Selective targeting of tumoral vasculature: Comparison of different formats of an antibody (L19) to the ED-B domain of fibronectin



We recently demonstrated that a human recombinant scFv, L19, reacting with the ED-B domain of fibronectin, a marker of angiogenesis, selectively targets tumoral vasculature in vivo. Using the variable regions of L19, we constructed and expressed a human “small immunoprotein” (SIP) and a complete human IgG1 and performed biodistribution studies in tumor-bearing mice to compare the blood clearance rate, in vivo stability and performance in tumor targeting of the 3 L19 formats [dimeric scFv (scFv)2, SIP and IgG1]. The accumulation of the different antibody formats in the tumors studied was a consequence of the clearance rate and in vivo stability of the molecules. Using the SIP, the %ID/g in tumors was 2–5 times higher than that of the (scFv)2, reaching a maximum 4–6 hr after injection. By contrast, the accumulation of IgG1 in tumors constantly rose during the experiments. However, due to its slow clearance, the tumor-blood ratio of the %ID/g after 144 hr was only about 3 compared to a ratio of 10 for the (scFv)2 and 70 for the SIP after the same period of time. The different in vivo behavior of these 3 completely human L19 formats could be exploited for different diagnostic and/or therapeutic purposes, depending on clinical needs and disease. Furthermore, the fact that ED-B is 100% homologous in human and mouse, which ensures that L19 reacts equally well with the human and the murine antigen, should expedite the transfer of these reagents to clinical trials. © 2002 Wiley-Liss, Inc.

Despite their enormous potential as therapeutic agents, monoclonal antibodies (mAbs) of nonhuman origin have not performed as well as expected in clinical trials as a result of their immunogenicity,1, 2 poor pharmacokinetic properties3, 4 and inefficiency in recruiting effector functions.5, 6 The recent prospect of isolating human antibody fragments from phage display libraries7, 8, 9, 10 transcends these problems, revitalizing studies and rekindling hopes of using these reagents to treat major diseases. Indeed, these molecules should serve as ideal building blocks for novel diagnostic and therapeutic tools.11, 12 Furthermore, these antibodies can be “matured” to reach affinities in the picomolar range,13 desirable, if not necessary, for their clinical use.14, 15

Clinical applications of human antibody fragments for the selective delivery of diagnostic or therapeutic agents nonetheless require highly specific targets. In the case of tumors, the most popular targets are cell-surface antigens, which are usually neither abundant nor stable. On the other hand, during tumor progression the microenvironment surrounding tumor cells undergoes extensive modification that generates a “tumoral environment” that could ultimately represent a suitable target for antibody-based tumor therapy.16 In fact, the concept that the altered tumor microenvironment is itself a carcinogen that can be targeted is increasingly gaining consensus. Molecules that are able to effectively deliver therapeutic agents to the tumor microenvironment thus represent promising and important new tools for cancer therapy.16, 17, 18

Fibronectin is an extracellular matrix (ECM) component that is widely expressed in a variety of normal tissues and body fluids. Different FN isoforms can be generated by the alternative splicing of the FN pre-mRNA, a process that is modulated by cytokines and extracellular pH.19, 20, 21, 22 The complete type III repeat ED-B may be entirely included or omitted in the FN molecule.23 ED-B is highly conserved in different species, having 100% homology in all mammalians thus far studied (human, rat, mouse) and 96% homology with a similar domain in chicken. The FN isoform containing ED-B (B-FN) is undetectable immunohistochemically in normal adult tissues, with the exception of tissues undergoing physiologic remodeling (e.g., endometrium and ovary) and during wound healing.20, 24 By contrast, its expression in tumors and fetal tissues is high.20 Furthermore, we demonstrated that B-FN is a marker of angiogenesis25 and that endothelial cells invading tumor tissues migrate along ECM fibers containing B-FN.26

We recently reported on the possibility to selectively target tumoral vasculature using a human recombinant antibody, L19(scFv),13 specific for the B-FN isoform.15, 27, 28, 29 This observation paved the way for the antibody's use in both in vivo diagnostic (immunoscintigraphy) and therapeutic approaches entailing the selective delivery of therapeutic radionuclides or toxic agents to tumoral vasculature. In addition, Birchler et al.30 showed that L19(scFv), chemically coupled to a photosensitizer, selectively accumulates in the newly formed blood vessels of the angiogenic rabbit cornea model and, after irradiation with near infrared light, mediates the complete and selective occlusion of ocular neovasculature. More recently, Nilsson et al.31 reported that the immunoconjugate of L19(scFv) with the extracellular domain of tissue factor mediates selective infarction in different types of murine tumor models. Furthermore, fusion proteins of L19(scFv) and IL-2 or IL-12 have shown the enhanced therapeutic efficacy of these 2 cytokines.32, 33 Finally, since L19 reacts equally well with mouse and human ED-B, it can be used for both preclinical and clinical studies.

Different antibody formats have shown diverse behavior in terms of in vivo stability, clearance and performance in tumor targeting.34 In order to select the formats suitable for different possible clinical applications, we prepared, characterized and investigated the in vivo biodistribution of 3 L19 human molecules, namely, the dimeric scFv (scFv)2, a small immunoprotein (SIP)35 and a complete IgG1.


Preparation and Expression of scFv, Small Immunoprotein (SIP) and IgG1 Constructs.


The L19(scFv) (Fig. 1) is an affinity matured (Kd = 5.4 × 10-11 M) antibody fragment specifically directed against the ED-B domain of fibronectin.13 The D1.3 (scFv),7, 28 a mouse-anti-hen egg white lysozyme scFv, was used as a control. These scFvs were expressed in E. Coli strain HB2151 (Maxim Biotech, San Francisco, CA) according to Pini et al.36

Figure 1.

(a) Model of the domain structure of an FN subunit. The protein sequences undergoing alternative splicing are indicated in gray. As indicated, the epitope of the recombinant antibody L19 is localized within the repeat ED-B. Schemes of the constructs used to express L19(scFv) (b); L19-SIP (c); and L19-IgG1/κ (d).


To construct the L19 small immunoprotein (L19-SIP) gene (Fig. 1), the DNA sequence coding for the L19(scFv) was amplified by polymerase chain reaction (PCR) using Pwo DNA Polymerase (Roche Diagnostics, Milan, Italy), according to the manufacturer's recommendations, with primers BC-618 (gtgtgcactcggaggtgcagctgttggagtctggg) and BC-619 (gcctccggatttgatttccaccttggtcccttggcc), containing the ApaLI and BspEI restriction sites, respectively. The amplification product was inserted ApaLI/BspEI in the pUT-ϵSIP vector, which provides the scFv gene with a secretion signal required for secretion of proteins in the extracellular medium. The pUT-ϵSIP vector was obtained from the previously described pUT-SIP-long35 after substituting the human constant γ1-CH3 domain with the CH4 domain of the human IgE secretory isoform IgE-S2 (ϵS2-CH4).37 CH4 is the domain that allows dimerization in the IgE molecule and the ϵS2 isoform contains a cysteine at the carboxyterminal end, which stabilizes the IgE dimer through an interchain disulphide bond. In the final SIP molecule, the L19(scFv) was connected to the ϵS2-CH4 domain by a short GGSG linker. The SIP gene was then excised from the plasmid pUT-ϵSIP-L19 with HindIII and EcoRI restriction enzymes and cloned into the mammalian expression vector pcDNA3 (Invitrogen, Groningen, The Netherlands), which contains the cytomegalovirus (CMV) promoter, in order to obtain the construct pcDNA3-L19-SIP.

The DNA sequence coding for D1.3(scFv) was amplified using the primers BC-721 (ctcgtgcactcgcaggtgcagctgcaggagtca) and BC-732 (ctctccggaccgtttgatctcgcgcttggt) and inserted ApaLI/BspEI in the pUT-ϵSIP vector. The D1.3-SIP gene was then excised from the pUT-ϵSIP-D1.3 with HindIII and EcoRI restriction enzymes and cloned into pcDNA3 in order to obtain the construct pcDNA3-D1.3-SIP.

These constructs were used to transfect SP2/0 murine myeloma cells (ATCC, American Type Culture Collection, Rockville, MD) using FuGENE 6 transfection reagent (Roche), following the protocol for adherent cells optimized by the manufacturer. Transfectomas were grown in DMEM supplemented with 10% FCS and selected using 750 μg/ml of Geneticin (G418, Calbiochem, San Diego, CA).


To prepare the complete IgG1, the variable region of the L19 heavy chain (L19-VH), together with its secretion peptide sequence, was excised with HindIII and XhoI from the previously described L19-pUTϵSIP and inserted in the pUC-IgG1 vector, containing the complete human γ1 constant heavy-chain gene. The recombinant IgG1 gene was then excised from the pUC-IgG1-L19-VH with HindIII and EcoRI and cloned into pcDNA3 to obtain the construct pcDNA3-L19-IgG1.

For the preparation of the complete L19 light chain, L19-VL was amplified from the L19-pUT-ϵSIP (described above) by PCR using the primers BC-696 (tggtgtgcactcggaaattgtgttgacgcagtc) and BC-697 (ctctcgtacgtttgatttccaccttggtcc), containing ApaLI and BsiWI restriction sites, respectively. After digestion with ApaLI and BsiWI, the amplification product was inserted in the vector pUT-SEC-hCκ containing the secretion signal sequence and the sequence of the human constant κ light chain. The recombinant light-chain gene was then excised from pUT-SEC-hCκ-L19-VL with HindIII and XhoI and inserted in the pCMV2Δ mammalian expression vector, derived from a pcDNA3 vector by removing the resistance gene to G418, to obtain the construct pCMV2Δ-L19-κ.

Equimolar amounts of these constructs were used to cotransfect SP2/0 murine myeloma cells as described above. Geneticin-selected clones were screened by ELISA for their ability to secrete chimeric immunoglobulin complete with heavy and light chains.

All DNA constructs were purified using the Maxiprep system from Qiagen (Hilden, Germany), and the DNA sequences of both strands of the constructs were confirmed using the ABI PRISM dRhodamine Terminator Cycle Sequencing Ready Reaction Kit (Perkin Elmer, Foster City, CA). All restriction enzymes (RE) were from Roche Diagnostics, with the exception of BsiWI (New England Biolabs, Beverly, MA). After RE digestion, inserts and vectors were recovered from agarose gels using the Qiaquick method (Qiagen).

Purification and Quality Control of Antibodies

Immunoaffinity chromatography was performed to purify the different antibodies according to the procedure described by Carnemolla et al.27

ED-B conjugated to Sepharose 4B (Amersham Pharmacia, Uppsala, Sweden) following the manufacturer's instructions27 was used to immunopurify all different L19 antibody formats, while a column of hen egg white lysozyme (Sigma, St. Louis, MO) conjugated to Sepharose 4B (Amersham Pharmacia) was used for D1.3 antibodies.

The immunopurified antibody formats L19-SIP and L19-IgG1 required no further purification and were dialyzed against PBS, pH 7.4, at +4°C. Since scFvs obtained from immunoaffinity chromatography are made up of 2 forms, monomeric and dimeric, a second purification step, as described by Demartis et al.,29 was required to isolate the latter dimeric form.

Batches of the different antibody formats were prepared and analyzed using SDS-PAGE under reducing and nonreducing conditions, immunohistochemistry, size exclusion chromatography (Superdex 200, Amersham Pharmacia) and ELISA experiments.

SDS-PAGE, ELISA, Size Exclusion Chromatography and Immunohistochemistry

Screening ELISA experiments on the conditioned culture media were performed according to Carnemolla et al.27 To reveal the expression of the different L19 antibody formats, the recombinant fragment 7B8927 containing the ED-B domain of FN, which includes the epitope recognized by L19, was immobilized on Maxisorp immunoplates (Nunc, Roskilde, Denmark). Hen egg white chicken lysozyme (Sigma) immobilized on NH2 surface EIA plates (Costar, Cambridge, MA) was used to detect D1.3 antibodies in ELISA experiments. A peroxidase-conjugated rabbit anti-human IgE (Pierce, Rockford, IL), diluted according to the manufacturer's recommendations, was used as secondary antibody to detect SIPs. A peroxidase-conjugated rabbit anti-human IgG (Pierce) was used in the case of IgG1. For the scFvs containing the tag sequence FLAG, we used a mouse anti-human FLAG monoclonal antibody (M2, Kodak) and a peroxidase-conjugated goat anti-mouse antibody (Pierce) as secondary and tertiary antibodies, respectively. In all cases the immunoreactivity with the immobilized antigen was detected using the substrate ABTS for peroxidase (Roche), and photometric absorbance at 405 nm was measured.

A Superdex 200 (Amersham Pharmacia) chromatography column was used to analyze the gel filtration profiles of the purified antibodies under native conditions using fast protein liquid chromatography (FPLC; Amersham Pharmacia). Immunohistochemistry on different tissue cryostat sections was performed as described by Castellani et al.25 and 4–18% gradient SDS-PAGE was carried out according to Carnemolla et al.20 under reducing and nonreducing conditions.

Animals and Cell Lines

Athymic nude mice (8-week-old nude/nude CD1 females) were obtained from Harlan Italy (Correzzana, Milano, Italy), 129 (clone SvHsd) strain mice (8–10 weeks old, female) were obtained from Harlan UK (Oxon, UK). Mouse embryonal teratocarcinoma cells (F9), human melanoma derived cells (SK-MEL-28) and mouse myeloma cells (SP2/0) were purchased from ATCC. To induce tumors, nude mice were subcutaneously injected with 16 × 106 SK-MEL-28 cells and 129 strain mice with 3 × 106 F9 cells. The tumor volume was determined with the following formula: (d)2 × D × 0.52, where d and D are the short and long dimensions (cm) of the tumor, respectively, measured with a caliper. Housing, treatments and sacrifice of animals were carried out according to national legislation (Italian law no. 116 of 27 January, 1992) on the protection of animals used for scientific purposes.

Radioiodination of Recombinant Antibodies

Radioiodination of proteins was achieved following the Chizzonite indirect method38 using IODO-GEN Pre-coated Iodination tubes (Pierce) to activate Na125I (NEN Life Science Products, Boston, MA) according to the manufacturer's recommendations. In the reported experiments, 1.0 mCi of Na125I was used for 0.5 mg of protein. The radiolabeled molecules were separated from free 125I using PD10 (Amersham Pharmacia) columns pretreated with 0.25% BSA and equilibrated in PBS. The radioactivity of the samples was established using a Crystal γ-counter (Packard Instruments, Milano, Italy). The immunoreactivity assay of the radiolabeled protein was performed on a 200 μl ED-B Sepharose column saturated with 0.25% BSA in PBS. A known amount of radioiodinated antibody, in 200 μl of 0.25% BSA in PBS, was applied on top and allowed to enter the column. The column was then rinsed with 1.5 ml of 0.25% BSA in PBS to remove nonspecifically bound antibodies. Finally, the immunoreactive bound material was eluted using 1.5 ml of 0.1M TEA, pH 11. The radioactivity of unbound and bound material was counted and the percentage of immunoreactive antibodies was calculated. Immunoreactivity was always higher than 90%.

To further analyze the radioiodinated antibodies, a known amount of radiolabeled protein in 200 μl was loaded onto the Superdex 200 column. The retention volume of the different proteins did not vary after radioiodination. For the 3 radioiodinated L19 antibody formats and their negative controls, the radioactivity recovery from the Superdex 200 column was 100% (Figs. 2 and 4a–c).

Figure 2.

Upper panel: SDS-PAGE analyses of the purified L19 antibodies. From left to right, L19(scFv)2, L19-SIP and L19-IgG1. r, reducing conditions; n-r, nonreducing conditions. The molecular masses of the standards are reported. Lower panel: Size exclusion chromatography profiles (Superdex 200) of the purified L19 formats, as indicated.

Figure 4.

(a–c) Size exclusion chromatography (Superdex 200) profiles of the indicated L19 formats after radioiodination. (d–f) Size exclusion chromatography (Superdex 200) profiles of plasma at the indicated times after i.v. injection of the radioiodinated L19 formats. No changes in the curve profiles of L19-SIP or L19-IgG1 were detected when loading plasma at different times after injection, while 3 hr after L19(scFv)2 injection a second peak of higher molecular mass was observed.

Biodistribution Experiments

To block nonspecific accumulation of 125Iodine in the stomach and concentration in thyroid, mice were given orally 20 mg of sodium perchlorate (Carlo Erba, Milano, Italy) in water 30 min before injection of the radiolabeled antibodies. This procedure was repeated at 24 hr intervals for the duration of biodistribution experiments. Tumor-bearing mice were injected in the tail vein with 0.1 nmoles of the different radiolabeled antibodies (corresponding to 6 μg for (scFvs)2, 8 μg for SIPs and 18 μg for IgGs) in 100 μl of 20 mM phosphate buffer, pH 7.3, 150 mM NaCl. Three animals were sacrificed per time point, the different organs including tumor were excised, weighed, counted in a γ-counter and then fixed with 5% formaldehyde in PBS, pH 7.4, to be processed for microautoradiographies, performed according to Tarli et al.26

The blood was also sampled for plasma preparation to determine the stability of the radiolabeled molecules in the bloodstream using the already described immunoreactivity test and the gel filtration analysis. In both cases, 200 μl of plasma were used. The radioactive content of the different organs was expressed as the percentage of injected dose per gram (%ID/g).

The blood clearance parameters of the radioiodinated antibodies were fitted with a least squares minimization procedure, using the MacIntosh program Kaleidagraph (Synergy Software, Reading, PA) and the equation: X (t) = A exp (−(alpha t)) + B exp (−(beta t)), where X (t) is the %ID/g of radiolabeled antibody at time t. This equation describes a biexponential blood clearance profile, in which the amplitude of the alpha phase is defined as A × 100 / (A + B) and the amplitude of the beta elimination phase is defined as B × 100 / (A + B). Alpha and beta are rate parameters related to the half-lives of the corresponding blood clearance phases. T1/2 (alpha phase) = ln2/alpha = 0.692…/alpha T1/2 (beta phase) = ln2/alpha = 0.692…/alpha. X(0) was assumed to be equal to 40%, corresponding to a blood volume of 2.5 ml in each mouse.


Antibody Preparation

Using the variable regions of L19,13 we prepared different antibody formats [(scFv)2, SIP and complete human IgG1] in order to analyze, in view of clinical applications, their performance in vivo in targeting tumoral vasculature. Figure 1 shows the constructs used to express the different L19 antibody formats.

To obtain SIPs and IgG1, SP2/0 murine myeloma cells were transfected with the constructs shown in Figure 1 and stable transfectomas were selected using G418. The best producers were selected by ELISA, and these clones were expanded for antibody purification. The purification of all 3 L19 antibody formats was based on immunoaffinity chromatography using recombinant ED-B conjugated to Sepharose. The yields were of about 8 mg/l for L19 (scFv)2, 10 mg/l for L19-SIP, 3 mg/l for L19-IgG1. For the control proteins, we used D1.3 (scFv)2 specific for hen-egg lysozyme,7, 28 and using the variable regions of scFv D1.3, we constructed D1.3-SIP. These 2 antibodies were purified on hen-egg lysozyme conjugated to Sepharose. The yields were of 8 and 5 mg/l, respectively. As control for L19-IgG1, we used commercially available human IgG1/κ (Sigma).

SDS-PAGE analysis of the 3 purified L19 formats under both reducing and nonreducing conditions is shown in the upper panel of Figure 2. For L19 (scFv)2, the apparent mass was, as expected, about 28 kDa, corresponding to the monomer, under both reducing and nonreducing conditions (not shown). The L19-SIP showed a molecular mass of nearly 80 kDa under nonreducing conditions and had a mass of about 40 kDa under reducing conditions. The results demonstrated that more than 95% of the native molecule exists as a covalently linked dimer. L19-IgG1 showed, as expected, a main band of about 180 kDa under nonreducing conditions, while, under reducing conditions, it showed 2 bands corresponding to the heavy chain of about 55 kDa and the light chain of about 28 kDa.

The lower panel of Figure 2 shows the elution profiles of the 3 L19 antibody formats analyzed by size exclusion chromatography (Superdex 200). In all 3 cases, a single peak with a normal distribution and representing more than 98% was detected. Using a standard calibration curve, the apparent molecular masses were 60 kDa for L19(scFv)2, 80 kDa for L19-SIP and 180 kDa for L19-IgG1. In addition, we demonstrated the absence of molecular aggregates that are often present in recombinant protein preparations and that may invalidate the results obtained in in vivo studies. SDS-PAGE and size exclusion chromatography (Superdex 200) performed on the purified control proteins gave similar results (not shown).

Using these 3 different L19 antibody formats, we performed immunohistochemic analyses on cryostat sections of SK-MEL-28 human melanoma induced in nude mice and of F9 murine teratocarcinoma induced in 129 strain mice. Optimal results were obtained at concentrations as low as 0.5–1.0 nM. All 3 purified L19 antibodies recognized identical structures (see Fig. 6a–c).

Figure 6.

(a–c) Immunohistochemic staining of SK-MEL-28 (a,c) and F9 (b) tumors using L19 (scFv)2(a), L19 SIP (b) and L19-IgG1 (c). (d) Microautoradiography of SK-MEL-28 tumor 3 hr after i.v. injection of the control 125I-D1.3 (scFv)2. (e,f) Microautoradiographies of normal liver (e) and normal kidney (f) 6 hr after i.v. injection of L19-SIP and L19-IgG1, respectively. (g–I) Microautoradiographies of F9 tumor 6 hr after i.v. injection of 125I-L19 (scFv)2(g), 125I-L19-SIP (h) and 125I-L19-IgG1 (i). Scale bars = 200 μm (a); 100 μm (g); 50 μm (b–f,h,i).

In Vivo Stability of the Radiolabeled L19 Antibody Formats

For in vivo biodistribution studies, SK-MEL-28 human melanoma and F9 murine teratocarcinoma were used. SK-MEL-28 tumor has a relatively slow growth rate, while F9 tumor grows rapidly (Fig. 3). Therefore, the use of SK-MEL-28 tumor enabled long-lasting experiments (up to 144 hr), while F9 tumor was induced for short biodistribution studies (up to 48 hr). All the biodistribution experiments were performed when the tumors were approximately 0.1–0.3 cm3. For comparison of the various antibody formats, equimolar amounts (0.1 nmol) in 100 μl of sterile 20 mM phosphate buffer, pH 7.3, 150 mM NaCl were injected. Before injection, the immunoreactivity and gel filtration profiles of the radioiodinated antibodies were checked (see Material and Methods). Immunoreactivity of the radiolabeled proteins was always more than 90%. Figure 4a–c reports the profiles of the gel filtration analysis (Superdex 200) of the radioiodinated L19 antibody formats.

Figure 3.

Growth curves of SK-MEL-28 tumor in nude mice (triangles) and of F9 tumor in 129 mouse strain (circles). The volume (mm3) is plotted vs. time (days). Each data point is the average of 6 mice ±SD.

Blood samples were taken from treated animals at the different time intervals from injection and the radioactivity present in plasma was analyzed for immunoreactivity and by gel filtration chromatography. Gel filtration profiles showed a single major peak, having the molecular mass of the injected protein, for all 3 L19 antibody formats. Only the profile of the (scFv)2 revealed a second peak having a higher molecular mass, suggesting formation of aggregates (Fig. 4d–f). Furthermore, the formation of large molecular mass aggregates not eluting from the Superdex 200 column was observed for L19(scFv)2. In fact, while the recovery from the Superdex 200 column was 90–100% of the applied radioactivity for both L19-SIP and L19-IgG, the yield of the loaded radioactivity of L19(scFv)2 was about 55%. The retained radioactivity was recovered only after washing the chromatography column with 0.5 M NaOH, demonstrating that large aggregates were blocked on the column filter (Table I). Table I also reports the results of the immunoreactivity test performed on plasma (see Material and Methods). Over the time of the experiments, L19-SIP and L19-IgG1 maintained the same immunoreactivity in plasma as the starting reagents. On the contrary, the immunoreactivity of L19(scFv)2 in plasma was reduced to less than 40% already 3 hr after injection.

Table I. Immunoreactivity (I)1 and Radioactivity Recovery (R) from Superdex 200 of the Radiolabeled Antibodies at Different Times after i.v. Injection
 3 hr6 hr24 hr48 hr72 hr144 hr
  • Immunoreactivity (%) and radioactivity recovery (%) from Superdex 200 were determined in plasma as described in Materials and Methods.

  • 1

    To normalize, the results of the immunoreactivity test are referred to the percentage values of the immunoreactivity before i.v. injection.–nd, not determined.


Comparative Biodistribution Experiments

Tables IIa–c and Figure 5 report the results obtained in the biodistribution experiments with the radiolabeled L19 antibodies in SK-MEL-28 tumor-bearing mice. Tables IIa–c show, at different times from i.v. injection of the radiolabeled antibodies, the average (±SD) of the %ID/g of tissues and organs, including tumors. Figure 5 depicts the variations of the %ID/g of the different antibody formats in tumor (a) and blood (b) at the different times of the experiments, as well as the ratios (c) between the %ID/g in tumor and blood. All 3 L19 antibody formats selectively accumulated in the tumor and the ratio of the %ID/g of tumor and other organs are reported in Table III.

Table IIa. Biodistribution Experiments of Radiolabeled L19 and D1.3 Antibody Fragments in SK-MEL-28 Tumor-Bearing Mice
 3 hr6 hr24 hr48 hr72 hr144 hr
  1. The results are expressed as percent of antibody injected dose per gram of tissue (%ID/g) ± SD.–nd, not determined.

L19 (scFv)2      
 Tumor2.47 ± 0.652.01 ± 0.721.62 ± 0.430.95 ± 0.140.68 ± 0.040.32 ± 0.14
 Blood1.45 ± 0.580.54 ± 0.120.10 ± 0.030.04 ± 0.010.03 ± 0.020.03 ± 0.01
 Liver0.48 ± 0.200.18 ± 0.050.04 ± 0.010.02 ± 0.000.02 ± 0.010.02 ± 0.00
 Spleen0.67 ± 0.280.27 ± 0.040.07 ± 0.020.03 ± 0.000.02 ± 0.010.02 ± 0.00
 Kidney4.36 ± 0.321.67 ± 0.080.16 ± 0.010.06 ± 0.010.04 ± 0.020.03 ± 0.00
 Intestine0.77 ± 0.210.57 ± 0.050.24 ± 0.060.17 ± 0.040.12 ± 0.050.09 ± 0.01
 Heart0.77 ± 0.200.31 ± 0.070.07 ± 0.020.02 ± 0.000.02 ± 0.010.02 ± 0.00
 Lung2.86 ± 0.341.50 ± 0.671.07 ± 0.420.73 ± 0.390.55 ± 0.110.51 ± 0.22
D1.3 (scFv)2      
 Tumor1.03 ± 0.740.87 ± 0.420.15 ± 0.100.07 ± 0.02ndnd
 Blood1.52 ± 0.860.81 ± 0.130.02 ± 0.000.01 ± 0.00ndnd
 Liver1.19 ± 0.650.66 ± 0.260.14 ± 0.040.03 ± 0.08ndnd
 Spleen1.05 ± 0.880.42 ± 0.330.07 ± 0.020.05 ± 0.01ndnd
 Kidney3.01 ± 2.481.83 ± 0.760.48 ± 0.010.18 ± 0.05ndnd
 Intestine0.56 ± 0.540.56 ± 0.130.17 ± 0.030.02 ± 0.01ndnd
 Heart0.86 ± 0.540.55 ± 0.840.02 ± 0.010.01 ± 0.00ndnd
 Lung1.28 ± 0.651.06 ± 0.880.04 ± 0.010.03 ± 0.01ndnd
Table IIb. Biodistribution Experiments of Radiolabeled L19-SIP and D1.3 SIP in SK-MEL-28 Tumor-Bearing Mice
 3 hr6 hr24 hr48 hr72 hr144 hr
  1. The results are expressed as percent of antibody injected dose per gram of tissue (%ID/g) ± SD.

L19 SIP      
 Tumor5.23 ± 0.656.14 ± 2.234.20 ± 2.472.57 ± 0.312.33 ± 0.901.49 ± 0.65
 Blood9.82 ± 0.685.03 ± 0.521.39 ± 0.060.29 ± 0.040.08 ± 0.020.02 ± 0.01
 Liver2.65 ± 0.141.74 ± 0.310.50 ± 0.040.19 ± 0.010.10 ± 0.020.05 ± 0.01
 Spleen3.76 ± 0.362.43 ± 0.240.71 ± 0.050.26 ± 0.040.13 ± 0.010.17 ± 0.18
 Kidney7.33 ± 0.913.87 ± 0.211.09 ± 0.050.30 ± 0.040.14 ± 0.020.05 ± 0.01
 Intestine1.45 ± 0.241.44 ± 0.291.06 ± 0.430.56 ± 0.080.40 ± 0.080.18 ± 0.00
 Heart4.16 ± 0.302.15 ± 0.080.52 ± 0.050.13 ± 0.030.06 ± 0.010.02 ± 0.01
 Lung7.72 ± 0.605.41 ± 0.551.81 ± 0.400.59 ± 0.290.19 ± 0.030.05 ± 0.01
D1.3 SIP      
 Tumor3.80 ± 0.301.65 ± 0.120.70 ± 0.000.26 ± 0.010.07 ± 0.010.04 ± 0.03
 Blood10.40 ± 0.814.45 ± 0.141.21 ± 0.010.32 ± 0.000.08 ± 0.010.06 ± 0.02
 Liver4.05 ± 0.982.73 ± 0.331.43 ± 0.070.51 ± 0.210.15 ± 0.080.02 ± 0.01
 Spleen3.31 ± 0.661.76 ± 0.500.82 ± 0.120.46 ± 0.200.15 ± 0.050.04 ± 0.02
 Kidney8.41 ± 0.494.64 ± 0.061.47 ± 0.050.36 ± 0.030.16 ± 0.030.06 ± 0.01
 Intestine2.03 ± 0.551.06 ± 0.201.02 ± 0.060.14 ± 0.030.08 ± 0.020.12 ± 0.04
 Heart3.28 ± 0.201.81 ± 0.020.29 ± 0.010.06 ± 0.000.05 ± 0.010.04 ± 0.01
 Lung6.16 ± 0.284.52 ± 0.071.16 ± 0.050.09 ± 0.000.06 ± 0.010.05 ± 0.01
Table IIc. Biodistribution Experiments of Radiolabeled L19 IgG1 and hIgG1k in SK-MEL-28 Tumor-Bearing Mice
 3 hr6 hr24 hr48 hr72 hr144 hr
  1. The results are expressed as percent of antibody injected dose per gram of tissue (%ID/g) ± SD.–nd, not determined.

L19 IgG1      
 Tumor4.46 ± 0.085.39 ± 1.016.70 ± 2.107.80 ± 2.518.90 ± 2.5211.22 ± 3.19
 Blood16.04 ± 0.8112.02 ± 1.658.31 ± 1.775.12 ± 1.425.02 ± 3.814.87 ± 0.26
 Liver6.03 ± 0.376.77 ± 0.532.41 ± 0.351.45 ± 0.411.26 ± 0.711.09 ± 0.16
 Spleen6.63 ± 1.346.37 ± 1.372.51 ± 0.472.01 ± 0.321.80 ± 1.021.51 ± 0.29
 Kidney6.47 ± 0.395.12 ± 0.473.07 ± 0.351.73 ± 0.631.54 ± 1.141.12 ± 0.44
 Intestine1.60 ± 0.391.35 ± 0.651.43 ± 0.191.13 ± 0.321.13 ± 0.980.97 ± 0.47
 Heart5.63 ± 0.674.77 ± 0.522.87 ± 0.451.48 ± 0.511.32 ± 1.090.92 ± 0.37
 Lung6.55 ± 0.655.15 ± 0.624.16 ± 0.662.28 ± 0.801.98 ± 1.601.42 ± 0.45
 Tumornd3.28 ± 0.384.00 ± 0.222.78 ± 0.20nd2.32 ± 0.26
 Bloodnd10.12 ± 0.357.87 ± 0.256.24 ± 0.34nd5.41 ± 0.51
 Livernd4.02 ± 0.092.06 ± 0.101.90 ± 0.24nd1.28 ± 0.03
 Spleennd4.47 ± 0.281.82 ± 0.011.42 ± 0.19nd1.24 ± 0.03
 Kidneynd5.40 ± 0.192.56 ± 0.062.08 ± 0.22nd1.30 ± 0.15
 Intestinend0.72 ± 0.070.46 ± 0.050.36 ± 0.03nd0.31 ± 0.01
 Heartnd3.80 ± 0.152.52 ± 0.210.99 ± 0.18nd1.48 ± 0.13
 Lungnd4.82 ± 0.923.64 ± 0.081.75 ± 0.32nd1.09 ± 0.13
Figure 5.

Results of biodistribution experiments in SK-MEL-28 tumor-bearing mice using the different radioiodinated L19 antibodies. The variations of the %ID/g in the tumor (a) and in blood (b) at the indicated times after i.v. injection are reported. In (c) the tumor-blood ratios of the %ID/g are plotted. The curves of L19 (scFv)2 are indicated by diamonds, of L19-SIP by squares and of L19-IgG1 by triangles.

Table III. Tumor-Organ Ratios of the % ID/g of the Radiolabeled L19 Antibody Formats in SK-MEL-28 Tumor-Bearing Mice
3 hr6 hr24 hr48 hr72 hr144 hr3 hr6 hr24 hr48 hr72 hr144 hr3 hr6 hr24 hr48 hr72 hr144 hr

As demonstrated by microautoradiography, the antibodies accumulated only in the tumor vasculature, whereas no specific accumulation in the vasculature of normal organs was seen (Fig. 6). By contrast, no specific accumulation of the radioiodinated control molecules in either tumors or normal tissues was found (Tables IIa–c and Fig. 6

All 3 L19 antibody formats showed a clearance that was mediated mainly by the kidney, as determined by counting the urine samples. As expected, clearance rate was faster for L19(scFv)2 and slower for the complete L19-IgG1. Fitting of the curve with a biexponential function yielded the half-life values reported in Table IV.

Table IV. Kinetic Parameters for Blood Clearance of the 3 L19 Antibody Formats
%1t1/2 (hr)%1t1/2 (hr)
  • 1

    Relative magnitude of the 2 half-life components.

L19 (scFv)296.70.533.38.00

Figure 7 depicts the variations in the %ID/g (±SD) of tumor and blood obtained with the radioiodinated L19(scFv)2 and L19-SIP using the F9 teratocarcinoma tumor model. Due to the high angiogenic activity of F9 teratocarcinoma, accumulation of radioactive molecules in this tumor was 3–4 times higher than in SK-MEL-28 tumor at 3 hr and 6 hr after i.v. injection and was persistently higher for the 48 hr duration of the experiment. As for SK-MEL-28 tumor, specific accumulation in tumoral vasculature was confirmed by microautoradiography (Fig. 6), while no specific tumor accumulation was seen after injection of the control molecules. Table V reports the %ID/g of L19(scFv)2and L19-SIP, at different times after i.v. injection, in F9 tumors and other organs.

Figure 7.

Results of biodistribution experiments in F9 tumor-bearing mice using radioiodinated L19 (scFv)2 (squares) and L19-SIP (diamonds). The variations of the %ID/g in the tumor (a) and in blood (b) at the indicated different times after i.v. injection are reported.

Table V. Biodistribution Experiments of Radiolabeled L19(scFv)2 and L19SIP in F9 Tumor-Bearing Mice
 3 hr6 hr24 hr48 hr
  1. The results are expressed as percent of antibody injected dose per gram of tissue (%ID/g) ± SD.

L19 (scFv)2    
 Tumor10.46 ± 1.758.15 ± 2.633.18 ± 0.832.83 ± 0.71
 Blood2.05 ± 0.381.88 ± 1.140.17 ± 0.010.06 ± 0.02
 Liver1.62 ± 1.670.73 ± 0.510.07 ± 0.010.04 ± 0.02
 Spleen1.53 ± 0.360.90 ± 0.540.11 ± 0.010.05 ± 0.01
 Kidney12.70 ± 0.734.37 ± 0.980.24 ± 0.030.18 ± 0.08
 Intestine0.68 ± 0.150.95 ± 0.230.24 ± 0.010.17 ± 0.06
 Heart1.35 ± 0.210.81 ± 0.380.08 ± 0.020.04 ± 0.01
 Lung2.88 ± 0.292.06 ± 0.690.38 ± 0.600.33 ± 0.05
 Tumor17.46 ± 1.9316.65 ± 2.5915.32 ± 2.1712.00 ± 1.91
 Blood13.51 ± 0.579.62 ± 1.181.73 ± 0.021.14 ± 0.20
 Liver2.81 ± 0.372.39 ± 0.130.54 ± 0.140.32 ± 0.00
 Spleen3.42 ± 0.262.66 ± 0.270.61 ± 0.090.37 ± 0.01
 Kidney9.18 ± 0.765.85 ± 0.501.16 ± 0.050.76 ± 0.06
 Intestine0.95 ± 0.031.36 ± 0.210.83 ± 0.111.04 ± 0.14
 Heart4.64 ± 0.243.67 ± 0.460.67 ± 0.060.46 ± 0.07
 Lung5.61 ± 0.015.93 ± 0.571.66 ± 0.190.91 ± 0.08


The observation that cytotoxic anticancer drugs localize more efficiently in normal tissues than in tumors39 prompted a wave of studies investigating the possibility of selective drug delivery to tumors. Among the different avenues that have been explored in order to achieve a selective tumor targeting, the ones featuring specific binding molecules (e.g., antibodies or peptides) that recognize tumor markers provide higher levels of selectivity, as measured by quantitative biodistribution analysis. The effective targeting of tumors, however, has 2 main requisites: (i) a target in the tumor that is specific, abundant, stable and readily available for ligand molecules coming from the bloodstream; (ii) a ligand molecule with suitable pharmacokinetic properties that is easily diffusible from the bloodstream to the tumor and that has a high affinity for the target to ensure its efficient and selective accumulation in the tumor.

Due to its distinctive features, we considered the tumor microenvironment as a possible pan-tumoral target. In fact, tumor progression induces (and subsequently needs) significant modifications in tumor microenvironment components, particularly those of the extracellular matrix (ECM). The molecules making up the ECM of solid tumors differ both quantitatively and qualitatively from those of the normal ECM. Moreover, many of these tumor ECM components are shared by all solid tumors, accounting for general properties and functions such as cell invasion (both normal cells into tumor tissues and cancer cells into normal tissues) and angiogenesis. Of the numerous molecules constituting the modified tumor ECM, we have focused our attention on an FN isoform containing the ED-B domain (B-FN). B-FN is widely expressed in the ECM of all solid tumors thus far tested and is constantly associated with angiogenic processes25 but is otherwise undetectable in normal adult tissues.20 These features potentially make B-FN an ideal tumor target, also because the targeted delivery of therapeutic agents to the subendothelial ECM overcomes problems associated with the interstitial hypertension of solid tumors.40, 41, 42

Today, some of the most promising ligands are the human antibody molecules that can be generated and customized using molecular engineering technologies. We have produced L19,13, 15, 26 an scFv with a high affinity (Kd = 5.4 × 10-11M) for the ED-B domain of FN, and demonstrated in vivo that it selectively and efficiently accumulates around tumor neovasculature and is able to selectively transport and concentrate in the tumor mass any one of a number of therapeutic molecules to which it is conjugated.30, 31, 32, 33 The ability of L19 to selectively target tumors has also been demonstrated in patients using scintigraphic techniques (data not shown).

In view of the potential diagnostic and therapeutic applications of L19, we report here on the tumor vascular targeting performance and pharmacokinetics of 3 different L19 human antibody formats: the (scFv)2, the small immunoprotein (SIP) and the complete human IgG1. The better performance in tumor targeting of dimeric vs. monomeric antibody formats was first demonstrated nearly 20 years ago by Buchegger et al.43 and later confirmed by numerous reports using molecules called “minibodies” or “flex minibodies” similar to the SIP reported here, as well as to dimeric scFvs.15, 34, 44, 45, 46, 47 The SIP molecule was obtained by fusion of the L19(scFv) to the ϵCH4 domain of the secretory isoform S2 of human IgE. The ϵCH4 is the domain that allows the dimerization of IgE molecules and the S2 isoform contains a cysteine at the COOH terminal that covalently stabilizes the dimer through an interchain disulphide bond, thus generating a stable bivalent molecule and in turn improving the performance of the antibody in vivo.15, 37, 48 The interaction of IgEs with their high-affinity cellular receptors (FcϵRI) may lead to deleterious allergic reactions upon crosslinking of the bound IgEs with their antigen. Since the IgE binding sites for FcϵRI reside in the CH3 domain,49, 50, 51 our scFv fused to ϵCH4 domain does not activate any signaling that could lead to hypersensitivity reactions.

We studied the performance of these 3 formats in 2 different tumor models in mice, murine F9 teratocarcinoma and human SK-MEL-28 melanoma. The first is a rapidly growing tumor that, once implanted, kills the animals in about 2 weeks. SK-MEL-28 tumor, on the other hand, presents a biphasic growth curve, with an early, fast, growth phase followed by a second, slower phase. We previously showed that the amount of ED-B in F9 teratocarcinoma remains stable during tumor growth;26 by contrast, ED-B accumulates in SK-MEL-28 melanoma proportionally to the ability of the tumor to grow,26 with abundant ED-B being found in the first phase and a lesser amount in the second. The use of SK-MEL-28 melanoma tumor allowed long-term biodistribution studies without dramatic variations of tumor mass (Fig. 3) that could give rise to misinterpretation of results.

Comparative studies of the 3 L19 antibody formats in terms of in vivo stability showed that L19-SIP and L19-IgG1 maintained, for the duration of experiments (144 hr), the same immunoreactivity and molecular mass in plasma as before injection. By contrast, L19 (scFv)2 rapidly lost its immunoreactivity in plasma and generated aggregates that were too large to enter the gel filtration chromatography column. Such aggregation of the scFv very likely accounts for the ratio between percent injected dose per gram (%ID/g) of tumor and lung, since aggregates could accumulate in the microvasculature of the lung (Table III). For all 3 formats, the blood clearance is mediated mainly via the kidney, showing a biphasic curve with an α and a β phase (Table IV), which is inversely proportional to molecular size, as expected.

The accumulation of the different antibody formats in the tumors studied was a consequence of the clearance rate and the in vivo stability of the molecules. Using the (scFv)2, the maximum %ID/g was observed 3 hr after injection of the radiolabeled antibody and then rapidly decreased. Using the SIP, the %ID/g in tumors was 2–5 times higher than that of the scFv, reaching a maximum 4–6 hr after injection. This pattern was observed in both F9 and SK-MEL-28 tumors. By contrast, the accumulation of IgG1 in tumors rose constantly during the experiments. However, due to its slow clearance, the tumor-blood ratio of the %ID/g after 144 hr was only about 3, compared to a ratio of 10 for the (scFv)2 and 70 for the SIP after the same period of time (Fig. 5).

The same distinctive properties of in vivo stability, clearance and tumor targeting performance shown by the 3 antibody formats studied here could be exploited for different diagnostic and/or therapeutic purposes, depending on the clinical needs and disease. For instance, radiolabeled antibodies showing good tumor-organ and tumor-blood ratios soon after injection are necessary for in vivo diagnostic immunoscintigraphy, mainly because short half-life isotopes are used in such procedures.

Two different approaches are possible using the antibody as a vehicle for therapeutic agents: delivery of substances that exert their therapeutic effects after reaching their targets (e.g., photosensitizers activated only on the targets), for which the absolute amount delivered to the tumor is relevant; delivery of substances that exert their therapeutic and toxic effects even before reaching the target (e.g., the β-emitter Yttrium-90), for which particular attention must be given to the ratio of the area under the curves of tumor and blood accumulation as a function of time, in order to minimize the systemic toxicity and to maximize the antitumor therapeutic effect.

L19-SIP, for instance, seems to offer the best compromise of molecular stability, clearance rate and tumor accumulation. Similar fusion proteins composed of scFv antibody fragments bound to a dimerizing domain have already been described,35, 47 but in both cases the human γ1CH3 was used as the dimerizing domain. Our use of the human ϵS2CH4 domain provides an easy means to obtain a covalent stabilization of the dimer. In addition, the disulphide bridge formed by the C-terminal cysteine residues can be easily reduced in mild enough conditions to preserve the overall structure of the molecule, thus providing a readily accessible reactive group for radiolabeling or chemical conjugation. This feature seems particularly promising in the view of a possible clinical utilization.

At the same time, the L19-IgG1 gathers abundantly in tumors, and even though this accumulation is offset by a slow blood clearance rate, the reported 3-step procedure to remove circulating antibodies could enable its use not only for therapeutic purposes but also for diagnostic immunoscintigraphy.52

In conclusion, we have generated and characterized 3 completely human L19 antibody formats that, thanks to the antigen they target, appear to provide a powerful means for the delivery of a variety of effector molecules to newly forming blood vessels. Moreover, the fact that ED-B is a naturally occurring angiogenesis marker identical in mouse and in humans should expedite the transfer of these promising reagents to clinical trials, thus providing a novel approach to the diagnosis and therapy of solid tumors.


We thank Dr. L. Vangelista (ICGEB, Trieste, Italy) for critical reading of the manuscript and Mr. T. Wiley for manuscript revision.