Induction of angiogenesis by implantation of encapsulated primary myoblasts expressing vascular...

RESEARCH ARTICLE

Induction of angiogenesis by implantation ofencapsulated primary myoblasts expressingvascular endothelial growth factor

Matthew L. Springer1

Gonzalo Hortelano2

Donna M. Bouley3

Jason Wong2

Peggy E. Kraft1

Helen M. Blau1*

1Department of MolecularPharmacology, Stanford UniversitySchool of Medicine, Stanford, CA,USA2Department of Pathology andMolecular Medicine, McMasterUniversity, Hamilton, Ontario,Canada3Department of ComparativeMedicine, Stanford University Schoolof Medicine, Stanford, CA, USA

*Correspondence to: H. M. Blau,Department of MolecularPharmacology, Stanford UniversitySchool of Medicine, Stanford,CA 94305-5174, USA.E-mail: [email protected]

Received: 24 November 1999

Revised: 4 April 2000

Accepted: 5 April 2000

Published online: 18 April 2000

Abstract

Background We previously demonstrated that intramuscular implantationof primary myoblasts engineered to express vascular endothelial growth factor(VEGF) constitutively resulted in hemangioma formation and the appearanceof VEGF in the circulation. To investigate the potential for using allogeneicmyoblasts and the effects of delivery of VEGF-expressing myoblasts to non-muscle sites, we have enclosed them in microcapsules that protect allogeneiccells from rejection, yet allow the secretion of proteins produced by the cells.

Methods Encapsulated mouse primary myoblasts that constitutivelyexpressed murine VEGF164, or encapsulated negative control cells, wereimplanted either subcutaneously or intraperitoneally into mice.

Results Upon subcutaneous implantation, capsules containing VEGF-expressing myoblasts gave rise to large tissue masses at the implantationsite that continued to grow and were composed primarily of endothelial andsmooth muscle cells directly surrounding the capsules, and macrophages andcapillaries further away from the capsules. Similarly, when injected intra-peritoneally, VEGF-producing capsules caused signi®cant localized in¯am-mation and angiogenesis within the peritoneum, and ultimately led to fatalintraperitoneal hemorrhage. Notably, however, VEGF was not detected in theplasma of any mice.

Conclusions We conclude that encapsulated primary myoblasts persist andcontinue to secrete VEGF subcutaneously and intraperitoneally, but thatthe heparin-binding isoform VEGF164 exerts localized effects at the site ofproduction. VEGF secreted from the capsules attracts endothelial and smoothmuscle cells in a macrophage-independent manner. These results, along withour previous results, show that the mode and site of delivery of the samefactor by the same engineered myoblasts can lead to markedly differentoutcomes. Moreover, the results con®rm that constitutive delivery of highlevels of VEGF is not desirable. In contrast, regulatable expression may lead toef®cacious, safe, and localized VEGF delivery by encapsulated allogeneicprimary myoblasts that can serve as universal donors. Copyright # 2000 JohnWiley & Sons, Ltd.

Keywords VEGF; myoblasts; encapsulation; gene therapy; angiogenesis

Introduction

Vascular endothelial growth factor (VEGF, VPF) is an angiogenic factor thatinduces blood vessel growth, increases vascular permeability, and controls the

THE JOURNAL OF GENE MEDICINEJ Gene Med 2000; 2: 279±288.

Copyright # 2000 John Wiley & Sons, Ltd.

formation of the embryonic vasculature [1±6, for review,see Ref. 7]. Thus, VEGF has become a prime candidate forgene delivery aimed at therapeutic angiogenesis due to itspotency and multiple clinical applications. Indeed,delivery of the VEGF gene to tissue via plasmid injectionor adenoviral infection, both of which resulted intransient expression of the gene, has been reported tocause angiogenesis in the target tissues [8±14]. Incontrast, as we previously demonstrated, intramuscularimplantation of mouse myoblasts that had been retro-virally transduced to constitutively secrete murineVEGF164 caused hemangioma formation at the implant-ation site [15], presumably because expression of thegene was strong, permanent, and highly localized withinthe muscle. This marked localized response of the tissuewas particularly striking because it occurred despite thenon-ischemic state of the muscle. Previously, VEGF-induced effects in muscle had only been reported underischemic conditions. Interestingly, VEGF was not detect-able in the bloodstream until several weeks after the onsetof the strong physiological response it had induced.

Primary myoblasts are well suited to muscle-mediatedgene therapy because of their ability to fuse into pre-existing muscle ®bers, resulting in a transfer of geneti-cally-altered nuclei directly into pre-existing muscle[16,17, for reviews, see 18,19]. This form of gene therapycurrently requires syngeneic myoblasts and a musclemilieu. To examine the potential for implanting allo-geneic primary myoblasts in non-muscle sites, we haveenclosed VEGF-expressing primary myoblasts in perme-able alginate microcapsules prior to implantation eithersubcutaneously (s.c.) or intraperitoneally (i.p.). The useof such capsules allows the enclosed cells to secreterecombinant proteins for systemic delivery while beingprotected from host immune responses [20±22]. Further-more, as previously shown for cell lines, myoblasts maybe particularly well-suited to encapsulation due to theirability to differentiate and remain viable for many monthsinside the capsules, in contrast to other cell types like®broblasts that overgrow and die in this environment[23±25]. Encapsulated myoblast cell lines are thus able topersist within the capsules in non-muscle sites. We reporthere the ®rst use, to our knowledge, of encapsulatedprimary myoblasts as opposed to myoblast cell lines. The®ndings suggest that the use of allogeneic, universaldonor primary myoblasts that evade the immuneresponse, may be possible when delivered using capsules.

In this report, we describe the results of implantingprimary myoblasts stably expressing VEGF within cap-sules either under the skin or in the peritoneal cavity. Weshow that such implantation results in potent neovascu-larization and in¯ammation, as well as recruitment ofendothelial and smooth muscle cells in a mannerindependent of macrophages. The lack of detectableVEGF in the plasma, taken together with our previousobservation that VEGF produced in muscle did notinitially appear in the blood [15], suggests that in bothcases VEGF164 binds and exerts its effects primarily at thesite of synthesis. These results con®rm the ability of

recombinant VEGF secreted from encapsulated primarymyoblasts to cause a marked physiological response,while underscoring the need for regulation of levels ofVEGF delivery. Such studies of potential deleteriouseffects are especially important now that VEGF genedelivery is the focus of several clinical trials. These resultsalso demonstrate that the same VEGF-expressing primarymyoblasts induce different physiological effects depend-ing on the site and mode of implantation.

Materials and Methods

Encapsulation

Alginate microcapsules were made following a previouslydescribed protocol [26], with modi®cations describedby Hortelano et al. [24]. In brief, cells were mixed with1.5 potassium alginate (Improved Kelmar, Kelco Inc.,Chicago, IL, USA) and extruded through a 27 G needlewith the aid of an air jet. The ®ne droplets produced weregelled upon contact with 1.1% CaCl2. Alginate beads werecoated with poly-L-lysine, and a second layer of alginate.Finally, the internal alginate core was dissolved with0.055 M sodium citrate to form the microcapsules.

ELISA analyses

ELISA analyses of cell culture supernatants and of mouseserum were carried out using a commercially availableanti-mouse VEGF ELISA kit (R&D Systems, Minneapolis,MN, USA) according to the manufacturer's instructions.

Implantation

Male SCID C.B-17 mice (Taconic, Germantown, NY, USA)were anesthetized with Metofane (methoxy¯urane) priorto the treatment. Microcapsules were washed in sterilephosphate-buffered saline (PBS) and allowed to settle inan upright syringe. For i.p. injections, the packed capsuleswere implanted in the peritoneal cavity with an i.v.catheter (18 G, Angiocath, Becton Dickinson, FranklinLakes, NJ, USA). For s.c. injections, microcapsules werelodged in a single subcutaneous pocket made with thecatheter. All mice recovered quickly from the anesthesiaand were soon mobile in their cages. All experimentsinvolving animals were performed in accordance with theguidelines of the Stanford University AdministrativePanel on Laboratory Animal Care.

Histology

Complete necropsies were performed on carbon dioxide-killed mice. In some cases, tissues were ®xed in 10%buffered neutral formalin for paraf®n embedding.Samples were routinely processed, paraf®n embedded,and 5 mm thick sections were cut and stained withhematoxylin and eosin (H&E). In other cases, tissues weresnap frozen in freezing isopentane, 10 mm thick sections

280 M. L. Springer et al.

Copyright # 2000 John Wiley & Sons, Ltd. J Gene Med 2000; 2: 279±288.

were cut with a cryostat, and subsequent histochemicalstaining was performed with H&E as previously described[27]. Unimplanted capsules in culture were stained withX-gal [27].

For immuno¯uorescent staining, 5 mm paraf®n sectionsof formalin-®xed tissue were deparaf®nized and re-hydrated by standard methods, or 10 mm frozen sectionswere ®xed in 1.5% formaldehyde for 15 min. Sectionswere then blocked with 0.5% casein in staining bufferconsisting of 2% normal goat serum and 0.3% Triton X-100 in PBS for 30 min. The slides were incubated for 2 hat room temperature with different combinations of thefollowing primary antibodies in staining buffer: a ratmonoclonal against mouse PECAM-1 (clone MEC 13.3;PharMingen, San Diego, CA,USA; 1 : 100 dilution), amouse monoclonal against mouse smooth muscle actin(clone 1A4; ICN Biomedicals, Aurora, OH, USA; 1 : 400dilution), a rat monoclonal against F4/80 antigen(Caltag, Burlingame, CA, USA; 1 : 20 dilution), and arabbit polyclonal against b-galactosidase (Eppendorf - 5Prime, Inc., Boulder, CO, USA; 1 : 400 dilution). Negativecontrols lacking primary antibody were always per-formed. Sections were rinsed in staining buffer, andthen incubated for 1±1.5 h at room temperature withAlexa488-conjugated goat anti-rabbit or goat anti-mousesecondary antibody at 1 : 200 dilution, or Texas Red-X-conjugated goat anti-rat secondary antibody at 1 : 100dilution (Molecular Probes, Eugene, OR, USA) in stainingbuffer also containing the nuclear dye Hoechst 33258.The slides were then rinsed and mounted.

Results

Expression of LacZ and VEGF genesafter encapsulation

Primary myoblasts that were retrovirally transduced withthe E. coli LacZ gene (control cells) and myoblaststransduced with both the LacZ and murine VEGF164 genes(VEGF cells) [15] were encapsulated in alginate andallowed to differentiate and fuse inside the capsules in amanner previously demonstrated to result in the long-term survival of a myogenic cell line [24]. Other cellsfrom the same myoblast population had previously beenshown to express their transgenes in culture and afterimplantation into mouse skeletal muscle [15]. To con®rmthat the differentiated myoblasts continued to express thetransgenes in culture after encapsulation, control cellcapsules were stained with X-gal 5 days after encapsula-tion to detect b-galactosidase (b-gal) activity, and theculture medium was assayed by ELISA to determine VEGFlevels. After staining with X-gal, the control cell capsulesdisplayed numerous dark blue-stained cells suspended intheir interior, as seen in Figure 1. Very few non-stainingcells were observed. The number of cells in capsules wasapproximated, since this value was determined at thetime of encapsulation, and the extent of cell death thatoccurred following encapsulation is unknown. ELISA

measurements of VEGF in the culture media from controland VEGF cell capsules revealed no VEGF in the mediumof controls, whereas VEGF cell capsules secreted approxi-mately 100 ng/106 cells/day into the medium.

Subcutaneous implantation ofencapsulated cells

To determine the physiological response to localized s.c.implantation of encapsulated VEGF-expressing cells,500 ml of control or VEGF capsule suspension wasimplanted under the skin of the abdomen (n=4 control,4 VEGF). Since the differentiated, multinucleate cellscontained therein were derived from 2r106 pre-differentiation myoblasts, the implanted capsules coulddeliver up to 200 ng VEGF/day. A sample mouse fromeach group was killed and analyzed after 14 days(Figure 2). In the control mouse, the capsules remainedlocalized as two distinct three-dimensional circularmasses (9r7r3 mm and 5r5r2 mm, respectively)under the skin and were associated with moderatein¯ammation. Over the course of the next 60 days, theremaining control mice exhibited similar tissue massesthat did not change size during that time. The VEGF cellcapsules also remained localized, but led to regions withmore severe in¯ammation associated with excessiveamounts of blood. These tissue masses grew steadily,achieving 14r11r8 mm in the sample mouse killed at14 days, and were accompanied by engorged subcuta-neous blood vessels. The masses continued to grow in theother mice harboring VEGF-producing capsules until theybegan to interfere with ambulation, at which time themice were killed (day 25). Thorough necropsies wereperformed on these mice by one of the authors (D.M.B),a board-certi®ed veterinary pathologist. Necropsiesrevealed no observable abnormalities in most internalorgans, although the spleens were enlarged in the micekilled on day 25, due to increased hematopoiesis in thered pulp. Extramedullary hematopoiesis is considered anormal ®nding in the spleens of mice, but may beincreased in cases of hemorrhage, or as seen in thesemice, blood pooling in an abnormal, peripheral site.

To determine the nature of the tissue surrounding theimplanted capsules, samples were isolated for histologicalanalysis from the control and VEGF tissue masses shownin Figure 2 (see Figures 3 and 4). Hematoxylin and eosin(H&E) staining of paraf®n-embedded tissue sections fromcontrol samples revealed a relatively small amountof subcutaneous in¯ammation around the capsules(Figure 3A,B). Similar analyses of sections containingencapsulated myoblasts expressing VEGF revealed a muchgreater extent of in¯ammation surrounding the capsules,along with engorged blood vessels, neovascularization,and localized pools of blood (Figure 3C±E).

Figure 4 shows the results of immunohistochemicalstudies performed on both paraf®n-embedded and frozentissue sections derived from similar s.c. implantation sitesto those shown in Figure 3, using antibodies againstPECAM-1 (endothelial cells), smooth muscle actin

Encapsulated VEGF-expressing Myoblasts 281


(smooth muscle cells), and F4/80 (macrophages). The

paraf®n sections, which exhibited superior morphology,

stained successfully only for smooth muscle actin

(Figure 4A,B). Immunostaining of frozen tissue sectionsshowed that VEGF cell capsules were directly surroundedby endothelial cells and smooth muscle cells (Figure4C±E), whereas macrophages were generally excludedfrom the region directly next to the capsules and wereobserved chie¯y in the regions further from the capsules(Figure 4F,G). In contrast, the small amount of in¯am-mation around the control cell capsules did not containsigni®cant numbers of endothelial cells and smoothmuscle cells (Figure 4H±J), but did contain macrophages.Further immunostaining with antibodies against b-gal

Figure 1. Encapsulated myoblasts expressing b-gal in culture.Encapsulated control myoblasts expressing b-gal were stainedin culture with X-gal, 5 days after encapsulation, revealingthat most of the differentiated myoblasts still expressed theLacZ gene. Capsule outlines can be faintly seen surroundingblue-stained cells. The capsules have been ¯attened under acover slip in this picture; the actual average diameter of thecapsules was slightly smaller. Bar =500 mm

Figure 2. Subcutaneous implantation of encapsulated myo-blasts. In¯amed tissue masses resulting from implantation ofcontrol cell capsules did not grow (arrows in left panel),whereas the tissue associated with implantation of VEGF cellcapsules continued to grow and became engorged with blood(middle panel; right panel shows side view). Bar =5 mm

Figure 3. Histological examination of subcutaneously implanted capsules and surrounding in¯amed tissue. Short yellow arrowsindicate capsules or regions that capsules had occupied before tissue processing. Long white arrows indicate blood vessels orblood pools. (A,B) Control cell capsule implantation sites and (C±E) VEGF cell capsule implantation sites after ®xation, paraf®nsectioning, and H&E staining. As seen in Figure 2, in¯ammation accompanies the implantation of both control and VEGF cellcapsules, but a much greater extent of in¯ammation is associated with the VEGF cell capsule implantation. Note in particularthe larger vessels (C) and blood pools (E) associated with the VEGF cell capsule implantation. Bar =100 mm



con®rmed that transgene expression was still strong inthe encapsulated cells after implantation (Figure 4L).

Intraperitoneal implantation ofencapsulated cells

Because i.p. implantation of encapsulated genetically-engineered myoblasts is a current strategy to deliverrecombinant therapeutic proteins to the entire body [24],we assessed the effects of i.p. implantation of VEGF-expressing encapsulated cells. 500 ml of control or VEGFcell capsules, containing the differentiated equivalent ofy2r106 myoblasts as described above, were implantedi.p. into groups of four control or four experimental mice.One mouse from each group was killed and analyzed atday 14 post-implantation and the remaining mice wereobserved for several weeks thereafter. Representativeanimals from each group are shown in Figure 5. Visceralorgans of the mice implanted with control cell capsuleswere normal in mice examined at days 14 and 55 post-implantation. Implantation of VEGF cell capsules causedmarked in¯ammation and neovascularization in regionsof the peritoneum by day 14, primarily around the testesand epididymi, in the mouse shown in Figure 5. A secondmouse implanted with VEGF cell capsules died on day 24after anesthetization with Metofane and venupuncturefor a small blood sample (100 ml). A third mouse similarlydied after being bled on day 38, and the fourth mouse waskilled on that day. All three of the day 24 and 38 mice haddeveloped hemoperitoneum (abdomens full of blood) andhad markedly enlarged spleens. Reddish-brown tissuemasses that oozed blood upon incision were found invarious parts of the viscera, with discrete masses up to1 cm in diameter. Additionally, adhesions were observedin some cases between viscera and the abdominal wall.

Histological examination of paraf®n-embedded sec-tions taken from the VEGF mouse pictured in Figure 5 areshown in Figure 6 at different magni®cations. The regionbetween and surrounding the testis and epididymis was amass of in¯ammatory cells, blood vessels, and bloodpools. Masses of proliferating cells were also apparent insome cases. Histological sections from the correspondingcontrol mice showed normal testes and epididymi with noextra tissue, in agreement with the ®ndings in Figure 5(data not shown). Further microscopic examination of theother VEGF mice at days 24 and 38 showed similar, butmore extreme, effects of VEGF on the viscera. Cellularproliferation occurred primarily around the testes, colon,and urinary bladder, and in some cases around thekidneys. These tissue masses, like those shown inFigure 6, consisted of cystic spaces containing capsulesthat were frequently lined by slightly ¯attened cuboidalcells, surrounded in turn by accumulations of denselypacked plump spindle cells, red blood cells, neutrophils,and mononuclear cells. The masses were vascularized andcontained pools of blood, with features consistent inappearance with hemangiosarcomas. Immunostaining ofparaf®n sections revealed that the VEGF cell capsuleswere surrounded by smooth muscle cells as they had been

in the s.c. implantation (data not shown); however,further characterization of cell types was not possible asfrozen tissue samples were not taken at necropsy of thesemice. All mice implanted i.p. with VEGF cell capsulesexhibited this physiological response.

Blood plasma levels of VEGF followings.c. and i.p. implantation

Blood samples were taken from all of the mice implanteds.c. and i.p. with VEGF-producing capsules, and also fromthe mice implanted s.c. with control capsules, on days 0,3, 10, 24, and 38 after implantation. Plasma was isolatedand assayed by ELISA for the presence of VEGF. Despitethe intense physiological response to the VEGF capsuleimplantation, there was no signi®cant increase in VEGFlevels in the experimental groups compared to the controlgroups (n=4 per group, p<0.05 at all time points).

Discussion

VEGF gene delivery shows promise as a potentiallytherapeutic approach; however, it can be deleterious ifthe levels of VEGF are too high. Thus, it is important tounderstand the physiological consequences of VEGFdelivery by different modes of introduction and atdifferent levels. We show here that the site and themeans of delivery of VEGF in¯uence the physiologicaleffect of this potentially therapeutic protein. We alsoshow that VEGF-secreting capsules can induce theaccumulation of endothelial cells and smooth musclecells, and that in¯ammation and the recruitment ofmacrophages alone is not suf®cient to cause this effect, asevidenced by the presence of macrophages in both VEGFand control s.c. implantation sites (see Figure 4F,G,K).The endothelial and smooth muscle cells were mostlyobserved directly surrounding the capsules, effectivelyexcluding the macrophages, further suggesting that theability of VEGF to recruit the endothelial and smoothmuscle cells is not dependent on the local presence ofmacrophages. While some of those endothelial cells mayhave represented capillaries, there were large regions ofsmooth muscle cells not associated with endothelial cellsthat were thus not vascular in nature. Therefore, VEGFsecreted by encapsulated myoblasts in non-muscle siteswas able to recruit multiple cell types in agreement withthe results that we previously reported in skeletal muscle[15], but the organization of these cells differed.

We had previously demonstrated that constitutivelocalized VEGF gene expression in muscle caused anextreme physiological response consisting of an accumu-lation of endothelial cells and the formation of largehemangiomas [15]. This most likely occurred because themuscle was implanted with primary myoblasts thatexpressed the VEGF gene constitutively from a strongretroviral promoter, in contrast to the intramuscularinjection of plasmid DNA, which has resulted in transientgene expression and the formation of normal blood



vessels [8]. Intramuscular implantation of genetically-

altered primary myoblasts is a very ef®cient gene delivery

technique because the myoblasts fuse directly into pre-

existing muscle ®bers, which are well-vascularized multi-

nucleate cells. The donor nuclei can enlist the biosyn-

thetic machinery of the entire muscle ®ber, which then

stably produces the recombinant protein. This approach

lends itself either to protein production in the muscle or

protein delivery to the bloodstream [15±17,28±33]. Other

advantages include the facility by which primary myo-

blasts are obtained and genetically engineered. Myoblasts

can be isolated and puri®ed with relative ease from

various species including human [27,34±36, W.E. Blanco-

Bose et al., manuscript in preparation], and can be

retrovirally transduced to express recombinant genes at

almost 100% ef®ciency before being implanted [37].

However, this approach requires that the myoblasts are

syngeneic, and requires the target tissue to be skeletal

muscle.To determine whether primary myoblasts that express

VEGF can elicit such a robust response in a non-skeletal

muscle environment, in which fusion into muscle ®bers

does not occur, we tested a means of cell encapsulation

for delivery to ectopic sites. We have shown that when

VEGF-expressing skeletal myoblasts are implanted into

the myocardium, they persist without fusing to the

cardiac muscle, and induce hemangioma formation

reminiscent of that observed in skeletal muscle [38].

For non-muscle implantation (s.c. and i.p.), we prepared

the cells by enclosing them in alginate microcapsules.

Such immunoisolation devices can improve the viability

of the transplanted recombinant cells by preventing the

elimination, by the cellular immune response, of allo-

geneic or inappropriately placed cells, thus allowing the

use of non-autologous, `universal' donor cells [22,39,40].

For example, polyether-sulfone macroscopic capsules

containing engineered myoblasts have been used to

deliver functional erythropoietin to mice [41], and

alginate microcapsules like those used in this study

have been used to deliver such therapeutic proteins as

blood clotting factors and growth hormones [20,21,

23±25]. Indeed, when microcapsules containing mouse

C2C12 myoblasts were injected into the peritoneal cavity

of mice, the viability of the enclosed cells in microcapsules

retrieved 7 months post-implantation was maintained

[24]. The myoblasts did not proliferate signi®cantly

within the alginate microcapsules over that period of

time. Unlike other types of cells, which can overgrow and

die within the capsules, the myoblasts tend to differ-

entiate, persisting as multinucleate `myoballs' such as that

clearly visible at the upper left of Figure 6D. The current

report constitutes the ®rst use of primary myoblasts in

capsules. Primary myoblasts provide the additional

advantage that occasional capsule rupture will not

result in tumor formation, which would be a potential

problem if cell lines were used [35]. Our results suggest

that encapsulated primary myoblasts may be useful for

the long-term delivery of therapeutic products in vivo in

non-muscle tissue.It is notable that the VEGF cell capsules induced

substantial in¯ammation, blood pooling, and vessel

growth after both s.c. and i.p. implantation. In contrast,

control cell capsules did not trigger an effect other than

mild in¯ammation at s.c. implantation sites. This mild

in¯ammation may have been due to the large quantity of

capsules concentrated in a single subcutaneous pocket, as

the same number of control cell capsules did not cause

gross or microscopic pathology when spread throughout

the peritoneum. Moreover, it is unlikely that the

in¯ammation observed was due to the alginate capsule

material alone, because mice implanted subcutaneously

with the same number of alginate microcapsules deliver-

ing a different protein, BMP-2, do not show obvious signs

of in¯ammation (Y. Zilberman, G. Hortelano, and D.

Gazit, unpublished results). BMP-2 is a heparin binding

protein, as is VEGF, so this observation also rules out the

possibility that the robust in¯ammatory response to the

VEGF-secreting capsules was due simply to VEGF's

heparin binding activity. However, we cannot discount

the possibility that the neovascularization seen was

Figure 4. Identi®cation of cell types responding to VEGF secreted from capsules. All panels show s.c. capsule implantation sitessimilar to those shown in Figure 3. A few representative capsules in A±G are identi®ed by asterisks (*); all capsules in H±L areidenti®ed in this manner. (A,B) Paraf®n section from a s.c. VEGF cell capsule implantation site, showing smooth muscle (SM)actin localization both alone and with Hoechst 33258 nuclear stain as a double exposure. Smooth muscle cells surround thecapsules but are relatively absent from the in¯ammation further from the capsules. Arrows point to two representative bloodvessels. (C±E) Frozen section from a VEGF cell capsule implantation site, stained to visualize PECAM-1 and SM actin, andshowing both markers individually and as a double exposure (areas of red and green overlap are yellow). The presence ofcells that are predominantly red or green as opposed to yellow con®rms that the two cell types are distinct populations.Again, large regions such as that at the top of the panels are relatively free of endothelial and SM cells. (F,G) Two differentregions of a frozen section from a VEGF cell capsule implantation site, immunostained to visualize the macrophage antigenF4/80. Macrophages are observed mainly in the region that was free of endothelial and SM cells, and are generally absentfrom the region directly next to the capsule (white bracket in F). Panel G shows a double exposure of F4/80 and Hoechststaining to further illustrate the exclusion of the macrophages from the region around the capsules. (H±J) Frozen section froma control cell capsule implantation site, stained with PECAM and SM actin antibodies as well as Hoechst dye. Aside from somenon-speci®c staining and possibly some capillaries, there is no endothelial or SM actin staining around the control capsules.(K) Frozen section from a control cell capsule implantation site, stained with antibodies to F4/80 as well as Hoechst dye,showing that macrophages are present in the in¯ammation around control capsules as well as those containing VEGF cells.(L) Frozen section from VEGF cell capsule implantation site, immunostained to visualize b-galactosidase (b-gal). Two capsulesare shown, each containing brightly-stained myoblasts/myoballs, con®rming continuation of transgene expression afterimplantation. Bar =100 mm for all panels



caused by the VEGF-induced in¯ammation, rather than

being caused directly by VEGF. Macrophages comprised a

substantial component of the cellular response to

intramuscular implantation of VEGF-expressing myo-

blasts [15]. As we pointed out in that study, macrophages

have receptors for VEGF [42], and can produce angio-

genic factors themselves [43]. Moreover, the pro-

angiogenic and pro-in¯ammatory effects of VEGF are

tied together by the increase in vascular permeability

caused by VEGF, also known as vascular permeability

factor [44]. However, the macrophages and slight

in¯ammation surrounding the control capsules in this

study were associated with a small number of blood

vessels but were not associated with the dramatic effects

elicited by VEGF cell capsules. The drastic nature of the

response to VEGF at all sites further reinforces our

previous conclusions that VEGF gene delivery that is both

ef®cient and safe will require controlled gene regulation

[15,19,38].It is interesting that while implantation of non-

encapsulated VEGF-expressing myoblasts into skeletal

or cardiac muscle led to hemangioma formation [15,38],

the s.c. or i.p. implantation of replicate VEGF-expressing

myoblasts in capsules led primarily to blood vessel

formation, extensive accumulation of endothelial and

smooth muscle cells, and macrophage-rich in¯ammation.

The explanation for this difference may lie in the

fundamental distinction between the VEGF source resid-

ing within the target tissue or being adjacent to it. Our s.c.

and i.p. implantations involved the placement of perma-

nent VEGF sources adjacent to the body wall and visceral

organs, respectively. This led to recruitment of endo-

thelial and smooth muscle cells and in¯ammation similar

to that observed in our earlier leg implantation experi-

ments, but resulted in vessel growth and in¯ammation

Figure 6. Histological examination of intraperitoneal VEGF-producing capsules. (A) Low magni®cation view of peritesticularregion of 14 day post-implantation VEGF mouse shown in Figure 5. (B±D) Higher magni®cation views of different regions of(A). In¯ammation and free blood surround the capsules. The corresponding regions of mice implanted with control cell cap-sules showed only normal testes and epididymi with no additional tissue (not shown). Key to symbols is as follows: cap =capsules, ep = epididymis, te = testis, bl = blood, p.c. = proliferating cells. Bars = (A) 1 mm, (B) 500 mm, (C,D) 100 mm

Figure 5. Intraperitoneal implantation of encapsulated myo-blasts. On the left is a representative mouse implanted withcontrol cell capsules, shown at 55 days post-implantation.The control mice were normal when necropsied on days 14and 55. On the right is a mouse implanted with VEGF cellcapsules, shown at day 14. The region around the testes (T)and epididymi is normal in the control mouse but is sur-rounded by severe in¯ammation and neovascularization inthe VEGF mouse (arrows)



with some blood pooling as opposed to the substantialhemangiomas seen in the leg. In contrast, our previousimplantations of non-encapsulated VEGF-producing pri-mary myoblasts into both skeletal and cardiac muscleresulted in the constant delivery of VEGF within the solidmuscle tissue, from which it could not easily dissipate.The fact that this hemangioma response was not observedafter injection of plasmid DNA or adenovirus into muscle[8,12] may be explained by the transient nature of thosegene delivery methods. These observations suggest thatconstant delivery of VEGF to a tissue may result indifferent effects depending not only on the identity of thetissue, but also on whether the source is within oradjacent to the tissue.

It is striking that neither i.p. nor s.c. implantation of thecapsules resulted in VEGF accumulation in the blood-stream, despite the signi®cant local physiologicalresponses. This observation was reminiscent of whatoccurred following injection of non-encapsulated VEGF-producing primary myoblasts into muscle [15], in whichVEGF was undetectable until the hemangiomas hadbecome quite large. The VEGF detected in the serummay have resulted from tissue damage and leakage out ofthe hemangiomas rather than active secretion. The factthat VEGF164 delivered from within muscle, under theskin, and within the peritoneum was not available to theblood suggests that it was quickly bound at its site ofproduction. The VEGF receptors (¯t-1, ¯k-1, and neuro-pilin [7]) may have played a role, but it is more likely thatthe VEGF was sequestered by heparan sulfate in theextracellular matrix. The VEGF delivered in our currentand previous studies was the 164 amino acid mouseisoform [45], which corresponds to the 165 amino acidhuman isoform, and is expected to be partially heparinbinding [46,47]. Although the shorter, non-heparin-binding isoform 121 of human VEGF has been used toinduce angiogenesis in the heart using an adenoviralvector [12], the VEGF164 isoform is the most commonlyused. When Takeshita et al. [9] compared delivery ofcDNA expression vectors encoding the three majorisoforms of human VEGF to the arterial wall in rabbits,they observed no difference in their ability to induceangiogenesis. However, elegant work with transgenicmice has shown that there are indeed functionaldifferences between the VEGF isoforms [48], and thereis almost certainly a difference in the ability of thedifferent isoforms to migrate from a solid tissue sourcedue to their differential af®nity for heparin.

We conclude from this study that VEGF164 can be stablyproduced and secreted from encapsulated primarymyoblasts in a biologically active form. The encapsulatedcells did not die but instead persisted at non-muscle sitesand delivered VEGF in a highly localized fashion, withoutsubsequent appearance of VEGF in the bloodstream. Theeffects of VEGF delivered in this fashion withoutregulation of the transgene were deleterious, includingin¯ammation and the uncontrolled accumulation ofendothelial and smooth muscle cells. However, withappropriate regulation of gene expression, the ability to

use primary muscle cells as universal donor cells makesthis a potentially useful approach for the local delivery ofVEGF or other proteins at non-muscle sites.

Acknowledgements

We thank Nong Xu and Desiree Monteiro for excellent technical

support. This work was supported by NIH grants AG09521,

HD18179, and CA59717 to HMB. GH is a Bayer/CBS/MRC

Scholar with the Canadian Blood Services.

References

1. Senger DR, Galli SJ, Dvorak AM, Perruzzi CA, Harvey VS,Dvorak HF. Tumor cells secrete a vascular permeability factorthat promotes accumulation of ascites ¯uid. Science 1983; 219:983±985.

2. Ferrara N, Henzel WJ. Pituitary follicular cells secrete a novelheparin-binding growth factor speci®c for vascular endothelialcells. Biochem Biophys Res Commun 1989; 161: 851±855.

3. Leung DW, Cachianes G, Kuang WJ, Goeddel DV, Ferrara N.Vascular endothelial growth factor is a secreted angiogenicmitogen. Science 1989; 246: 1306±1309.

4. PloueÈt J, Schilling J, Gospodarowicz D. Isolation and character-ization of a newly identi®ed endothelial cell mitogen producedby AtT-20 cells. EMBO J 1989; 8: 3801±3806.

5. Conn G, Soderman DD, Schaeffer MT, Wile M, Hatcher VB,Thomas KA. Puri®cation of a glycoprotein vascular endothelialcell mitogen from a rat glioma-derived cell line. Proc Natl AcadSci U S A 1990; 87: 1323±1327.

6. Wilting J, Christ B, Weich HA. The effects of growth factors onthe day 13 chorioallantoic membrane (CAM): a study ofVEGF165 and PDGF-BB. Anat Embryol 1992; 186: 251±257.

7. Neufeld G, Cohen T, Gengrinovitch S, Poltorak Z. Vascularendothelial growth factor (VEGF) and its receptors. FASEB J1999; 13: 9±22.

8. Tsurumi Y, Takeshita S, Chen D, Kearney M, Rossow ST, PasseriJ, Horowitz JR, Symes JF, Isner JM. Direct intramuscular genetransfer of naked DNA encoding vascular endothelial growthfactor augments collateral development and tissue perfusion.Circulation 1996; 94: 3281±3290.

9. Takeshita S, Tsurumi Y, Couf®nahl T, Asahara T, Bauters C,Symes J, Ferrara N, Isner JM. Gene transfer of naked DNAencoding for three isoforms of vascular endothelial growthfactor stimulates collateral development in vivo. Lab Invest 1996;75: 487±501.

10. Isner JM, Pieczek A, Schainfeld R, Blair R, Haley L, Asahara T,Rosen®eld K, Razvi S, Walsh K, Symes JF. Clinical evidence ofangiogenesis after arterial gene transfer of phVEGF165 inpatient with ischaemic limb. Lancet 1996; 348: 370±374.

11. Magovern CJ, Mack CA, Zhang J, Rosengart TK, Isom OW,Crystal RG. Regional angiogenesis induced in nonischemic tissueby an adenoviral vector expressing vascular endothelial growthfactor. Hum Gene Ther 1997; 8: 215±227.

12. Mack CA, Patel SR, Schwarz EA, Zanzonico P, Hahn RT, Ilercil A,Devereux RB, Goldsmith SJ, Christian TF, Sanborn TA, KovesdiI, Hackett N, Isom OW, Crystal RG, Rosengart TK. Biologicbypass with the use of adenovirus-mediated gene transfer of thecomplementary deoxyribonucleic acid for vascular endothelialgrowth factor 121 improves myocardial perfusion and functionin the ischemic porcine heart. J Thorac Cardiovasc Surg 1998;115: 168±177.

13. Baumgartner I, Pieczek A, Manor O, Blair R, Kearney M, WalshK, Isner JM. Constitutive expression of phVEGF165 afterintramuscular gene transfer promotes collateral vessel develop-ment in patients with critical limb ischemia. Circulation 1998;97: 1114±1123.

14. Losordo DW, Vale PR, Symes JF, Dunnington CH, Esakof DD,Maysky M, Ashare AB, Lathi K, Isner JM. Gene therapy formyocardial angiogenesis: initial clinical results with directmyocardial injection of phVEGF165 as sole therapy formyocardial ischemia. Circulation 1998; 98: 2800±2804.

15. Springer ML, Chen AS, Kraft PE, Bednarski M, Blau HM. VEGF



gene delivery to muscle: potential role for vasculogenesis inadults. Mol Cell 1998; 2: 549±558.

16. Barr E, Leiden JM. Systemic delivery of recombinant proteins bygenetically modi®ed myoblasts. Science 1991; 254: 1507±1509.

17. Dhawan J, Pan LC, Pavlath GK, Travis MA, Lanctot AM, BlauHM. Systemic delivery of human growth hormone by injectionof genetically engineered myoblasts. Science 1991; 254:1509±1512.

18. Blau HM, Springer ML. Muscle-mediated gene therapy. New EnglJ Med 1995; 333: 1554±1556.

19. Ozawa CR, Springer ML, Blau HM. Ex vivo gene therapy usingmyoblasts and regulatable retroviral vectors. In Gene Therapy:Therapeutic Mechanisms and Strategies, Templeton NS, Lasic DD(eds). Marcel Dekker, Inc., New York, in press.

20. Liu HW, Ofosu FA, Chang PL. Expression of human factor IX bymicroencapsulated recombinant ®broblasts. Hum Gene Ther1993; 4: 291±301.

21. Chang PL, Shen N, Westcott AJ. Delivery of recombinant geneproducts with microencapsulated cells in vivo. Hum Gene Ther1993; 4: 433±440.

22. Chang PL, Van Raamsdonk JM, Hortelano G, Barsoum SC,MacDonald NC, Stockley TL. The in vivo delivery of heterologousproteins by microencapsulated recombinant cells. TrendsBiotechnol 1999; 17: 78±83.

23. al-Hendy A, Hortelano G, Tannenbaum GS, Chang PL. Correc-tion of the growth defect in dwarf mice with nonautologousmicroencapsulated myoblasts ± an alternate approach to somaticgene therapy. Hum Gene Ther 1995; 6: 165±175.

24. Hortelano G, al-Hendy A, Ofosu FA, Chang PL. Delivery ofhuman factor IX in mice by encapsulated recombinantmyoblasts: a novel approach towards allogeneic gene therapyof hemophilia B. Blood 1996; 87: 5095±5103.

25. Hortelano G, Xu N, Vandenberg A, Solera J, Chang PL, Ofosu FA.Persistent delivery of factor IX in mice: gene therapy forhemophilia using implantable microcapsules. Hum Gene Ther1999; 10: 1281±1288.

26. Chang PL, Hortelano G, Tse M, Awrey D. Growth of recombinant®broblasts in alginate microcapsules. Biotechnol Bioeng 1994;43: 925±933.

27. Springer ML, Rando TA, Blau HM. Gene delivery to muscle. InCurrent Protocols in Human Genetics, Boyle AL (ed.). John Wiley& Sons: New York, 1997; 13.14.11±13.14.19.

28. Dai Y, Roman M, Naviaux RK, Verma IM. Gene therapy viaprimary myoblasts: Long-term expression of factor IX proteinfollowing transplantation in vivo. Proc Natl Acad Sci U S A 1992;89: 10892±10895.

29. Yao S-N, Smith KJ, Kurachi K. Primary myoblast-mediated genetransfer: persistent expression of human factor IX in mice. GeneTher 1994; 1: 99±107.

30. Hamamori Y, Samal B, Tian J, Kedes L. Persistent erythropoiesisby myoblast transfer of erythropoietin cDNA. Hum Gene Ther1994; 5: 1349±1356.

31. Vilquin JT, Kinoshita I, Roy B, Goulet M, Engvall E, Tome F,Fardeau M, Tremblay JP. Partial laminin alpha2 chain restora-tion in alpha2 chain-de®cient dy/dy mouse by primary musclecell culture transplantation. J Cell Biol 1996; 133: 185±197.

32. Bohl D, Naffakh N, Heard JM. Long-term control of erythro-poietin secretion by doxycycline in mice transplanted withengineered primary myoblasts. Nat Med 1997; 3: 299±305.

33. Gussoni E, Blau HM, Kunkel LM. The fate of individualmyoblasts after transplantation into muscles of DMD patients.Nat Med 1997; 3: 970±977.

34. Webster C, Pavlath GK, Parks DR, Walsh FS, Blau HM. Isolationof human myoblasts with the ¯uorescence-activated cell sorter.Exp Cell Res 1988; 174: 252±265.

35. Rando T, Blau HM. Primary mouse myoblast puri®cation,characterization, and transplantation for cell-mediated genetherapy. J Cell Biol 1994; 125: 1275±1287.

36. Rando TA, Blau HM. Methods for myoblast transplantation.Methods Cell Biol 1997; 52: 261±272.

37. Springer ML, Blau HM. High ef®ciency retroviral infection ofprimary myoblasts. Som Cell Mol Genet 1997; 23: 203±209.

38. Lee RJ, Springer ML, Blanco-Bose WE, Shaw R, Ursell PC, BlauHM. VEGF gene delivery to myocardium: deleterious effects ofunregulated expression. Circulation, in press.

39. Christensen L, Dionne KE, Lysaght MJ. Biomedical applicationsof immobilized cells. In Fundamentals of Animal Cell Encapsula-tion and Immobilization, Goosen MFA (ed.). CRC Press: BocaRaton, FL, 1993; 7±29.

40. Lysaght MJ, Aebischer P. Encapsulated cells as therapy. Sci Am1999; 280: 76±82.

41. Regulier E, Schneider BL, Deglon N, Beuzard Y, Aebischer P.Continuous delivery of human and mouse erythropoietin in miceby genetically engineered polymer encapsulated myoblasts.Gene Ther 1998; 5: 1014±1022.

42. Shen H, Clauss M, Ryan J, Schmidt AM, Tijburg P, Borden L,Connolly D, Stern D, Kao J. Characterization of vascularpermeability factor/vascular endothelial growth factor receptorson mononuclear phagocytes. Blood 1993; 81: 2767±2773.

43. Moore JW, Sholley MM. Comparison of the neovascular effectsof stimulated macrophages and neutrophils in autologous rabbitcorneas. Am J Pathol 1985; 120: 87±98.

44. Dvorak HF, Brown LF, Detmar M, Dvorak AM. Vascularpermeability factor/vascular endothelial growth factor, micro-vascular hyperpermeability, and angiogenesis. J Pathol 1995;146: 1029±1039.

45. Claffey KP, Wilkison WO, Spiegelman BM. Vascular endothelialgrowth factor. Regulation by cell differentiation and activatedsecond messenger pathways. J Biol Chem 1992; 267:16317±16322.

46. Houck KA, Leung DW, Rowland AM, Winer J, Ferrara N. Dualregulation of vascular endothelial growth factor bioavailabilityby genetic and proteolytic mechanisms. J Biol Chem 1992; 267:26031±26037.

47. Shima DT, Kuroki M, Deutsch U, Ng YS, Adamis AP, D'AmorePA. The mouse gene for vascular endothelial growth factor.Genomic structure, de®nition of the transcriptional unit, andcharacterization of transcriptional and post-transcriptionalregulatory sequences. J Biol Chem 1996; 271: 3877±3883.

48. Carmeliet P, Ng YS, Nuyens D, Theilmeier G, Brusselmans K,Cornelissen I, Ehler E, Kakkar VV, Stalmans I, Mattot V, PerriardJC, Dewerchin M, Flameng W, Nagy A, Lupu F, Moons L, CollenD, D'Amore PA, Shima DT. Impaired myocardial angiogenesisand ischemic cardiomyopathy in mice lacking the vascularendothelial growth factor isoforms VEGF164 and VEGF188. NatMed 1999; 5: 495±502.



Induction of angiogenesis by implantation of encapsulated primary myoblasts expressing vascular...

Documents

Transcript of Induction of angiogenesis by implantation of encapsulated primary myoblasts expressing vascular...