Wilms Tumor NCAM-Expressing Cancer Stem Cells as Potential Therapeutic Target for Polymeric Nanomedicine Free

Neural cell adhesion molecule, NCAM (CD56), an immunoglobulin superfamily cell adhesion molecule, is associated with an aggressive biological behavior, elevated expression of stem-cell markers, and increased metastatic capacity in several tumor types (1). In neuroblastoma, glioblastoma, melanoma, and some ovarian tumors, NCAM was found to be overexpressed in the entire cancer cell population (2, 3). Neuroblastoma tumors, for example, show high NCAM expression, which is frequently related to disease progression and inhibition of tumor cell adhesion to the endothelium (1). In other tumors, NCAM expression was found to be associated with poor prognosis (1, 3–8).

Wilms tumor (WT; nephroblastoma) is a type of cancer that initiates in the kidneys. It is the most common type of pediatric renal malignancy. The majority of WT are characterized by a tri-phasic histological pattern consisting of blastemal, stromal, and epithelial elements. A small subset of WT contains widespread anaplastic histology consisting of nuclear enlargement, and irregular mitotic cells, which are associated with increased relapse risk and are classified as unfavorable histology tumors. The absence of such anaplastic features categorizes these patients to having favorable histology WT, which represents the patient population analyzed in this study (9). The standard of care for children of any age has been laid out by two main strategies. Most children in Europe are treated with preoperative chemotherapy, according to the International Society of Pediatric Oncology- Renal Tumor Study Group (SIOP-RTSG). In North America, patients are treated with upfront surgery prior to administration of chemotherapy, as per the National Wilms Tumor Study/Children's Oncology Group (NWTS/COG) protocols (10). These approaches, which combine a wide variety of chemotherapy agents, have a similar overall survival of nearly 85%. Current treatment of WT principally centers around vincristine and actinomycin-D (to include doxorubicin for higher risk categories) for several months. In general, surgery is usually followed by multimodal adjuvant therapy that varies according to molecular biology and histology of the tumor. In cases of diffuse (widespread) anaplasia, WT patients are treated with more intense types of chemotherapy using the drugs etoposide, cyclophosphamide, and carboplatin, along with mesna in addition to the standard drugs vincristine and doxorubicin. However, none of these standard drugs is cell- or target-specific. As a result, many off-target effects may be encountered using these agents, which could be avoided with greater drug specificity. Therefore, there is a great medical need to develop new potent, selective, and safe drugs for WT patients that are not cured as well as patients who experience recurrence of disease and suffer from toxicity of current regimens. This motivation led to the rational design of our current study.

Our previous work has indicated that the human WT blastema, the undifferentiated component of the tumor, is locked in a renal progenitor state, overexpressing genes that characterize early renal development and nephron progenitors in the nephrogenic cortex of the developing kidney (11–14). This differentiation arrest is not complete so that mature lineages appear in the tumor at varying proportions, mimicking mesenchymal to epithelial transition (MET) that occurs in nephrogenesis (15, 16). NCAM marks the multipotent nephron stem cell population and its epithelial progeny that appear in fetal life during MET and nephrogenesis (15, 17, 18). Following completion of nephrogenesis at 34th week of human gestation, it is not expressed in the nephron neither postnatal nor during adulthood. Mouse models introducing gene mutations into renal lineage have confirmed human data linking a transformed early renal stem/progenitor cell to development of WT (19–21).

Cancer stem cells (CSC) are a population of cancer cells within the tumor that have similar characteristics to normal stem cells, specifically in their ability to self-renew and give rise to all cell lineages found in the tumor. Additional features of CSCs include increased ability to migrate and form metastases, and increased resistance to chemotherapy (22, 23). Consequently, even a small number of CSCs remaining after therapy can repopulate the tumor and cause recurrence of the disease (23). Our previous findings have indicated that WT CSC's self-renew, reside in the tumor blastema, share features with multipotent nephron progenitors, and are plastic, e.g., differentiating to epithelial lineage in transplantation assays but also being able to transit towards earlier mesenchymal lineage (12, 15, 17, 24). What allows the isolation of these cells from the WT blastema is the expression of NCAM and ALDH1 (12, 17, 24). Hence, specific targeting of NCAM in the tumor has the potential to eliminate a cell population containing the CSCs responsible for propagation and recurrence of the tumor (14). Accordingly, it was shown that an NCAM-targeting antibody conjugate (IMGN901, lorvotuzumab mertansine) led to tumor eradication in a human WT xenograft model (17). Studies with IMGN901 conducted by the Pediatric Preclinical Testing Program on WT xenografts confirmed these results (25). These studies have sparked clinical trials with lorvotuzumab mertansine for relapsing WT (https://clinicaltrials.gov/ct2/show/NCT02420873). However, there exists a need for newer and safer NCAM targeting compounds, especially as CSCs have been characterized by the expression of NCAM in other cancers as well, including hepatocellular carcinoma, hepatoblastoma and pleuro-pulmonary blastoma, indicating it may serve as a broader therapeutic target (26, 27).

Here, we hypothesized that an NCAM-targeted polymer drug conjugate can be used to target both tumors with high NCAM expression as well as NCAM-CSC populations within tumors and the tumor blood vessels. An attractive polymeric carrier, which uses NCAM as a selective target for drug delivery to the tumor, should be water-soluble, nontoxic and multivalent such as polyglutamic acid (PGA; refs. 28–30). The polymeric backbone is composed of units of naturally occurring L-glutamic acid linked together through amide bonds. The pendant free γ-carboxyl groups in each repeating unit are negatively charged at a neutral pH, which renders the polymer water-soluble, they also provide functionality for drug attachment. PGA is enzymatically biodegradable by cathepsin B, highly expressed in most tumor tissues (31–34). Moreover, it is nonimmunogenic and enables multivalent binding of drugs as well as targeting moieties. Hence, when used at an appropriate nanoscaled size, PGA conjugate can allow selective extravasation-dependent delivery to tumors via leaky angiogenic tumor vessels (29, 35, 36).

Several study cases report dramatic and favorable responses of WT patients to the tubulin inhibitor, paclitaxel (PTX), in addition to the more commonly used vincristine (37–39). Therefore, we selected to use PTX in this study for which the chemical reaction required for binding it to a polyglutamic acid nanocarrier is relatively simple and elegant as we reported previously (35, 36). To that end, this approach is attractive from the translational clinical point of view. We have previously developed an NCAM-targeted-peptide (NTP) conjugate of PGA with PTX named here PGA–PTX–NTP conjugate. This conjugate effectively inhibited the growth of neuroblastoma, a high NCAM-expressing tumor in mice, and did not have the toxic side effects associated with free PTX such as weight loss, decreased WBC count and neurotoxicity (36). Our goal in this work was to determine the ability of PGA–PTX–NTP to inhibit WT by targeting the NCAM-expressing CSCs.

All animal procedures were performed in compliance with Tel Aviv University and Sheba Medical Center guidelines approved by the Institutional Animal Care and Use Committees (IACUC).

Two known peptides (40) were synthesized using solid phase peptide synthesis (SPPS) method on Sieber amide resin: (i) an NCAM targeting peptide with the sequence ASKKPKRNIKA (NTP) and (ii) a control peptide with the sequence ASKKPAANIKA (cNTP). Glycine was added at the N-terminal as a linker to allow conjugation to the PGA as previously described (36).

The conjugate was synthesized as described in refs. (35, 36). Briefly, the NCAM-targeting peptide and PTX were conjugated to PGA by carbodiimide coupling. The conjugate was treated with NaHCO₃ to transform it into the sodium salt, water-soluble form, and purified on a size exclusion chromatography column.

To achieve single-cell suspensions, tumor samples were first dissolved by mincing them in Iscove's Modified Dulbecco's Medium (IMDM) supplemented with 1% penicillin–streptomycin (pen-strep) 100 mol/L (Biological Industries), and soaked in collagenase type IV (Sigma-Aldrich) for 1 hour at 37°C. Next, the digested tissues were gradually forced through 100 μm and 70 μm cell strainers to achieve single-cell suspension. After removal of the digestion medium, cells were washed in red blood cells (RBC) lysis solution (comprised of 8.3 g NH₄Cl, 1.0 g KHCO₃, 1.8 mL of 5% EDTA in double distilled H₂O) at 1 mL/5 mL cell suspension ratio for 2 minutes at 4°C. Cells were then resuspended in IMDM growth medium supplemented with 10% fetal bovine serum (Invitrogen), 1% pen-strep 100 mol/L, 1% L-glutamine (both from Biological Industries), 50 ng/mL epidermal growth factor (EGF), 50 ng/mL basic fibroblast growth factor (bFGF) and 5 ng/mL stem cell factor (SCF; all growth factors were purchased from PeproTech Asia).

For this study, two patient derived xenografts (PDX) lines were established out of two separate WT patients, from Sheba Medical Center, Tel Hashomer, and propagated in mice for five generations. These xenograft lines were generated from favorable histology WT. They were expanded in vivo to achieve more blastema predominance (expansion of blastema at the expense of differentiated elements). Importantly, WT with blastemal predominance (after preoperative chemotherapy) is regarded as high-risk tumors and require more aggressive treatment (41). These xenograft lines were termed W011 and W038 and were used for the in vitro and in vivo experiments at the fifth course of propagation. W011 cells were isolated in 2008 from a recurrent Wilms tumor of a 7-year-old female originally bearing a Triphasic-Blastemal predominance favorable histology tumor that showed features of local anaplasia. Its loss of heterozygosity (LOH) status 1p/16q was not known. The recurrence indicates a more aggressive treatment-resistant form of disease. W038 cells were isolated in 2013 from a primary tumor resected from a 5-year-old male bearing a Triphasic-Blastemal predominance with favorable histology tumor, treated using COG protocol and with a negative LOH status. When used in vitro, the patient-derived cells were plated after suspension and used immediately for the relevant assays; therefore, it was irrelevant to test for mycoplasma contamination.

Cells were plated onto 24-well plates (3 × 10⁴ WT cells/well) and allowed to attach for 24 hours. Cells were incubated with the conjugates and free drugs, dissolved in cell culture medium at serial concentrations, for 72 hours. PTX treatment concentrations ranged from 0.01 to 10,000 nmol/L, according to PTX-equivalent concentration. PGA was used at concentrations equivalent to the PGA content in PGA-PTX or PGA-PTX-NTP conjugate. Following incubation, cells were washed, detached by trypsin, and counted by Coulter Counter (Beckman Coulter). Each experiment included at least 3 replicates and was repeated 3 times.

The migration of cancer cells in the presence of conjugates and controls was evaluated using the scratch assay. A scratch was done on a confluent cell monolayer using a tip, and cells were incubated with the conjugates and the free drugs at PTX-equivalent concentration of 100 nmol/L for 24 hours. Plates were imaged and the gap width was measured at the beginning and end of the experiment, at the same reference points. Each experiment included at least 3 replicates and was repeated 3 times.

Tumors were established in NOD/SCID mice from a single primary tumor that was propagated in mice for five passages (W038 xn5). The tumor was enzymatically dissociated into single-cell suspension and 1 × 10⁶ cells in 100 μL 1:1 serum free medium/Matrigel (BD Biosciences) were injected subcutaneously into the left lower flank of NOD/SCID mice. Mice bearing palpable xenografts were randomized prior to treatment initiation. Five weeks after tumor inoculation, mice were treated intravenously (i.v.) with conjugates (PGA-PTX-NTP (n = 5), PGA-PTX-cNTP (n = 4), PGA-PTX (n = 6), free PTX (n = 5), and saline (n = 4)). Mice were treated every other day, five treatments in total. All treatments were administered at 15 mg/kg PTX equivalent dose. Tumors were measured by a digital caliper and tumor volume was calculated as follows: width² × length × 0.52. Body weight and tumor size were monitored every other day. Two days after the last treatment, mice were euthanized and tumors were measured. NCAM expression in tumors treated with PGA-PTX-NTP was measured before, after three treatments, and after four treatments (tumors were taken for analysis at each time point).

Freshly removed patient-derived tumors (Xns) were dissociated into a single-cell suspension as described above and immediately analyzed. Cells were suspended in FACS buffer consisting of 0.5% BSA (Sigma-Aldrich) and 2 mmol/L EDTA (Biological Industries) in Dulbecco's Phosphate Buffered Saline (PBS; Biological Industries), and blocked with a mixture of FcR blocking reagent (MiltenyiBiotec) and human serum (1:1). For NCAM (eBiosience, 1205942) labeling, cells were incubated with the antibody at a concentration of 5 μg antibody per 10⁶ cells for 30 minutes, in the dark, at 4°C. Cell viability was tested using 7-amino-actinomycin-D (7AAD) viability staining solution (eBioscience). Cell labeling was detected using FACS Calibur (BD Pharmingen). FACS results were analyzed using FlowJo analysis software. Viable cells were defined by their Forward Scatter/Side Scatter (FSC/SSC) profiles and lack of 7AAD staining. All quantitative measurements were made in comparison with the antibody IgG isotype (eBioscience).

H&E, caspase 3, Ki-67, and Pax8 histological staining were performed on formalin-fixed paraffin-embedded tumors. Staining was done with Caspase-3 (CP 229 A) mouse monoclonal IgG primary antibody (Biocare Medical) for apoptosis, Rabbit anti-Ki-67 (RBK027-05, Zytomed) for proliferation, Pax8 was stained using mouse monoclonal antibody (Cell Marque), anti-human CD56 primary antibody for NCAM (CD56; BioLegend), and mouse anti-ALDH for Aldehyde dehydrogenase 1 (ALDH1) staining (BD Biosciences) as was previously described (24, 36). All were followed by Histostain Broad spectrum kit (Invitrogen) and Liquid DAB substrate kit (Invitrogen).

In vitro assays data were expressed as mean ± SD and in vivo experiments as ±SEM. Statistical significance was determined using an analysis of variance (ANOVA).

A PGA-based conjugate bearing PTX and NTP (PGA–PTX–NTP) and control conjugates (PGA–PTX–cNTP and PGA–PTX) were synthesized and characterized. Structure of the conjugate is depicted in Fig. 1. Physico-chemical characterization results were described previously (36).

The conjugate's activity was evaluated in vitro on proliferation of WT cells isolated from two patient-derived tumors, W038 xn5 and W011 xn4. Prior to evaluation of the antiproliferative effect of the conjugates, we determined the presence of an NCAM-expressing population fraction. Both patient-derived cells expressed high levels of NCAM (Fig. 2).

Cells were incubated with the conjugates (PGA–PTX–NTP, PGA–PTX–cNTP, PGA–PTX), free PTX and PGA for 72 hours. PGA–PTX–NTP had a high cytotoxic activity on both patient-derived WT cell cultures, showing that the drugs are released from the polymer and retain their activity. PGA–PTX–NTP inhibited the proliferation of WT cells at a slightly lower, yet significant, IC₅₀ value compared with control PGA–PTX–cNTP conjugate and nontargeted PGA–PTX. The similar IC₅₀ values at the same order of magnitude suggest that the PTX release from all conjugates retained its antiproliferative (tubulin inhibitory) activity. PGA had no cytotoxic effect in all the concentrations examined (Fig. 3A and B). IC₅₀ values of the compounds are summarized in Table 1.

The migration of patient-derived WT cells in the presence of the PGA–PTX–NTP conjugate was evaluated using the scratch assay. A scratch was done on a 90% confluent cell monolayer, and cells were incubated with the conjugates and the free drug at PTX-equivalent concentration of 100 nmol/L for 24 hours. Plates were imaged, and gap width was measured at the beginning and end of the experiment. PGA–PTX–NTP significantly inhibited the migration of the cells (Fig. 3C).

In W038 cells, the fraction of NCAM-expressing cells in the tumor decreased dramatically following treatments with PGA–PTX–NTP: from over 90% before treatment initiation, to 7% after only 3 treatments as measured by flow cytometry (Fig. 4A). Moreover, FACS analysis of cells that were suspended from all treated-groups' tumors at the endpoint of the study revealed that PGA–PTX–NTP treatment markedly decreased NCAM expression compared with the other treatments. NCAM expression at the end point was: Saline, 51%; PGA-PTX, 40%; PGA–PTX–cNTP, 62.5%; PGA–PTX–NTP, 8.9% (Fig. 4B). It can be seen that there was a marked reduction in NCAM expression, from ∼90% to ∼50%, even in the saline-treated group due to tumor progression and differentiation to a more epithelial phenotype, which expresses less NCAM. The acceptable differences between the NCAM levels amongst the mice treated with PGA–PTX–NTP (Fig. 4A and B, showing 0.29% and 8.97%, respectively) result from an anticipated biological variability between tumors' response to treatments and yet shows a marked decrease from control groups. NCAM-targeted PGA–PTX–NTP conjugate inhibited W038 tumor growth more than the control PGA–PTX–cNTP conjugate and the nontargeted PGA–PTX conjugate and was able to reduce the tumor size. Free PTX exhibited higher reduction of tumor size; however, mice in the PTX-treated group suffered from toxic side effects and significant weight loss following treatments (Fig. 5). This implies that the elimination of NCAM-expressing cells from the tumors can account for not only growth inhibition but also tumor volume reduction observed in three of the five PGA–PTX–NTP-treated tumors (Fig. 5C). Obviously, PTX caused reduction in all surviving treated mice; however, this treatment was generally toxic and it can be seen that at the time of the acquired data, only 3 mice were left in the PTX-treated group which also suffered from severe weight loss (therefore euthanized).

H&E staining of tumors shows similar histology characteristics in the different treated groups. Xenografts are composed of mainly blastemal cells, while PGA–PTX–NTP tumors are devoid of differentiated elements. Ki-67 staining shows complete absence of proliferation in PGA–PTX–NTP-treated tumors, while in all other treatment groups there is substantial proliferation. Caspase-3 staining shows extensive apoptosis in tumors treated with free PTX, PGA–PTX–NTP and PGA–PTX. In tumors treated with PGA–PTX–NTP, apoptosis is seen in the core of the tumors, while in PGA–PTX–treated tumors, apoptosis is only seen at the margins. NCAM (CD56) staining was positive in mice treated with saline, while treatment with PGA–PTX and PGA–PTX–cNTP resulted in few cells stained for NCAM and treatment with PGA–PTX–NTP displayed no positive NCAM staining. This is in agreement with our conclusion from previous experiments that the targeted conjugate depletes NCAM-expressing cells. Aldehyde dehydrogenase 1 (ALDH1) staining shows positive staining in all treatments except of PGA–PTX–NTP, suggesting that the conjugate eliminated the CSC population defined in this tumor as NCAM⁺ cells. Pax8, a nephric cell-lineage transcription factor, staining was positive in tumors from all the control groups, while in the PGA–PTX–NTP- treated group, very few cells were positively stained, suggesting that almost all NCAM⁺ cells were targeted and eliminated by the conjugate, leaving no epithelial cells in the lesion (Fig. 6).

Several antibody-based systems for NCAM targeting were previously developed (42), among them humanized anti-NCAM antibodies (43–45), radioimmunoconjugates, and immunotoxins (46–51). To date, the most advanced NCAM-targeting therapeutic is huN901-DM1 (IMGN901), a humanized antibody conjugated to the cytotoxic drug maytansine (ImmunoGenInc). However, failure to achieve a sufficient improvement over current therapy (etoposide/carboplatin) using IMGN901 in a phase II clinical trial for small-cell lung carcinoma treatment resulted in a discontinuation of the trial in 2013.

We have successfully synthesized and characterized a novel conjugate that consists of an NCAM-targeting peptide (NTP) and paclitaxel (PTX) bound to the biodegradable PGA polymer (36). NTP is bound to PGA via a noncleavable amide bond, while PTX is bound via an ester bond that can be hydrolyzed by esterases in acidic pH. The peptide loading was 1.5 mol% in both PGA–PTX–NTP and PGA–PTX–cNTP, and drug loading was 7.5 mol% and 11 mol%, respectively, allowing a reliable comparison between the two conjugates. The conjugates' (with peptides) hydrodynamic radius was ∼10 nm, whereas of PGA-PTX was ∼5 nm. We predicted that 10 nm size would be sufficient to achieve selective accumulation via the enhanced permeability and retention (EPR) effect, while 5 nm size is at the lower range.

PGA–PTX–NTP had a strong cytotoxic activity on all patient-derived WT cell cultures, showing that the drug is released from the polymer and retains its activity. PGA–PTX–NTP inhibited the proliferation of WT cells at a higher extent than the control PGA–PTX–cNTP conjugate and the nontargeted PGA–PTX, suggesting a faster and more efficient internalization of the NCAM-targeted conjugate. The free drug was more active than the conjugated one, because it rapidly crosses the cell membrane by diffusion. As expected, the advantage of the conjugate over the free drug was seen in vivo in terms of activity and safety.

NCAM-targeted conjugate exhibited enhanced antitumor activity compared with control PGA–PTX–cNTP conjugate on mice bearing a human WT patient-derived xenograft (PDX). We speculate that due to the preferential binding to NCAM-expressing cells, the targeted conjugate was able to accumulate and internalize more efficiently into tumor cells and their supporting tumor endothelial cells. This speculation is based on the current data and on previous data showing that the targeted conjugate, PGA–PTX–NTP, internalized preferentially and more rapidly to NCAM-expressing cancer cells as shown by FACS, and to endothelial cells as shown by inhibition of capillary-like tube formation compared with the nontargeted conjugate PGA–PTX–cNTP (36). PGA–PTX did not exhibit any activity, perhaps due to its small hydrodynamic size that did not allow for effective accumulation in the tumor.

Free PTX displayed increased activity; however, it was much more toxic, as evidenced by the high weight loss, as well as toxic deaths and WBC count decrease that was previously published (36). Therefore, while the free drug was already given at the maximum tolerated dose (MTD), we hypothesized that the dose of the conjugate can be increased further to improve its activity. Mice that were treated with saline and conjugates suffered some weight loss as well; however, this weight loss can be attributed to the tumor's effect, as in this experiment, the treatments were initiated when the tumors were already very large (4 weeks after tumor inoculation) as clinically, Wilms tumors often become quite large before they are noticed. Therefore, this study design represents more realistically the clinical settings for a “regression trial” as opposed to a design representing a “prevention trial.” In an additional experiment on a different cancer type (neuroblastoma), treatments were initiated at an earlier time point when tumors were smaller. In that experiment, PGA–PTX–NTP was compared with PGA–PTX–cNTP and PGA–PTX at 10 mg/kg PTX-equivalent dose, for 5 treatments q.o.d. (total dose of 50 mg/kg), and, additionally, given at 30 mg/kg (total dose of 150 mg/kg; ref. 36). The results confirmed that the NCAM-targeted conjugate is more effective than control PGA–PTX–cNTP and nontargeted PGA–PTX conjugates. Moreover, these results clearly demonstrate that the conjugate's dose can be significantly increased to obtain enhanced activity without aggravating toxicity, while free PTX cannot be given at a dose over 15 mg/kg at this dosing schedule. The complete absence of toxic effects for the PGA–PTX–NTP, i.e., weight loss, decrease in WBC count, and toxic death, suggests that the dose can be increased even further.

NCAM can also be exploited to target the tumor vasculature. We have previously shown that NCAM is expressed on human umbilical vein endothelial cells (HUVEC) during the process of capillary-like tube formation, but not in HUVEC cultured in regular conditions (36). These results are in line with the work of Bussolati and colleagues, which showed that normal endothelial cells transiently acquired NCAM when organized in vessel-like structures. They also demonstrated that NCAM was constitutively expressed by tumor-derived endothelial cells but not by normal endothelial cells (52). These results suggest that an NCAM-targeting conjugate can have a dual effect by inhibiting both the proliferation of tumor cells and formation of tumor vasculature. Indeed, we have found that the PGA–PTX–NTP conjugate effectively inhibited the organization of HUVEC into capillary-like tubes (36). These findings are relevant to WT as demonstrated here.

Targeting CSCs in WT was shown to have a great potential by using humanized anti-NCAM antibody–drug conjugate huN901-DM1 (lorvotuzumab mertansine; ref. 17). In these experiments, huN901-DM1 inhibited the in vitro stemness properties of WT cell cultures that varied in the extent of NCAM expression. Furthermore, its in vivo targeting of multiple WT xenograft models was highly effective: WT bearing high NCAM expression, where NCAM population was enriched by passaging the tumors repeatedly in vivo, were completely or significantly eradicated. These results demonstrated the potential of targeting NCAM in a population of cells fueling the tumor and suggested that it can be a key target in appropriate developmental cancers bearing a primitive undifferentiated NCAM⁺ “blastema” phenotype. Indeed, we hypothesized that eradication by targeting the NCAM⁺ blastema, especially if >98% positive, would change the histologic appearance of the xenografts. However, the PGA–PTX–NTP treated tumors still appear to have predominantly blastema (small round blue cells) and with minimal stroma and no obvious or very few epithelia (Fig. 6) as seen by PAX8 staining. The explanation for this phenomenon might be that eradication of NCAM⁺ blastema WT CSCs allows selection of a more committed blastemal cell that is NCAM⁻ and lacks differentiation capacity to give rise to epithelial and stromal structure. In fact, it seems like the distribution of NCAM⁺ cells is such that they reside principally in the center of xenografts as can be seen in the saline-treated group (Fig. 6), and that after treatment with PGA–PTX–NTP, all that remains are the more peripheral NCAM⁻ cells that compress and comprise the surviving lesion. Moreover, PGA–PTX treatment resulted in apoptosis displayed only at peripheral margins, although we expected it to have a global effect. It seems that PGA–PTX follows recruited murine blood vessels, while the PGA–PTX–NTP effect occurred mainly in the center of xenografts supporting the cell specificity of the targeted strategy. These findings suggest that future studies should include treatments for a longer time period and an alternate dosing schedule.

Moreover, xenografts used in the in vivo experiments were blastema predominant and contained a minimal number of premature tubules (as seen in Fig. 6). Following the PGA–PTX–NTP treatment, we observed a change in the histological appearance of the xenografts that includes the disappearance of differentiated elements (from which NCAM expression is absent). Thus, targeting the NCAM⁺ population may not eradicate all blastema, but elimination of the CSCs is likely to result in the disappearance of differentiated epithelial elements that are derivatives of the WT CSCs (15, 17, 18, 24). Previous results have indicated that differentiated elements within the blastema arise from the NCAM⁺ WT-CSC population. This population fuels the propagation of WT (15, 17, 18, 24); therefore, targeting NCAM⁺ WT-CSC depletes the tumor of these structures as can be seen in Fig. 6 (right, treatment with PGA–PTX–NTP). This suggests that anti-NCAM therapy should be implemented alongside other modalities so as to afford complete eradication of all tumor compartments. In fact, lorvotuzomab mertansine is currently under trial for relapsing WT (https://clinicaltrials.gov/ct2/show/NCT02452554). Importantly, NCAM is variably expressed as different isoforms and with extensive posttranslational modifications, and these factors may influence antibody–drug conjugate binding and/or internalization. Thus, even if a valid therapeutic target such as NCAM exists, the target molecule may not be ideal from a clinical perspective. Accordingly, additional compounds should be developed and introduced to overcome the dependency on antibody specificity. PGA–PTX–NTP conjugate represents an attractive alternative as its binding to NCAM is dependent on a peptide and not an antibody, and is therefore less sensitive to posttranslational modifications.

In conclusion, our conjugate was highly effective in inhibiting proliferation and migration of xenograft-derived WT cells. Most importantly, the NCAM-targeted conjugate caused a reduction in tumor size in a patient-derived WT model and dramatically reduced the NCAM-expressing cell fraction, which was previously established as containing the CSC fraction of the tumor (12, 17). These results demonstrate that the PGA–PTX–NTP conjugate can be an effective tool to target and eradicate CSCs that are characterized by NCAM. Moreover, the activity seen on W011 cells is promising as this WT sample was originated from a recurrent disease, highlighting the PGA–PTX–NTP ability to target a treatment-resistant population of cells underscoring its therapeutic effect.

No potential conflicts of interest were disclosed.

Conception and design: E. Markovsky, E. Vax, O. Harari-Steinberg, N. Pode-Shakked, B. Dekel, R. Satchi-Fainaro

Development of methodology: E. Markovsky, E. Vax, I. Barshack, N. Pode-Shakked, R. Satchi-Fainaro

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Markovsky, E. Vax, A. Eldar-Boock, E. Yeini, I. Barshack, R. Caspi

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Markovsky, E. Vax, D. Ben-Shushan, A. Eldar-Boock, R. Shukrun, I. Barshack, O. Harari-Steinberg, R. Satchi-Fainaro

Writing, review, and/or revision of the manuscript: E. Markovsky, E. Vax, D. Ben-Shushan, A. Eldar-Boock, E. Yeini, I. Barshack, B. Dekel, R. Satchi-Fainaro

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Vax, D. Ben-Shushan, E. Yeini, R. Caspi, R. Satchi-Fainaro

Study supervision: B. Dekel, R. Satchi-Fainaro

We thank Dr. Camila Avivi and Liat Anafi for their professional assistance with the histology staining and Dr. Hana Golan for her contribution to this study. E.M. thanks Tel Aviv University Center for Nanoscience and Nanotechnology for an excellence doctoral fellowship.

R. Satchi-Fainaro is supported by The Israel Science Foundation (grant no. 918/14), The European Research Council (ERC) Consolidator Award (617445, PolyDorm). B. Dekel is supported by The Ziering Foundation, The Israel Cancer Association (grant no. 20150916), The Israel Cancer Research Fund (ICRF) project grant (grant no. PG14112). This study was partially supported by the Tel-Aviv University, Sheba Medical Center Collaboration grant, the Israel Science Foundation (Grant No. 918/14), The European Research Council (ERC) Consolidator Award (617445, PolyDorm), and by grants from the Israeli National Nanotechnology Initiative (INNI), The Leona M. and Harry B. Helmsley Nanotechnology Research Fund and by Focal Technology Area (FTA) program: Nanomedicine for Personalized Theranostics.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.