBO J 1999; 18: 3964–3972.
2 Arbab AS, Pandit SD, Anderson SA, Yocum GT, Bur M, Khuu
HM et al. MRI and confocal microscopy studies of magnetically
labeled endothelial progenitor cells trafficking to sites of tumor
angiogenesis. Stem Cells 2006; 24: 671–678.
3 Asahara T, Masuda H, Takahashi T, Kalka C, Pastore C, Silver M
et al. Bone marrow origin of endothelial progenitor cells
responsible for postnatal vasculogenesis in physiological and
pathological neovascularization. Circ Res 1999; 85: 221–228.
4 Coussens LM, Tinkle CL, Hanahan D, Werb Z. MMP-9 supplied
by bone marrow-derived cells contributes to skin carcinogenesis.
Cell 2000; 103: 481–490.
5 Davidoff AM, Ng CY, Brown P, Leary MA, Spurbeck WW, Zhou
J et al. Bone marrow-derived cells contribute to tumor
neovasculature and, when modified to express an angiogenesis
inhibitor, can restrict tumor growth in mice. Clin Cancer Res 2001;
6 Lyden D, Hattori K, Dias S, Costa C, Blaikie P, Butros L et al.
Impaired recruitment of bone-marrow-derived endothelial and
hematopoietic precursor cells blocks tumor angiogenesis and
growth. Nat Med 2001; 7: 1194–1201.
7 De Palma M, Venneri MA, Roca C, Naldini L. Targeting
exogenous genes to tumor angiogenesis by transplantation of
genetically modified hematopoietic stem cells. Nat Med 2003; 9:
8 Garcia-Barros M, Paris F, Cordon-Cardo C, Lyden D, Rafii S,
Haimovitz-Friedman A et al. Tumor response to radiotherapy
regulated by endothelial cell apoptosis. Science 2003; 300:
9 Machein MR, Renninger S, de Lima-Hahn E, Plate KH. Minor
contribution of bone marrow-derived endothelial progenitors
to the vascularization of murine gliomas. Brain Pathol 2003; 13:
10 Ruzinova MB, Schoer RA, Gerald W, Egan JE, Pandolfi PP, Rafii
S et al. Effect of angiogenesis inhibition by Id loss and the
contribution of bone-marrow-derived endothelial cells in spon-
taneous murine tumors. Cancer Cell 2003; 4: 277–289.
11 Gothert JR, Gustin SE, van Eekelen JA, Schmidt U, Hall MA, Jane
SM et al. Genetically tagging endothelial cells in vivo: bone
marrow-derived cells do not contribute to tumor endothelium.
Blood 2004; 104: 1769–1777.
12 Hilbe W, Dirnhofer S, Oberwasserlechner F, Schmid T, Gunsilius
E, Hilbe G et al. CD133 positive endothelial progenitor cells
contribute to the tumour vasculature in non-small cell lung
cancer. J Clin Pathol 2004; 57: 965–969.
13 Li H, Gerald WL, Benezra R. Utilization of bone marrow-derived
endothelial cell precursors in spontaneous prostate tumors
varies with tumor grade. Cancer Res 2004; 64: 6137–6143.
14 Rajantie I, Ilmonen M, Alminaite A, Ozerdem U, Alitalo K,
Salven P. Adult bone marrow-derived cells recruited during
angiogenesis comprise precursors for periendothelial vascular
mural cells. Blood 2004; 104: 2084–2086.
15 Strobel P, Hartmann M, Jakob A, Mikesch K, Brink I,
Dirnhofer S et al. Thymic carcinoma with overexpression of
mutated KIT and the response to imatinib. N Engl J Med 2004;
16 Yang L, DeBusk LM, Fukuda K, Fingleton B, Green-Jarvis B,
Shyr Y et al. Expansion of myeloid immune suppressor
Gr+CD11b+ cells in tumor-bearing host directly promotes tumor
angiogenesis. Cancer Cell 2004; 6: 409–421.
17 Anderson SA, Glod J, Arbab AS, Noel M, Ashari P, Fine HA et al.
Noninvasive MR imaging of magnetically labeled stem cells to
directly identify neovasculature in a glioma model. Blood 2005;
18 De Palma M, Venneri MA, Galli R, Sergi LS, Politi LS,
Sampaolesi M et al. Tie2 identifies a hematopoietic lineage of
proangiogenic monocytes required for tumor vessel formation
and a mesenchymal population of pericyte progenitors. Cancer
Cell 2005; 8: 211–226.
19 Larrivee B, Niessen K, Pollet I, Corbel SY, Long M, Rossi FM et al.
Minimal contribution of marrow-derived endothelial precursors
to tumor vasculature. J Immunol 2005; 175: 2890–2899.
20 Peters BA, Diaz LA, Polyak K, Meszler L, Romans K, Guinan EC
et al. Contribution of bone marrow-derived endothelial cells to
human tumor vasculature. Nat Med 2005; 11: 261–262.
21 Shinde Patil VR, Friedrich EB, Wolley AE, Gerszten RE, Allport
JR, Weissleder R. Bone marrow-derived lin(?)c-kit(+)Sca-1+
stem cells do not contribute to vasculogenesis in Lewis lung
carcinoma. Neoplasia 2005; 7: 234–240.
22 Duda DG, Cohen KS, Kozin SV, Perentes JY, Fukumura D,
Scadden DTet al. Evidence for bone marrow-derived endothelial
cells incorporation into perfused blood vessels in tumors. Blood
2006; 107: 2774–2776.
23 Ferrari N, Glod J, Lee J, Kobiler D, Fine HA. Bone marrow-
derived, endothelial progenitor-like cells as angiogenesis-selec-
tive gene-targeting vectors. Gene Ther 2003; 10: 647–656.
24 Arafat WO, Casado E, Wang M, Alvarez RD, Siegal GP, Glorioso
JC et al. Genetically modified CD34+ cells exert a cytotoxic
bystander effect on human endothelial and cancer cells. Clin
Cancer Res 2000; 6: 4442–4448.
25 Moore XL, Lu J, Sun L, Zhu CJ, Tan P, Wong MC. Endothelial
progenitor cells’ ‘homing’ specificity to brain tumors. Gene
Therapy 2004; 11: 811–818.
26 Tabatabai G, Frank B, Mohle R, Weller M, Wick W. Irradiation
and hypoxia promote homing of haematopoietic progenitor cells
towards gliomas by TGF-beta-dependent HIF-1alpha-mediated
induction of CXCL12. Brain 2006; 129: 2426–2435.
27 Jevremovic D, Gulati R, Hennig I, Diaz RM, Cole C, Kleppe L et
al. Use of blood outgrowth endothelial cells as virus-producing
vectors for gene delivery to tumors. Am J Physiol Heart Circ
Physiol 2004; 287: H494–H500.
BOEC cancer gene therapy
J Wei et al
28 Le Ricousse-Roussanne S, Barateau V, Contreres JO, Boval B,
Kraus-Berthier L, Tobelem G. Ex vivo differentiated endothelial
and smooth muscle cells from human cord blood progenitors
home to the angiogenic tumor vasculature. Cardiovasc Res 2004;
29 Lin Y, Weisdorf DJ, Solovey A, Hebbel RP. Origins of circulating
endothelial cells and endothelial outgrowth from blood. J Clin
Invest 2000; 105: 71–77.
30 Lin Y, Chang L, Solovey A, Healey JF, Lollar P, Hebbel RP. Use of
blood outgrowth endothelial cells for gene therapy for hemo-
philia A. Blood 2002; 99: 457–462.
31 Hicklin DJ, Ellis LM. Role of the vascular endothelial growth
factor pathway in tumor growth and angiogenesis. J Clin Oncol
2005; 23: 1011–1027.
32 Ferrara N, Gerber HP, LeCouter J. The biology of VEGF and its
receptors. Nat Med 2003; 9: 669–676.
33 Gill M, Dias S, Hattori K, Rivera ML, Hicklin D, Witte L et al.
Vascular trauma induces rapid but transient mobilization of
VEGFR2(+)AC133(+) endothelial precursor cells. Circ Res 2001;
34 Heissig B, Hattori K, Dias S, Friedrich M, Ferris B, Hackett NR
et al. Recruitment of stem and progenitor cells from the bone
marrow niche requires MMP-9 mediated release of kit-ligand.
Cell 2002; 109: 625–637.
growth factor andits receptor,
growth factor receptor-1: novel targets for stimulation of
ischemic tissue revascularization and inhibition of angiogenic
and inflammatorydisorders. J
36 Clauss M, Weich H, Breier G, Knies U, Rockl W, Waltenberger J
et al. The vascular endothelial growth factor receptor Flt-1
mediates biological activities. Implications for a functional role
of placenta growth factor in monocyte activation and chemo-
taxis. J Biol Chem 1996; 271: 17629–17634.
37 Carmeliet P, Moons L, Luttun A, Vincenti V, Compernolle V, De
Mol M et al. Synergism between vascular endothelial growth
factor and placental growth factor contributes to angiogenesis
and plasma extravasation in pathological conditions. Nat Med
2001; 7: 575–583.
38 Hiratsuka S, Maru Y, Okada A, Seiki M, Noda T, Shibuya M.
Involvement of Flt-1 tyrosine kinase (vascular endothelial
growth factor receptor-1) in pathological angiogenesis. Cancer
Res 2001; 61: 1207–1213.
39 Adini A, Kornaga T, Firoozbakht F, Benjamin LE. Placental
growth factor is a survival factor for tumor endothelial cells and
macrophages. Cancer Res 2002; 62: 2749–2752.
40 Zhang L, Chen J, Ke Y, Mansel RE, Jiang WG. Expression of
Placenta growth factor (PlGF) in non-Small cell Lung cancer
(NSCLC) and the clinical and prognostic significance. World J
Surg Oncol 2005; 3: 68.
41 Ikai T, Miwa H, Shikami M, Hiramatsu A, Tajima E, Yamamoto
H et al. Placenta growth factor stimulates the growth of
leukemia cells by both autocrine and paracrine pathways. Eur
J Haematol 2005; 75: 273–279.
42 Donnini S, Machein MR, Plate KH, Weich HA. Expression and
localization of placenta growth factor and PlGF receptors in
human meningiomas. J Pathol 1999; 189: 66–71.
43 Lacal PM, Failla CM, Pagani E, Odorisio T, Schietroma C,
Falcinelli S et al. Human melanoma cells secrete and respond to
placenta growth factor and vascular endothelial growth factor.
J Invest Dermatol 2000; 115: 1000–1007.
ThrombHaemost 2003; 1:
44 Wei SC, Tsao PN, Yu SC, Shun CT, Tsai-Wu JJ, Wu CH et al.
Placenta growth factor expression is correlated with survival of
patients with colorectal cancer. Gut 2005; 54: 666–672.
45 Chen CN, Hsieh FJ, Cheng YM, Cheng WF, Su YN, Chang KJ
et al. The significance of placenta growth factor in angiogenesis
and clinical outcome of human gastric cancer. Cancer Lett 2004;
46 Matsumoto K, Suzuki K, Koike H, Okamura K, Tsuchiya K,
Uchida T et al. Prognostic significance of plasma placental
growth factor levels in renal cell cancer: an association with
clinical characteristics and vascular endothelial growth factor
levels. Anticancer Res 2003; 23: 4953–4958.
47 Taylor AP, Osorio L, Craig R, Raleigh JA, Ying Z, Goldenberg
DM et al. Tumor-specific regulation of angiogenic growth factors
and their receptors during recovery from cytotoxic therapy.
Clin Cancer Res 2002; 8: 1213–1222.
48 Luttun A, Tjwa M, Moons L, Wu Y, Angelillo-Scherrer A, Liao F
et al. Revascularization of ischemic tissues by PlGF treatment,
and inhibition of tumor angiogenesis, arthritis and atherosclero-
sis by anti-Flt1. Nat Med 2002; 8: 831–840.
49 Aicher A, Brenner W, Zuhayra M, Badorff C, Massoudi S,
Assmus B et al. Assessment of the tissue distribution of
transplanted human endothelial progenitor cells by radioactive
labeling. Circulation 2003; 107: 2134–2139.
50 Virgolini I, Angelberger P, Li SR, Koller F, Koller E, Pidlich J et al.
Indium-111-labeled low-density lipoprotein binds with higher
affinity to the human liver as compared to iodine-123-low-
density-labeled lipoprotein. J Nucl Med 1991; 32: 2132–2138.
51 Erbs P, Regulier E, Kintz J, Leroy P, Poitevin Y, Exinger F et al. In
vivo cancer gene therapy by adenovirus-mediated transfer of a
bifunctional yeast cytosine deaminase/uracil phosphoribosyl-
transferase fusion gene. Cancer Res 2000; 60: 3813–3822.
52 Huber BE, Austin EA, Richards CA, Davis ST, Good SS.
Metabolism of 5-fluorocytosine to 5-fluorouracil in human
colorectal tumor cells transduced with the cytosine deaminase
gene: significant antitumor effects when only a small percentage
of tumor cells express cytosine deaminase. Proc Natl Acad Sci
USA 1994; 91: 8302–8306.
53 Pasqualini R, Arap W, McDonald DM. Probing the structural
and molecular diversity of tumor vasculature. Trends Mol Med
2002; 8: 563–571.
54 Rumpold H, Wolf D, Koeck R, Gunsilius E. Endothelial
progenitor cells: a source for therapeutic vasculogenesis? J Cell
Mol Med 2004; 8: 509–518.
55 Ishizawa K, Kubo H, Yamada M, Kobayashi S, Suzuki T, Mizuno
S et al. Hepatocyte growth factor induces angiogenesis in injured
lungs through mobilizing endothelial progenitor cells. Biochem
Biophys Res Commun 2004; 324: 276–280.
56 Li H, Gerald WL, Benezra R. Utilization of bone marrow-derived
endothelial cell precursors in spontaneous prostate tumors
varies with tumor grade. Cancer Res 2004; 64: 6137–6143.
57 Gorski DH, Beckett MA, Jaskowiak NT, Calvin DP, Mauceri HJ,
Salloum RM et al. Blockage of the vascular endothelial growth
factor stress response increases the antitumor effects of ionizing
radiation. Cancer Res 1999; 59: 3374–3378.
58 Sonveaux P, Brouet A, Havaux X, Gregoire V, Dessy C, Balligand
JL et al. Irradiation-induced angiogenesis through the up-
regulation of the nitric oxide pathway: implications for tumor
radiotherapy. Cancer Res 2003; 63: 1012–1019.
59 Heissig B, Rafii S, Akiyama H, Ohki Y, Sato Y, Rafael Tet al. Low-
dose irradiation promotes tissue revascularization through
VEGF release from mast cells and MMP-9-mediated progenitor
cell mobilization. J Exp Med 2005; 202: 739–750.
Supplementary Information accompanies the paper on Gene Therapy website (http:/ /www.nature.com/gt)
BOEC cancer gene therapy
J Wei et al