Download Viral Vector Delivery in Solid-State Vehicles: Gene Expression in a

Viral Vector Delivery in Solid-State Vehicles: Gene
Expression in a Murine Prostate Cancer Model
D. Robert Siemens, J. Christopher Austin, Sean P. Hedican, James Tartaglia,
Timothy L. Ratliff
Background: Although there are increasingly more clinical
trials involving gene therapy, efficient gene transfer remains
a major hurdle to success. To enhance the efficiency of delivery of viral vectors in gene therapy protocols, we evaluated the effect of various matrices to act as a vehicle for
recombinant virus during intratumoral injection. Methods:
The ability of several vehicles (catgut spacer, polyglycolic
acid, chromic catgut, and gelatin sponge matrix) to deliver
the canarypox virus ALVAC to the cells of the murine prostate cancer cell line RM-1 was studied in vitro and in vivo.
ALVAC recombinants encoding the murine cytokines interleukin 2 (IL-2), interleukin 12 (IL-12), and tumor necrosis
factor-␣ (TNF-␣) were used to assess enhancement of antitumor activity after intratumoral inoculation. Confirmatory
experiments were conducted by use of another mouse prostate cancer cell line, RM-11, and a mouse bladder cancer cell
line, MB-49. All statistical tests were two-sided. Results: The
gelatin sponge matrix proved to be the most effective solidstate vehicle for delivering viral vectors to cells in culture. In
addition, this matrix statistically significantly enhanced expression of ALVAC-delivered reporter genes in tumor models when compared with fluid-phase delivery of virus (P =
.037 for the RM-1 model and P = .03 for the MB-49 model).
Statistically significant growth inhibition of established tumors was observed when a combination of the three recombinant ALVAC viruses expressing IL-2, IL-12, and TNF-␣
was delivered with the matrix in comparison with 1) fluidphase intratumoral injection of the ALVAC recombinants,
2) no treatment, or 3) treatment with parental ALVAC (all
P<.05). Conclusions: Viral vector delivery in a solid-state
vehicle resulted in improved recombinant gene expression in
vivo and translated to greater inhibition of tumor growth in
an immunotherapy protocol for heterotopic tumor nodules.
The efficient delivery of reporter genes described herein may
prove useful in many solid tumor gene therapy protocols. [J
Natl Cancer Inst 2000;92:403–12]
Gene therapy protocols for cancer are based on eradicating
tumor cells either directly (e.g., toxic genes) or indirectly (e.g.,
genes that elicit antitumor immune responses). Alternatively,
corrective gene therapy involves the replacement or inactivation
of defective genes in neoplastic cells (e.g., p53 [also known as
TP53]). There are increasingly more clinical gene therapy trials
under way and, although many investigations have demonstrated
a great promise in preclinical studies, the efficient and accurate
delivery of therapeutic genes remains a formidable task in all
solid-tumor oncology.
We have previously reported one approach to cancer immunotherapy involving the transfer of genes encoding the cytokines
interleukin 2 (IL-2) and tumor necrosis factor-␣ (TNF-␣) utilizing the canarypox viral vector ALVAC (1). The ALVAC virus
Journal of the National Cancer Institute, Vol. 92, No. 5, March 1, 2000
was shown to efficiently infect murine prostate cancer cells,
RM-1, and to produce high levels of extrinsic gene product.
In addition, antitumor immunity was induced when tumor
cells were infected by ALVAC cytokine recombinants and injected subcutaneously in the flanks of male C57BL/6 mice. The
ALVAC virus is particularly well suited for the direct injection
of tumors because the inability of the canarypox virus vectors
to replicate in human cells greatly reduces the risk of systemic
adverse events. The absence of replication also means that
the virus will not propagate through the tumor mass; therefore,
the expression of the delivered gene products is restricted to
the sites of delivery. Our preliminary studies have shown a restricted pattern of expression after direct injection of fluid-phase
ALVAC into the prostate or a tumor at a subcutaneous site.
Moreover, any clinical gene therapy protocol for prostate cancer
involving the intraprostatic delivery of therapeutic genes will be
hampered by the multifocal and often cryptic nature of neoplastic and preneoplastic lesions in the prostate. Thus, methods to
enhance the distribution and expression of recombinant genes
are needed.
Numerous substances have been used as carriers to enhance
and sustain the delivery of soluble products to both neoplastic
and non-neoplastic tissue (2,3). In these studies, a gelatin sponge
matrix was determined to be the most efficient in vitro and was
shown to enhance delivery and, hence, reporter gene expression
when injected intratumorally. The previously described tumor
suppression in this model by recombinant virus encoding genes
for IL-2, interleukin 12 (IL-12), and TNF-␣ was also markedly
improved when the vector was delivered by the gelatin matrix as
compared with fluid-phase injection.
In this study, we tested the ability of different matrices to act
as a carrier vehicle for the canarypox virus (ALVAC) to improve
the delivery of the vector in a heterotopic murine prostate cancer
model.
MATERIALS
AND
METHODS
Animals and Tumor Cells
The murine prostate cancer cell line RM-1 used for these studies mimics
multistep carcinogenesis by activating the ras and myc oncogenes and is used to
induce an aggressive prostate carcinoma in vivo. This cell line retains many
features of prostate cancer, including androgen responsiveness early in culture,
expression of androgen receptor, and progression to androgen independence with
Affiliations of authors: D. R. Siemens, J. C. Austin, S. P. Hedican, Department
of Urology, The University of Iowa, Iowa City; J. Tartaglia, Virogenetics Corporation, Troy, NY; T. L. Ratliff, Department of Urology, The University of
Iowa Cancer Center and The University of Iowa Prostate Cancer Research
Group, Iowa City.
Correspondence to: Timothy L. Ratliff, Ph.D., Department of Urology, The
University of Iowa, 200 Hawkins Dr., 3 RCP, Iowa City, IA 52242-1089 (e-mail:
[email protected]).
See “Notes” following “References.”
© Oxford University Press
ARTICLES 403
time (4). MB-49, a chemically induced mouse bladder tumor, was used in concert with RM-1 throughout the in vitro and in vivo gene expression experiments.
Both RM-1 and MB-49 are syngeneic to C57BL/6 mice. The myc- and rastransformed BALB/c RM-11 prostate cancer cell line was used for complementary in vivo tumor outgrowth studies. Cultured cells were maintained in Dulbecco’s modified Eagle medium (DMEM) containing 10% fetal calf serum (FCS).
Mice (6–8 weeks old at the time of study initiation) were obtained through the
National Cancer Institute, Bethesda, MD, and were allowed free access to food
and water. All animal studies were approved by the Animal Review Board of the
University of Iowa and were performed in accordance with institutional guidelines.
Gene Transfer Vectors
ALVAC is a canarypox virus that can infect mammalian cells but is restricted
to avian species for replication (5). It has been shown to be a safe and effective
vector in both humans and animals (6,7). The viral strain from which ALVAC
was obtained was isolated from a pox lesion on an infected canary. Parental
ALVAC, ALVAC vectors encoding murine IL-2, murine IL-12 and murine
TNF-␣, as well as the reporter gene constructs ␤-galactosidase (ALVAC-lacZ),
green fluorescent protein (ALVAC-GFP), and luciferase (ALVAC-luciferase)
were developed at Virogenetics Corporation (Troy, NY).
Delivery Systems
Four solid-state delivery matrices were compared with fluid-phase delivery of
viral vector to cells in vitro and in vivo: polyglycolic acid, chromic catgut, catgut
spacer, and gelatin sponge. These substances were chosen because of their
absorbable nature and relatively low tissue reaction. Polyglycolic acid (Davis
and Geck, Inc., Wayne, NJ) and chromic catgut (Ethicon, Inc., Somerville, NJ)
are both absorbable suture materials. Plain catgut spacer material (MDTech,
Gainesville, FL) is a commercially available product used for prostate cancer
brachytherapy protocols. An absorbable gelatin sponge (Gelfoam; Pharmacia
and Upjohn, Kalamazoo, MI) is prepared from purified pork skin gelatin granules and is used as a hemostatic agent. All delivery systems were prepared in
6-mm lengths and were tested in vitro and in vivo through an 18-gauge B-D
spinal needle (Becton Dickinson and Co., Franklin, NJ) to better mimic intraprostatic injection in a clinical trial.
As previously described for the carrier delivery for insulin (2), the virion
concentration to be delivered by an individual matrix was determined by weighing the matrices before and after viral absorption. The dry and wet weights of the
matrices after 1 minute of viral absorption were recorded, and the subsequent
volume delivered was calculated. Subsequently, to ensure that the gelatin sponge
matrix was reliably delivering this weight-calculated number of viral particles,
comparisons were made with the use of particle determination by measurement
of the optical density (OD) at 260 nm after digestion of the gelatin matrix. A
calculated particle concentration was delivered into solution by the gelatin
sponge matrix and then digested by a combination of collagenase (0.16%),
bovine serum albumin (2.5%), and deoxyribonuclease (0.001%) in phosphatebuffered saline (PBS) (pH 7.2); these three reagents were obtained from Sigma
Chemical Co., St. Louis, MO.
Infection and Reporter Gene Assays
For in vitro analysis, RM-1 or MB-49 cells were harvested from tissue culture
plates and replated with DMEM containing 10% FCS and 10 mM HEPES buffer
(pH 7.0) on the day before infection. The medium was changed to DMEM with
2% FCS at the time of viral infection with either ALVAC-luciferase or ALVAClacZ delivered directly into the culture or via the delivery matrices. The viral
vectors were added to the cells at the multiplicity of infection (MOI)–plaqueforming units (pfu) per cell shown in each experiment. The cells were then
incubated for 6 hours at 37 °C in an atmosphere of 5% CO2 when the medium
was changed back to DMEM with 10% FCS. Reporter assays were performed 48
hours after virus addition (unless otherwise stated). All in vitro experiments were
performed in triplicate and repeated in at least two independent experiments.
For the in vivo gene expression studies, RM-1 or MB-49 cells were harvested
from the tissue culture plates by treatment with 10 mM EDTA and were washed
with PBS. The cells were then resuspended in DMEM in a concentration of 5 ×
106 pfu/mL, and 0.1 mL was injected subcutaneously into the backs of mice. The
ALVAC vectors recombinant for luciferase or ␤-galactosidase were injected
either directly (fluid phase) or via the delivery systems at a concentration of 3 ×
106 pfu/mL approximately 10 days after tumor implantation. Tumors at that time
were approximately 8 mm by 8 mm (approximately 200 mg wet weight). Tumors
404 ARTICLES
were harvested for reporter gene assays at various times after infection as described for individual experiments. Experiments were performed at least twice
for the RM-1 and the MB-49 tumors in vivo.
We determined ␤-galactosidase transgene expression in vitro after briefly
fixing the cells with 0.5% glutaraldehyde for 10 minutes, washing them with
PBS, then incubating them in X-gal (5-bromo-4-chloro-indolyl ␤- D galactopyranoside) at 37 °C for at least 4 hours. ␤-Galactosidase cleaves this
substrate into an indigo compound, such that cells producing the ␤-galactosidase
gene product are stained blue. We quantified ␤-galactosidase expression by
visualizing a representative area of the culture plate under high power (×40) and
recording the percentage of blue cells. In vivo, after the tumor was harvested and
weighed, sections were incubated in the X-gal solution containing the detergents
sodium deoxycholate (0.01%) and Nonidet P-40 (0.02%). Representative tissue
sections were then scored for percentage of blue-stained cells. Tumors infected
with the parental ALVAC (not recombinant for the lacZ gene) were used as
negative controls.
The luciferase assay from cell lysates was performed with the use of a commercial luciferase assay kit (Promega Corp., Madison, WI) following the manufacturer’s recommendations. The Monolight 2010 Luminometer (Analytical Luminescence Laboratory, Ann Arbor, MI) was used for the luciferase assay.
Internal controls were performed by reading background luminescence for each
assay as well as the periodic evaluation of the variation between replicates.
Luciferase assays of infected subcutaneous tumors were performed after each
tumor was homogenized with the Tissue Tearor (Biospeck Products, Inc., Fisher
Scientific, Itasca, IL) in 0.5 mL of cell lysis buffer.
Tumor Inhibition Studies
RM-1 (5 × 105) or RM-11 (1 × 105) cells were injected subcutaneously in the
backs of mice in a volume of 0.1 mL as described above. Approximately 10 days
after tumor implantation, a total of 8.4 × 106 pfu of recombinant ALVAC vector
was injected intratumorally. The ALVAC vectors used were the IL-2, IL-12, and
TNF-␣ constructs in equal concentrations (2.8 × 106 pfu each). The vectors
absorbed by the gelatin sponge matrix were injected in an 18-gauge needle and
were compared with three separate 33-␮L injections of the fluid-phase product
(8.4 × 106 total pfu). Other controls included parental ALVAC absorbed by the
gelatin sponge matrix, matrix only, and a no treatment group.
Tumor outgrowth, determined by tumor size as a function of time, was measured approximately three times a week. Survival of the tumor-bearing mice was
also determined. Mice were killed for humane reasons if a single tumor was
greater than 25 mm in any dimension or if the mice appeared to be ill from the
tumor burden. Each experimental group contained four to six mice, and experiments were repeated at least twice.
Statistical Evaluation
Differences were analyzed by the Mann–Whitney rank sum test, including the
nonparametric data for ␤-galactosidase and luciferase reporter assays. The gene
expression data are recorded as the means and the 95% confidence intervals
(CIs). The rank sum test (Mann–Whitney) was also used to compare average
tumor volume between matrix-delivered and fluid-phase-delivered treatment
groups at individual time points. These data were also presented in the figures as
the means and the 95% CIs. A one-way analysis of variance of log-transformed
data was also used to compare all control groups with the gelatin sponge matrixdelivered treatment group at individual time points for the tumor outgrowth
studies. Survival data were analyzed for significance with the use of the Cox
proportional hazards regression model. For all statistical analyses, we used a
computer software program, SAS (SAS Institute, Inc., Cary, NC) (8), or Statistix
(Analytical Software, Seattle, WA; Version 1.0, 1996). All reported P values are
two-sided, and statistical significance was determined as a P value of less than
.05.
RESULTS
Determination of Virion Concentration
The calculated volume delivered by the individual matrices
was determined by the wet weight after absorption of the virus
in the 18-guage delivery needle. The mean wet weight and dry
weight of at least five samples were assessed for each delivery
matrix, and the results of three separate determinations are preJournal of the National Cancer Institute, Vol. 92, No. 5, March 1, 2000
sented in Table 1. The known volumes absorbed by the individual matrices allowed comparisons to fluid-phase delivery
both in vitro and in vivo by calculation of the number of viral
particles or pfu. Thus, pfu delivery via either matrix or fluid was
equivalent, given the absorbable capacity of each matrix.
Equivalency was verified in the gelatin sponge matrix system as
determined by the particle count read as OD at 260 nm (data not
shown).
Delivery of Reporter Genes In Vitro
Table 1. Calculation of volume of vector absorbed by individual matrices
Solid-state matrices
Wet weight, mg*
Dry weight, mg*
Volume delivered,
␮L†
Chromic
catgut
Polyglycolic
acid
Catgut
spacer
Gelatin
sponge
2.15 ± 1.14
0.69 ± 0.01
1.46 ± 1.14
2.74 ± 0.98
0.56 ± 0.02
2.18 ± 0.97
4.47 ± 0.68
2.67 ± 0.05
1.80 ± 0.66
3.57 ± 0.78
0.93 ± 0.03
2.63 ± 0.80
*Wet weight (mg) was measured after solid-state matrix absorption of vector
in the fluid phase for 1 minute. Dry weight is the measurement (mg) of the
solid-state matrix before absorption of fluid. Results ⳱ the mean wet and dry
weights ± standard deviation of three separate measurements.
†Volume delivered (␮L) calculated from wet weight minus dry weight. Results ⳱ mean ± standard deviation.
To examine the ability of each matrix to deliver the calculated
virion concentration in cell culture, we determined ␤-galactosidase and the firefly luciferase expression of cells 48 hours after
infection with ALVAC-lacZ or ALVAC-luciferase vectors. The
virus delivered by the matrices was compared with the addition
of a known MOI (pfu) of fluid-phase virus in cell culture. As
shown in Fig. 1, A, no significant differences were found with
the percentage of RM-1 cells infected by the ␤-galactosidase
vector (MOI 10 : 1) in the gelatin sponge matrix-delivered group
(mean ⳱ 48%; 95% CI ⳱ 41%–55%) versus the fluid-phase
group (mean ⳱ 49%; 95% CI ⳱ 44%–55%). It is interesting
that there was a trend to less ␤-galactosidase expression for the
other delivery systems, especially for the chromic catgut. These
results were consistent over a wide range of MOI (10 : 1 to
Fig. 1. Comparison of in vitro delivery by different matrices. A) Percentage (95% confidence
interval) of ␤-galactosidase-expressing RM-1
cells detected 48 hours after in vitro infection
with ALVAC-␤-galactosidase. No statistically
significant difference (P ⳱ .66, Mann–Whitney
rank sum test) was seen between the gelatin
sponge matrix delivery and direct infusion of
the fluid-phase product. B) Luciferase activity
(relative light units) of RM-1 cell lysates 48
hours after infection with ALVAC-luciferase.
No statistically significant difference in luciferase activity was observed between the gelatin
sponge matrix delivery and direct infusion of
fluid-phase product (P ⳱ .08, Mann–Whitney
rank sum test). Gene expression shown represents experiments performed in triplicate at a
10 : 1 multiplicity of infection (MOI). Experiments at different MOIs (10 : 1 to 200 : 1) in
both the RM-1 and MB-49 cell lines demonstrated similar in vitro gene expression between
the gelatin sponge matrix delivery and fluidphase delivery.
Journal of the National Cancer Institute, Vol. 92, No. 5, March 1, 2000
ARTICLES 405
200 : 1) for the RM-1 cell line, as well as for experiments using
the MB-49 cell line.
Similarly, there was no significant difference between the
luciferase activity (relative light units [RLU]) of RM-1 tumor
cell lysates infected with ALVAC-luciferase (MOI 10 : 1) delivered by the gelatin sponge (mean ⳱ 1.8 × 107 RLU; 95% CI ⳱
9.2 × 106–2.5 × 107 RLU) compared with delivery of the fluidphase product in vitro (mean ⳱ 1.1 × 107 RLU; 95% CI ⳱ 7.7
× 106–1.5 × 107 RLU) (Fig. 1, B). No statistically significant difference was found between these two groups (P ⳱ .08,
Mann–Whitney rank sum test). Similarly, no differences in gene
transfer were observed between the gelatin sponge and fluidphase groups in the MB-49 cell line (P ⳱ .46, Mann–Whitney
rank sum test). Presumably, the entire viral load delivered into
culture by this matrix was available to infect the tumor cells.
Again, there was a trend to lower expression with the other
delivery systems. The lower gene expression observed with the
use of these systems most likely reflects either the inability to
deliver the entire calculated volume or the inability of the matrix
to release the absorbed virus in cell culture.
Transgene Expression in a Heterotopic Tumor Model
To determine the ability of these carrier systems to deliver the
viral vectors in vivo, we first injected ALVAC-luciferase (3 ×
106 pfu) into established subcutaneous RM-1, either in fluidphase or via the various delivery systems. The chromic system
was not tested, given its consistently poor transfer of virus to
cells in culture. Forty-eight hours after infection, the tumors
were harvested and the luciferase assay was performed. The
gelatin sponge matrix delivery consistently resulted in significantly (P ⳱ .037, Mann–Whitney rank sum test) enhanced gene
expression (mean ⳱ 34 226 RLU; 95% CI ⳱ 14 673–52 012
RLU) over fluid-phase delivery (mean ⳱ 2961 RLU; 95% CI ⳱
29–7707 RLU) (Fig. 2, A). The other delivery systems (polyglycolic acid and catgut spacer) did not consistently enhance gene
expression in injected tumor nodules. Similar experiments in the
Fig. 2. Comparison of in vivo delivery by different
matrices. A) Mean luciferase activity (95% confidence interval) of harvested subcutaneous RM-1
tumors (n ⳱ 5 per group) 48 hours after infection
by 3 × 106 plaque-forming units of ALVACluciferase delivered by different matrices. Statistically significant differences (P ⳱ .037, Mann–
Whitney rank sum test) in mean luciferase activity
were observed only between the gelatin sponge matrix delivery and fluid-phase injection. Similar experiments in MB-49 subcutaneous tumors also resulted in statistically significant increases (P<.03,
Mann–Whitney rank sum test) in luciferase gene
expression with gelatin sponge delivery compared
with fluid-phase delivery. B) Mean luciferase activity (95% confidence interval) in log scale of
RM-1 subcutaneous tumors (n ⳱ 4 per group) 48
hours after infection with ALVAC-luciferase at different virion concentrations.
406 ARTICLES
Journal of the National Cancer Institute, Vol. 92, No. 5, March 1, 2000
heterotopic MB-49 tumor model confirmed the ability of the
gelatin sponge matrix to significantly improve (P ⳱ .03, Mann–
Whitney rank sum test) the delivery of viral vectors compared
with direct injection of the fluid-phase product. Fig. 2, B, shows
the improved luciferase expression with gelatin sponge delivery
in the RM-1 tumor model at different doses (pfu) of the
ALVAC-luciferase vector (note log scale for RLU in Fig. 2, B).
To determine if the improved in vivo gene expression consistently observed at 48 hours was conserved over time, we
harvested subcutaneous RM-1 tumors infected with 3 × 106
ALVAC-luciferase for luciferase assays 24, 48, 72, and 96 hours
after injection. Those tumors that were injected with virus in the
gelatin sponge matrix had remarkably greater luciferase activity
than intratumoral fluid-phase virus injection at each period of
time (Fig. 3).
Enhanced gene expression was consistently demonstrated under various conditions (virion concentration, time) when the
gelatin sponge matrix was used to deliver the vectors. To determine if improved biodistribution throughout the tumor was
partly responsible for this improved expression, heterotopic tumors were infected by ALVAC encoding the reporter genes GFP
or ␤-galactosidase. Viewing tumors under the fluorescent microscope after ALVAC-GFP infection revealed much brighter
and more widespread fluorescence when the vectors were delivered by the gelatin sponge matrix than when the vectors were
delivered by fluid phase (Fig. 4, A and B). For comparisons,
controls for background fluorescence were determined by infecting tumors with parental ALVAC (data not shown).
Similarly, fluid-phase injection of tumors by ALVAC-lacZ
consistently revealed ␤-galactosidase activity only within a relatively narrow distribution along the needle tract. Delivery by the
gelatin sponge matrix, however, resulted in substantially more
widespread distribution. These results were also controlled for
endogenous ␤-galactosidase activity by staining tumors infected
with parental ALVAC vector (Fig. 4, C–E).
Tumor Outgrowth Studies
To determine if the enhanced gene expression found with the
matrix delivery of vectors translated into improved biologic effect, we treated established subcutaneous RM-1 tumors (mean
tumor volume ⳱ 110 mm3) with ALVAC vector expressing
IL-2, IL-12, and TNF-␣ delivered in the gelatin sponge matrix or
via fluid-phase injection. Control groups included 1) parental
ALVAC delivered via the matrix, 2) a matrix only, and 3) a
group with no treatment. Statistically significant (P values for all
comparisons <.05, Mann–Whitney rank sum test) tumor inhibition, as determined by tumor volume over time, was seen in the
treatment group only when delivered by the matrix (Fig. 5, A).
This tumor inhibition was greatest within the first 6 or 7 days
after infection with the recombinant virus, although the inhibitory effects remained significant through 13 days. Further comparison between groups was not possible because many control
mice were killed after day 13. Several tumors (three of five) in
the treatment group delivered by the gelatin matrix demonstrated
substantial inhibition of growth for a period of 10 days after gene
transfer, although all tumors did eventually grow out. Tumor
volumes at days 4–13 in the gelatin sponge matrix-delivered
group were statistically significantly smaller (P values for all
comparisons <.05, Mann–Whitney rank sum test) than tumors in
the fluid-phase injection group, as well as those in the control
groups. Similarly, a statistically significant (P<.005; Cox proportional hazards regression model) increase in survival was
seen for mice treated with the recombinant virus delivered by
gelatin matrix as compared with those mice treated with the
fluid-phase injection (Fig. 5, B). The differences observed between matrix and fluid-phase delivery have been confirmed in
four separate tumor outgrowth experiments.
For confirmation of these results in a different prostate tumor
model, subcutaneous RM-11 tumor nodules in BALB/c mice
were infected with the recombinant ALVAC vectors as previously described. Impressive, statistically significant tumor inhibition and regression (P values for all comparisons 艋.036,
Mann–Whitney rank sum test) were demonstrated in this model
when ALVAC recombinant for the IL-2, IL-12, and TNF-␣
cytokines was delivered with the gelatin sponge matrix compared with fluid-phase delivery (Fig. 6, A). Inhibition of tumor
growth was statistically significantly greater in the matrixdelivered group (P<.045, Cox proportional hazards regression
model) than in the fluid-phase and control groups, resulting in a
substantial survival benefit for those mice (Fig. 6, B).
Fig. 3. Mean luciferase activity (95% confidence interval) of heterotopic RM-1 tumor nodules after infection
by 3 × 106 plaque-forming units of ALVAC-luciferase.
Tumors (n ⳱ 5 per group) were harvested at various
times after infection. Values represent combined data
from two separate experiments. A statistically significant difference (all P values <.05, Mann–Whitney rank
sum test) was observed at each time point between the
gelatin sponge-delivered groups and the fluid-phasedelivered groups except at 96 hours (P ⳱ .058, Mann–
Whitney rank sum test).
Journal of the National Cancer Institute, Vol. 92, No. 5, March 1, 2000
ARTICLES 407
Fig. 4. Histopathology sections
of heterotopic RM-1 tumor nodule 24 hours after infection with
3 × 106 plaque-forming units of
ALVAC-green fluorescent protein (GFP) or ALVAC-␤galactosidase (lacZ). Parental
ALVAC control infected RM-1
tumor under fluorescent microscopy showed no fluorescence
(data not shown). A) Limited and
localized GFP expression demonstrated under fluorescent microscopy when ALVAC-GFP
delivered by fluid-phase injection. B) Greater gene expression
and wider distribution seen when
ALVAC-GFP vector delivered
by the gelatin sponge matrix. C)
RM-1 tumor infected with parental ALVAC control, stained with
nuclear fast red (original magnification ×63). D) ALVAC-lacZinfected tumor stained with Xgal and counterstained with
nuclear fast red. Limited ␤-galactosidase expression was seen
in needle tract after injection (arrows) of the fluid-phase product
(original magnification ×63). E)
Lower power magnification of
tumor infected with ALVAClacZ vector delivered by gelatin
sponge matrix, demonstrating
qualitatively higher gene expression with greater distribution
(original magnification ×25).
DISCUSSION
Prostate cancer is an important public health concern in the
United States and represents the most common visceral cancer
and the second leading cause of cancer deaths among men in
this country. The American Cancer Society estimates that, in
1999, approximately 179 300 new cases of prostate cancer will
be diagnosed and about 37 000 men could die of the disease (9).
Despite this enormous prevalence, management of the disease
remains controversial, and 5-year biochemical failure rates
for radical prostatectomy range anywhere from 27% (10) to
57% (11). It is imperative to develop alternative or adjuvant
treatment strategies, such as the introduction of therapeutic
genes, to better manage both clinically localized disease and
metastatic disease.
Several reports investigating prostate cancer immunotherapy
have been encouraging. Early studies by Sanda et al. (12)
showed that granulocyte–macrophage colony-stimulating factor
(GM-CSF)-transfected rat prostatic adenocarcinomas grew more
slowly than parental tumors. Subsequently, Vieweg et al. (13)
showed that IL-2-transfected rat R3327-MatLyLu prostate cancer cells also induced antitumor activity. Existing tumors had a
decreased rate of outgrowth, and protection was gained against
408 ARTICLES
subsequent tumor challenge. Using the parental R3327G tumor
that exhibits hormone responsiveness, Yoshimura et al. (14) also
observed antitumor activity; however, neither cytotoxic Tlymphocyte activity nor protection against subsequent tumor
challenge was observed. In a previous study (1), we have shown
that the ALVAC virus can efficiently infect prostate cancer cells,
produce high levels of extrinsic gene product, and induce antitumor immunity. RM-1 tumor cells were infected by ALVAC
cytokine recombinants and injected subcutaneously in the flanks
of male C57BL/6 mice. As single agents, ALVAC-IL-2,
ALVAC-IL-12, ALVAC-GM-CSF, and ALVAC-TNF-␣ were
effective in partially inhibiting tumor outgrowth. As a combination therapy of ALVAC-TNF-␣ with ALVAC-IL-2, ALVACIL-12, or ALVAC-GM-CSF, the tumor outgrowth inhibition
was optimized. Subsequent studies assessing the ability of the
ALVAC cytokine vectors to induce regression of existing tumors showed only limited effects (data not shown). Although
this system seems perfectly suited for intratumoral treatment of
established tumors, given the inability of the ALVAC vector to
replicate in mammalian species, we hypothesized that the fluidphase injection of the viral vector resulted in limited expression
and/or distribution of the gene product, as has been reported in
other models (15).
Journal of the National Cancer Institute, Vol. 92, No. 5, March 1, 2000
Fig. 5. RM-1 tumor inhibition studies. A)
Tumor growth of preestablished RM-1 subcutaneous nodules in C57BL/6 mice (n ⳱ 5
per group) infected by ALVAC virus encoding murine interleukin 2, interleukin 12, and
tumor necrosis factor-␣ (8.4 × 106 total
plaque-forming units) delivered either by
the gelatin sponge matrix or by the fluidphase product. Controls include matrix only,
parental ALVAC virus, and a no treatment
group. Data are presented as mean tumor
volumes (95% confidence intervals) for the
gelatin sponge-delivery group and the fluidphase-delivery group. The average tumor
volume in the matrix-delivered treatment
group was significantly different (P values
for all comparisons were <.05, Mann–
Whitney rank sum test) from that in the
fluid-phase-delivered group at each of the
measurement days 4–13. A one-way analysis of variance of log-transformed data was
also used to compare the average tumor volume of the matrix-delivered treatment group
with all of the other experimental groups
and was found to be significantly different
at each time point (P values for all comparisons were <.001, one-way analysis of variance). B) Mice in tumor outgrowth studies
were followed for survival and were killed if
their tumor measured more than 25 mm in
any dimension or if they became ill from the
tumor burden. Improved survival was observed in the gelatin sponge matrix treatment group (P<.005, Cox proportional hazards regression model).
Although numerous clinical gene therapy trials for prostate
cancer have been initiated and many investigators have demonstrated great promise for both tumoricidal and corrective gene
therapies, improving delivery and enhancing gene expression are
imperative. Most preclinical investigations and ongoing phase I
clinical protocols for localized prostate cancer rely on direct
intraprostatic injection for the delivery of viral vectors (16,17).
However, the relative inaccessibility of the prostate and the often
isoechoic nature of cancer foci on transrectal ultrasound make
detection and localization difficult. Moreover, examination of
autopsy and radical prostatectomy specimens has revealed multiple and separate tumor sites, suggesting the possibility of a
field change in the prostate (18,19) and necessitating even more
efficient and accurate distribution of any therapeutic vector.
Unfortunately, the efficient delivery of genetic material to
tumors remains a formidable task for all solid-tumor oncology.
Asgari et al. (20) have found significant inhibition of preestablished subcutaneous tumor nodules of human prostate cancer by
Journal of the National Cancer Institute, Vol. 92, No. 5, March 1, 2000
a single injection of the recombinant adenovirus p53 vector. It is
interesting that they were unable to detect p53 overexpression in
these adenovirus wild-type p53-infected tumors 48 hours after
intratumoral injection. Moreover, multiple injections (1 week
apart) did not improve the antitumor effect. These results most
likely reflect the poor efficiency of transfer when direct intratumoral injections are employed. These difficulties with vector
delivery are not isolated to the prostate. The effective use of
gene therapy in glial tumors of the brain is hampered by the
inefficient delivery of viral vectors. High titers of adenovirus are
required for gene transfer to gliomas (15), and the resulting
inflammation of the high inoculum decreases the efficiency of
the gene transfer. To improve delivery, Beer et al. (21) found
sustained release of recombinant adenovirus when coupled to
biodegradable microspheres, allowing the administration of
lower doses of viral vectors to glioma tissue.
Numerous methods have been investigated to better deliver
soluble products to both neoplastic and non-neoplastic cells. A
ARTICLES 409
Fig. 6. RM-11 tumor inhibition studies.
A) Tumor growth of preestablished
RM-11 subcutaneous nodules in
BALB/c mice (n ⳱ 6 per group) infected with ALVAC virus encoding
murine interleukin 2, interleukin 12,
and tumor necrosis factor-␣ (8.4 × 106
total plaque-forming units) delivered either by the gelatin sponge matrix or by
the fluid-phase product. Data are presented as mean tumor volumes (95%
confidence intervals) for the gelatin
sponge-delivery group and the fluidphase delivery group. Again, a statistically significant difference in tumor
volume was found between the matrixdelivered treatment group and the fluidphase-delivered treatment group (P values for all comparisons were 艋.036,
Mann–Whitney rank sum test) at days
3–13. B) Survival of mice from the tumor outgrowth studies demonstrates a
significant increase in the gelatin
sponge matrix treatment group. Longterm survival (tumor-free) of approximately 20% has been observed in three
separate tumor outgrowth studies.
polymer-based paste has been found to enhance local delivery of
chemotherapeutic agents and decrease recurrence rates at tumor
resection sites (3). A fibrin- and gelatin-based drug-delivery system has been shown to more slowly release and to improve the
therapeutic effect of antibiotics (22). Poloxamer 407 has been
shown to improve the delivery of adenoviral vectors in vascular
smooth muscle based on ␤-gal reporter gene expression (23).
The absorbable gelatin sponge employed in this study is primarily used as an intraoperative hemostatic agent, but it has also
been used to deliver a number of different compounds, including
insulin (2) and various cytokines and growth factors (24,25), in
order to improve and sustain delivery.
In our study, the ability of a known quantity of recombinant
ALVAC absorbed by the gelatin sponge matrix to infect tumor
cells in vitro was similar to that seen with infusion of fluid-phase
410 ARTICLES
virus into the culture medium. The retained viral particles were
able to freely efflux from the gelatin sponge matrix in this liquid
environment over a 4- to 6-hour incubation time. The other
matrices showed markedly less consistency as a delivery vehicle, most likely explained by some loss of viral particles in the
transfer of the matrix to the culture dish or the inability of the
retained virus to reenter the media. It is interesting that the
gelatin matrix mediated enhanced gene expression and also enhanced biodistribution when compared with the fluid-phase injection of the viral vector in preestablished subcutaneous tumor
nodules. The expression of the respective products of ALVACGFP or ALVAC-lacZ delivered in fluid phase was limited, i.e.,
restricted to the needle tract (Fig. 4) and around the tumor margin. Moriuchi et al. (26) have described a similar distribution of
viral vector infection after intratumoral injection of an intraceJournal of the National Cancer Institute, Vol. 92, No. 5, March 1, 2000
rebral tumor, resulting in poor gene expression and biologic
response.
In contrast to delivery in the fluid phase, the vectors delivered
by the gelatin matrix showed a broad biodistribution, which
often would extend the full breadth of small tumors (approximately equal to 0.5 cm in diameter). This enhanced gene transfer
was shown to translate into an improved biologic effect in the
heterotopic RM-1 murine prostate cancer model with the use of
a cytokine-based immunotherapy protocol. RM-1 tumors with a
mean volume of 100–500 mm3 (depending on the experiment)
were significantly inhibited when the gelatin matrix was used to
deliver ALVAC-IL-2, ALVAC-IL-12, and ALVAC-TNF-␣.
The inhibitory effects in these treated mice also resulted in a
significant survival advantage with a single injection. Despite
this dramatic increase in gene expression observed in these heterotopic tumors when viral vectors were delivered with the gelatin sponge matrix, systemic distribution of vectors was not found
to be increased over fluid-phase delivery. Selected organs
(spleen, liver, and kidneys) harvested at the time of these experiments demonstrated no increased luciferase gene expression
when compared with fluid-phase delivery. Similarly, in experiments involving orthotopic delivery of viral vectors in the ventral prostate of mice, gelatin sponge matrix delivery resulted in
only minimal gene expression in surrounding organs, including
the testes, bladder, and seminal vesicles, that was no greater and
often less than that of fluid-phase delivery (data not shown).
Although gene therapy approaches to immunotherapy have
enhanced immune activation and provided enhanced therapy results, the control of preestablished tumors has remained problematic (27–29). Control of the immunogenic RENCA kidney
tumors was limited to tumors established for a maximum of 7
days [nonpalpable tumors; (29)]. Likewise, the control of tumor
growth by herpes simplex virus thymidine kinase/gancyclovir
has been limited to a reduction in tumor growth rate but not
tumor regression (30). Our studies show the induction of substantial tumor growth inhibition with a single injection of 8.4 ×
106 pfu of ALVAC cytokine vaccine (Fig. 5, A). While the
control of tumor growth does not result in a cure, it does provide
a significant extension of survival for treated mice. This result is
accomplished in spite of the use of the poorly differentiated,
highly aggressive RM-1 tumor model (approximate in vitro doubling time of 12 hours) in which treatment was initiated when
tumors were large (approximately 215 mm3). The differences
between the two means of vector delivery were even more dramatic in the RM-11 tumor model (Fig. 6) and have resulted in
the long-term tumor-free survival in approximately 20% of mice
over three separate experiments.
Improving the delivery of viral vectors is paramount in any
clinical gene therapy trial, and this is especially true for prostate
cancer. Delivery in a solid-state matrix, such as the one that we
have described, may allow for more efficient and widespread
gene transfer for all nonreplicative vectors. The increased efficiency of delivery may also result in circumventing antiviral
host immune responses by decreasing the viral concentration
needed to attain a desired effect.
REFERENCES
(1) Kawakita M, Rao GS, Ritchey JK, Ornstein DK, Hudson MA, Tartaglia J,
et al. Effect of canarypox virus (ALVAC)-mediated cytokine expression on
murine prostate tumor growth. J Natl Cancer Inst 1997;89:428–36.
(2) Lee YC, Simamora P, Yalkowsky SH. Systemic delivery of insulin via an
enhancer-free ocular device. J Pharm Sci 1997;86:1361–4.
Journal of the National Cancer Institute, Vol. 92, No. 5, March 1, 2000
(3) Hunter WL, Burt HM, Machan L. Local delivery of chemotherapy: a
supplement to existing cancer treatments. A case for surgical pastes and
coated stents. Adv Drug Delivery Rev 1997;26:199–207.
(4) Thompson TC, Southgate J, Kitchener G, Land H. Multistage carcinogenesis induced by ras and myc oncogenes in a reconstituted organ. Cell
1989;56:917–30.
(5) Cadoz M, Strady A, Meignier B, Taylor J, Tartaglia J, Paoletti E, et al.
Immunization with canarypox virus expressing rabies glycoprotein. Lancet
1992;339:1429–32.
(6) Taylor J, Trimarchi C, Weinber, R, Languet B, Guillemin F, Desmettre P,
et al. Efficacy studies on a canarypox rabies recombinant virus. Vaccine
1991;9:190–3.
(7) Cox WI, Tartaglia J, Paoletti E. Induction of cytotoxic T lymphocytes by
recombinant canarypox (ALVAC) and attenuated vaccinia (NYVAC) virus
expressing the HIV-1 envelope glycoprotein. Virology 1993;195:845–50.
(8) SAS version 6.12, SAS/STAT user’s guide, version 6, 4th ed. Vols. 1 and
2. Cary (NC): SAS Institute, Inc.; 1995.
(9) Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics, 1999. CA
Cancer J Clin 1999;49:8–31.
(10) Ohori M, Wheeler TM, Kattan MW, Goto Y, Scardino PT. Prognostic
significance of positive surgical margins in radical prostatectomy specimens. J Urol 1995;154:1818–24.
(11) Zeitman AL, Edelstein RA, Coen JJ, Babayan RK, Krane RJ. Radical
prostatectomy for adenocarcinoma of the prostate: the influence of preoperative and pathologic findings on biochemical disease free outcome. Urology 1994;43:828–33.
(12) Sanda MG, Ayagari SR, Jeffer EM, Epstein FI, Clift SL, Cohey LK, et al.
Demonstration of a rational strategy for human prostate cancer gene
therapy. J Urol 1994;151:622–8.
(13) Vieweg J, Rosenthal FM, Bannerji R, Heston WD, Fair WR, Gansbacher B,
et al. Immunotherapy of prostate cancer in the Dunning rat model: Use of
cytokine gene modified tumor vaccines. Cancer Res 1994;54:1760–5.
(14) Yoshimura I, Heston WD, Gansbacher B, Fair WR. Cytokine mediated
immuno-gene therapy in rat prostate cancer model [abstract]. J Urol
1996;155(suppl):510A.
(15) Shewach DS, Zerbe LK, Hughes TL, Roessler BJ, Breakefield XO, Davidson BL. Enhanced cytotoxicity of antiviral drugs mediated by adenovirus
directed transfer of the herpes simplex virus thymidine kinase gene in rat
glioma cells. Cancer Gene Ther 1994;1:107–12.
(16) Ko SC, Gotoh A, Thalmann GN, Zhau HE, Johnston DA, Zhang WW, et
al. Molecular therapy with recombinant p53 adenovirus in an androgenindependent, metastatic human prostate cancer model. Hum Gene Ther
1996;7:1683–91.
(17) Timme TL, Hall SJ, Barrios R, Woo SL, Aguilar-Cordova E, Thompson
TC. Local inflammatory response and vector spread after direct intraprostatic injection of a recombinant adenovirus containing the herpes simplex
virus thymidine kinase gene and gancyclovir therapy in mice. Cancer Gene
Ther 1998;5:74–82.
(18) Miller GJ, Cygan JM. Morphology of prostate cancer: the effects of multifocality on histological grade, tumor volume and capsule penetration. J
Urol 1994;152:1709–13.
(19) Byar DP, Mostofi FK. Veterans Administration Cooperative Urologic Research Group: carcinoma of the prostate: prognostic evaluation of certain
pathological features in 208 radical prostatectomies examined by stepsection technique. Cancer 1972;30:5–13.
(20) Asgari K, Sesterhenn IA, McLeod DG, Cowan K, Moul JW, Seth P, et al.
Inhibition of the growth of pre-established subcutaneous tumor nodules of
human prostate cancer cells by single injection of the recombinant adenovirus p53 expression vector. Int J Cancer 1997;71:377–82.
(21) Beer SJ, Hilfinger JM, Davidson BL. Extended release of adenovirus from
polymer microspheres: potential use in gene therapy for brain tumor. Adv
Drug Delivery Rev 1997;27:59–66.
(22) Park MS, Kim YB. Sustained release of antibiotic from a fibrin–gelatin–antibiotic mixture. Laryngoscope 1997;107:1378–81.
(23) Feldman LJ, Pastore CJ, Aubailly N, Kearney M, Chen D, Perricaudet M,
et al. Improved efficiency of arterial gene transfer by use of poloxamer 407
as a vehicle for adenoviral vectors. Gene Ther 1997;4:189–98.
(24) Segal DH, Germano IM, Berderson JB. Effects of basic fibroblast growth
factor on in vivo cerebral tumorigenesis in rats. Neurosurgery 1997;40:
1027–33.
ARTICLES 411
(25) Watanabe M, McCormick KL, Volker K, Ortaldo JR, Wigginton JM,
Brunda MJ, et al. Regulation of local host-mediated anti-tumor mechanisms by cytokines: direct and indirect effects on leukocyte recruitment and
angiogenesis. Am J Pathol 1997;150:1869–80.
(26) Moriuchi S, Oligino T, Krisky D, Marconi P, Fink D, Cohen J, et al.
Enhanced tumor cell killing in the presence of gancyclovir by herpes simplex virus type 1 vector-directed coexpression of human tumor necrosis
factor-␣ and herpes simplex virus thymidine kinase. Cancer Res 1998;58:
5731–7.
(27) Zbar B, Rapp HJ. Immunotherapy of guinea pig cancer with BCG. Cancer
1974;34:1532–40.
(28) Fearon ER, Pardoll DM, Itaya T, Golumbek P, Levitsky HI, Simons JW, et
al. Interleukin-2 production by tumor cells bypasses T helper function in
the generation of an antitumor response. Cell 1990;60:397–403.
(29) Golumbek PT, Lazenby AJ, Levitsky HI, Jaffee LM, Karasuyama H, Baker
412 ARTICLES
M, et al. Treatment of established renal cancer by tumor cells engineered to
secrete interleukin-4. Science 1991;254:713–6.
(30) Eastham JA, Chen SH, Sehgal I, Yang G, Timme TL, Hall SJ, et al.
Prostate cancer gene therapy: herpes simplex virus thymidine kinase gene
transduction followed by gancyclovir in mouse and human prostate cancer
models. Hum Gene Ther 1996;7:515–523.
NOTES
Supported by grant 98-84 from the Carver Foundation.
We thank Jan Rodgers (Department of Pathology, University of Iowa) for her
help with histopathology and Dr. Charles Davis (Department of Biostatistics,
University of Iowa) for his assistance with statistical analysis.
Manuscript received May 6, 1999; revised November 29, 1999; accepted
December 8, 1999.
Journal of the National Cancer Institute, Vol. 92, No. 5, March 1, 2000