A novel, safe, fast and efficient treatment for Her2-positive and negative bladder cancer utilizing an EGF-anthrax toxin chimera

Abstract

Bladder cancer is the sixth most common cancer in the United States, and it exhibits an alarming 70% recurrence rate. Thus, the development of more efficient antibladder cancer approaches is a high priority. Accordingly, this work provides the basis for a transformative anticancer strategy that takes advantage of the unique characteristics of the bladder. Unlike mucin-shielded normal bladder cells, cancer cells are exposed to the bladder lumen and overexpress EGFR. Therefore, we used an EGF-conjugated anthrax toxin that after targeting EGFR was internalized and triggered apoptosis in exposed bladder cancer cells. This unique agent presented advantages over other EGF-based technologies and other toxin-derivatives. In contrast to known agents, this EGF-toxin conjugate promoted its own uptake via receptor microclustering even in the presence of Her2 and induced cell death with a LC₅₀ < 1 nM. Furthermore, our data showed that exposures as short as ≈3 min were enough to commit human (T24), mouse (MB49) and canine (primary) bladder cancer cells to apoptosis. Exposure of tumor-free mice and dogs with the agent resulted in no toxicity. In addition, the EGF-toxin was able to eliminate cells from human patient tumor samples. Importantly, the administration of EGF-toxin to dogs with spontaneous bladder cancer, who had failed or were not eligible for other therapies, resulted in ~30% average tumor reduction after one treatment cycle. Because of its in vitro and in vivo high efficiency, fast action (reducing treatment time from hours to minutes) and safety, we propose that this EGF–anthrax toxin conjugate provides the basis for new, transformative approaches against bladder cancer.

Introduction

Bladder cancer poses a significant public health concern with an estimated 81,190 new cases and 17,240 related deaths in 2018 just in the United States.¹ Importantly, 70% of newly diagnosed patients will suffer disease recurrence after transurethral resection of a bladder tumor and more than 20% will develop invasive bladder cancer.² In fact, the majority of patients return frequently for office visits, cystoscopic procedures and intravesical treatments, making bladder cancer one of the most expensive malignancies to treat.^3–5

Despite efforts to address these issues with adjuvant therapies involving the intravesical instillation of chemotherapeutic and immunomodulatory agents, high rates of disease recurrence and progression continue to persist. These strategies (e.g., using mitomycin C or Bacillus Calmette–Guerin [BCG]) require at least 2 hr-long treatment per session^6,7 and have been linked to local and systemic side effects including urinary symptoms, cystitis, fever and inflammatory response.^8–12 This situation is worsened by the current BCG shortage.^13,14 Therefore, there is a pressing need for more efficient therapeutic approaches against bladder cancer. Importantly, novel efficient countermeasures against this malignancy should be able to exploit the unique advantages that the bladder architecture presents while overcoming its specific challenges.

Umbrella cells protect the epithelia from contact with urine by forming tight junctions, depositing an apical semicrystalline uroplakin coat and by bearing an insulating glycosaminoglycans (GAG) layer.^15–17 However, since malignant bladder cells are less differentiated, they present a poorly assembled GAG layer, virtually no uroplakin deposits and are loosely associated with each other.^16,18 These characteristics result in a differential bladder lumen exposure of cancer vs. normal cells,¹⁶ thereby presenting a great opportunity for treating tumors while minimizing effects on normal cells. Nevertheless, dilution of the bladder content by urine influx and elimination of therapeutics due to bladder voiding requires the use of targeted anticancer drugs against tumor cells.¹⁹

Since bladder cancer cells are known to upregulate epidermal growth factor (EGF) receptor (EGFR) expression,²⁰ we tested the use of an EGF-targeted bacterial toxin designed to bypass the limitations of other EGF/toxin-based approaches as an antibladder cancer agent: Since EGFR-targeting therapeutics require receptor endocytosis to be active, impaired EGFR uptake (e.g., by Her2/neu upregulation or the presence of mutations affecting EGFR internalization) are predicted to negatively impact their efficacy.^21–23 Importantly, to bypass such limitations we previously devised an approach relying on receptor microclustering to induce nanoparticle uptake by bladder tumor cells.²⁴ These microclustering effects can also be elicited by multivalent agents such as anthrax toxin that, in contrast to monovalent toxins (e.g., diphtheria toxin^25–27), forms oligomers at the cell surface promoting its own internalization.²⁸

Here we present evidence that indicates that an EGF–anthrax toxin fusion efficiently targeted and eliminated human, mouse and canine bladder tumor cells, offering a novel, efficacious and fast strategy (minutes vs. hours with current treatments) against both superficial and invasive bladder cancer. Furthermore, our data indicate that Her2 was overexpressed in dog tumors and nevertheless the toxin was highly effective in vitro and in vivo against canine bladder cancer cells. Indeed, six dogs with terminal, treatment-resistant bladder cancer exhibited consistent tumor mass reduction because of exposure to this EGF-toxin. Importantly, in addition to being very specific for EGFR overexpressing cells, this agent is safer than others²⁷ due to its binary nature.^28,29 Although this approach is intended to be administered intravesically (NOT systemically) and at a low nM dosage, if any leakage into the bloodstream/adjacent organs would occur, each agent’s component would be independently diluted, making toxin reassembly virtually impossible.²⁹

In conclusion, we believe that the studies presented in this work provide the basis for the design of an innovative and transformative therapeutic strategy against bladder cancer.

Materials and Methods

Protein expression and purification

EGF-PA′ and lethal factor N-terminus fused to the catalytic domain of diphtheria toxin A (LF_N-DTA) were expressed by the BL21 (DE3) E. coli strain (New England Biolabs, Ipswich, MA), under the induction of 1 mM IPTG for 2 hr at 30°C. EGF-PA′ was purified as described previously.³⁰ Recombinant LFN-DTA was expressed as a His₆-SUMO fusion using the Champion pet-SUMO expression system (Invitrogen, Carlsbad, CA) and purified as previously described.³¹

Cell sources, origin, maintenance and transfection

Human T24 (RRID: CVCL_0554) bladder cancer cell line was obtained from ATCC while murine MB49 (RRID: CVCL_7076) was provided by the Ratliff lab. The MB49 low EGFR-expressing (MB49^LE) cell line is a spontaneous variant isolated during our study. All cell lines were cultured in Dulbecco’s modified Eagle medium (DMEM; Sigma, St. Louis, MO), 5% Penicillin/Streptomycin (Corning, Corning, NY), 5% l-glutamine (Corning) and 10% fetal bovine serum (Atlas Biologicals, Fort Collins, CO) while maintained at 37°C in the presence of 5% CO₂.

All experiments were performed with mycoplasma-free cells and the human T24 cell line was authenticated using STR profiling.

Patient tumor disaggregation.

Samples from freshly resected human or canine bladder tumors were washed three times with phosphate-buffered saline (PBS), transferred onto a 150 mm-dish where they were minced with a sterile razor blade, 10 ml of 5 mg/ml or 5% w/v collagenase type 4 (Worthington Biochemical, Lakewood, NJ) were added and incubated at RT for 45 min. The tissue culture dish was then placed at 37°C for 30 min. The sample was transferred to a 50 ml tube and centrifuged at 1,200 rpm for 5 min. The supernatant was discarded, and the pellet was washed three times with PBS and centrifuged once more. After being resuspended using a 10 ml transfer pipette, cells were then passed through a 70 μm cell strainer (BD Biosciences, San Jose, CA) to achieve a single-cell suspension. Cells were resuspended in fully supplemented DMEM+20% fetal bovine serum (FBS) and plated on fibronectin-coated six-well plates (Biomedical Technologies Inc., Stoughton, MA).

Transfection.

MB49^LE cells were seeded in a six-well plate containing glass coverslips and allowed to grow to 70% confluence to then be transfected with an EGFR-green fluorescent protein (GFP) plasmid using the Fugene6 (Promega, Madison, WI) transfection reagent according to the manufacturer’s protocol. Then, 24 hr later, the cells were fixed in 4% formaldehyde, mounted on slides and the transfectants were visualized via epifluorescence microscopy using a GFP filter.

EGF–TMR and EGF-toxin binding and internalization

All experiments with human cells were done in the presence of PBS and 50% human urine, while those using murine and canine cells were performed in PBS supplemented with 50% murine and canine urine, respectively. When needed, cytotoxicity was assessed using standard MTT assays.

Stepwise incubation format.

(A). Fluorescently tagged EGF:

2.0 × 10⁵ cells were seeded in fully supplemented media on coverslips in six-well plates, the cells were serum-starved for 12 hr just before use. Cells were rinsed with ice-cold PBS and incubated with 0.5 μg/ml (70 nM) EGF conjugated with tetra-methyl rhodamine (EGF-TMR; Invitrogen, Carlsbad, CA) in PBS/50% urine at 4°C for 45 min and then washed with ice-cold PBS to remove any unbound ligand. Next, coverslips were transferred to 37°C in prewarmed PBS/50% urine for 0 and 30 min to allow internalization. Coverslips were rinsed in ice-cold PBS and the bound, but noninternalized ligand was removed by washing with ice-cold acidic buffer (0.2 M glycine, 0.15 M NaCl, pH = 3.0) for 45 sec, washed and fixed in 4% formaldehyde in PBS for 10 min at RT. Coverslips were mounted on slides and 15 randomly chosen fields were imaged using a Zeiss Axiovert 200 M epifluorescence microscope.

(B). EGF-toxin:

Approximately 2–3 × 10⁴ cells (to yield MTT assay’s high sensitivity medium linear range) were seeded at least 24 hr prior to experiments, and they were serum starved for 12 hr before treatment. Cells were then washed with ice-cold PBS to remove residual FBS and treated with varying concentrations of EGF-PA′ for 45 min at 4°C. After the incubation period, cells were washed with ice-cold PBS to remove any unbound ligand. LF_N-DTA (10 nM) in PBS (at 37°C) was added to cells and incubated for times ranging from 0 to 30 min (controls without LF_N-DTA were also performed). After incubation period, cells were washed with ice-cold PBS. Bound but noninternalized ligand was removed by washing with ice-cold 0.2 M glycine pH = 3.0, 0.15 M NaCl for 45 sec. Cells were then washed with ice-cold PBS and placed in fully supplemented DMEM media for times ranging from 48 to 72 hr at which point viability was assessed using MTT assays. Experiments were performed by triplicate and at least three times.

(C). EGF-toxin/EGF competition assay:

As described above, cells were plated in six-well plates and serum starved for at least 12 hr prior to treatment. Cells were incubated with 0.63 nM EGF-PA′ with and without 100-fold higher concentrations of wild-type, unlabeled EGF (Cytoskeleton). After EGF-PA′ binding in either the absence or the presence of wild-type EGF for 45 min at 4°C, the cells were washed in PBS and then incubated with 10 nM LF_N-DTA for 30 min at 37°C. After rinsing with PBS, the cells were washed with acidic buffer (pH = 3) to strip off the bound, but noninternalized, EGF-PA′/LF_N-DTA complexes as described above. Cells were washed in PBS once more and placed in fully supplemented DMEM. Then, 48 hr later, cell viability was determined using MTT assays.

Simultaneous incubation format.

(A). Fluorescently-tagged EGF:

2.0 × 10⁵ cells were plated and serum starved as indicated above. Cells were then rinsed with PBS and incubated with 0.5 μg/ml (70 nM) EGF-TMR at 37°C in PBS/urine for the desired length of time. Cells were then washed with PBS and fixed in 4% formaldehyde in PBS for 10 min at RT and imaged as described above.

(B). EGF-toxin:

2–3 × 10⁴ cells were seeded, and serum starved as indicated above. Cells were then treated and washed in PBS and treated with a single cocktail comprising of EGF-PA′ + LF_N-DTA at varying concentrations for the desired length of time. Cells were then washed in PBS and placed in fully supplemented DMEM for 48–72 hr. Viability was assessed using MTT assays. Dose–response curves were plotted and nonlinear regression was applied using the GraphPad Prism software.

Alternatively, apoptotic, necrotic and healthy cell populations were determined using the Apoptosis, Necrosis & Healthy Cell Quantitation Kit (Biotium, Fremont, CA) according to the manufacture’s protocol. Cells were then visualized by fluorescence microscopy.

In vivo antitumor effect of the EGF-toxin on canine spontaneous bladder cancer

After approval of the Purdue Animal Care and Use Committee, the safety of the EGF toxin was confirmed in four laboratory dogs, and then the EGF toxin was tested in a pilot study of six tumor-bearing dogs. The dogs had naturally occurring histopathologically confirmed urothelial carcinoma expressing EGFR detected by immunohistochemistry. These dogs were ones presented to the Purdue University Veterinary Teaching Hospital for treatment of their bladder cancer and they had either failed other treatments or were not eligible for other bladder cancer treatments. The dog owners were informed of the trial and they elected to enroll their dog, the dog owner provided written informed consent. If the dogs had previously received cyclooxygenase (COX)-inhibitors, and had experienced tumor progression on the COX-inhibitor, but needed the medications to control arthritis pain or other painful conditions, they were allowed to continue to receive the COX-inhibitor while in the pilot study. Since COX-inhibitors have antitumor activity in dogs with bladder cancer, it was important to only allow the drugs if cancer progression had occurred on that medication prior to the pilot study.³² Before and after treatment, the dogs were evaluated with blood work (CBC, serum biochemical profile), urinalysis, thoracic radiography, abdominal ultrasonography and a standardized bladder ultrasound protocol³³ to measure the bladder masses.³³ The ultrasound exams were standardized for machine, operator, dog position, probe position and angle, degree of bladder distension and image analysis program. When this standardized protocol is followed, the interassay variability was <10% and was used to ascertain tumor size.

The dogs were treated with the EGF toxin for up to 5 days in a week (one cycle). Prior to the intravesical administration of EGF-toxin, dogs were sedated with 0.2 mg/kg butorphenol-IV and 3–5 mg/kg dexdomotor-IV and a urinary catheter was passed transurethrally. The bladder was then emptied of its contents and washed with 50 ml of saline solution for 5 min; this step was repeated twice. The EGF-toxin was then instilled intravesically via the urinary catheter into the bladder at a concentration of 200 nM EGF-PA′/400 nM LF_N-DTA (in 25–75 ml total volume depending on the size of the dog) and left for 1 hr. During this time the dog was positioned in right lateral recumbency, then shifted to dorsal recumbency and then shifted to left lateral recumbency, being shifted every 20 min. After the 1 hr-dwell time, the solution was removed from the bladder and the bladder washed with 50–100 ml of PBS. The sedation was then reversed by administering Atipamazole (the same dosage as Dexdomotor-IV), and the dog was allowed to recover.

Statistical analysis

Normally distributed data were represented as the mean ± standard deviation of triplicate measurements. Statistical significance of value differences was assessed by applying the t-test. Some biological data analyzed within this work did not follow a normal distribution (failed normality tests). Given their nature, data distributions were shown as box-plots (constructed with Sigmaplot-14) depicting the median, 25–75 (bottom and top of box, respectively) and 10–90 (whiskers) percentiles. When appropriate, statistical significance was evaluated using the nonparametric Wilcoxon test.

Supplemental Methods can be found in the Supporting Information Materials section.

Data availability

Data is available upon request.

Results

The EGF-PA′/LFN-DTA toxin system

To design efficient antibladder cancer approaches it is necessary to consider the bladder’s unique advantages and challenges as a target for therapy: on the one hand, and in contrast to normal bladder epithelia, cancer cells are exposed to the lumen of the bladder (Fig. 1a). On the other hand, the potential beneficial impact of the instillation of nontargeted therapeutics on tumor cells is limited by agent dilution due to constant urine influx, and elimination caused by periodic voiding of the bladder. Therefore, only high affinity targeted therapeutics are expected to overcome these challenges.

Figure 1.

Bladder tumor cells can be targeted by an EGF-toxin. (a) In contrast to normal differentiated epithelial umbrella cells, bladder tumor cells are known to be deficient for GAG layer synthesis, uroplakin deposition and formation of tight junctions. Therefore, tumor cells are exposed to the lumen of the bladder and they overexpress EGFR. (b) Mechanism of action of the EGF-toxin. The specificity of a mutated version of the anthrax protective antigen (PA′: Blue oval—unable to recognize the anthrax receptor), is redirected to EGFR by fusion to EGF (star). After EGFR binding, PA′ is proteolytically processed and activated (not shown) allowing the assembly of an oligomer on the plasma membrane. This complex binds the anthrax lethal factor N-terminus (LF_N) fused to the catalytic domain of diphtheria toxin (DTA—the LF_N-DTA fusion is represented as a green triangle). After internalization, the lower pH of the endosome induces a conformational change in the EGF-PA′ oligomer that promotes its insertion into the endosomal membrane and LF_N-DTA translocation into the cytosol. LF_N-DTA catalyzes the ADP-ribosylation of eukaryotic elongation factor 2 (eEF2) triggering apoptosis. (c) Bladder cancer cells can be targeted by EGF and the EGF-toxin. Upper panel: Fluorescently (tetra-methyl-rhodamine: TMR)-labeled EGF was bound and internalized by the human T24 and mouse MB49 bladder cancer cell lines and canine primary bladder tumor cells in the presence of saline and 50% of corresponding urine, as detected by epifluorescence microscopy. Scale bar: 20 μm. Lower panel: 20,000 cells of the indicated origin were plated per well on six-well plates and incubated for 8 min in presence or absence (controls) of the indicated reagents in saline supplemented with 50% corresponding urine. After 48 hr cell viability was measured by MTT assays and represented as % of the control. (d) The presence of EGFR and Her2 was investigated by Western blotting on whole lysates from canine tumor cells using specific antibodies (detection of Actin was used as a loading control). Representative results are shown.

Since it is well-known that bladder cancer cells overexpress EGFR,²⁰ we speculated that an EGF-targeted bacterial toxin would be effective in bladder cancer therapeutics. The EGF-toxin system mechanism of action^30,31,34 is depicted in Figure 1b. Briefly, an EGF–anthrax protective antigen mutant fusion-protein (EGF-PA′; where PA′ stands for a mutant PA unable to bind anthrax receptor) recognizes EGFR, is proteolytically activated by a cellular, furin-family protease, and assembles as an heptameric or octameric prepore complex on the plasma membrane and recruits up to three molecules of LF_N-DTA (Fig. 1b). In contrast to other EGF-agents,²⁷ endocytosis of this EGF-toxin is triggered by PA′ oligomer-mediated microclustering of EGFR. After internalization, the lower pH of the endosome induces a conformational change in the EGF-PA′ oligomer leading to pore formation and translocation of LF_N-DTA molecules into the cytosol, which in turn initiates apoptosis by inactivation of elongation factor-2 (EF2; Fig. 1b).

Binding and internalization of EGF-TMR and EGF-toxin in the presence of instillation buffer and urine

To determine the viability of the proposed strategy, we took advantage of the fact that human (h) EGF can also bind mouse and dog EGFR. Specifically, we tested if under conditions that emulate the environment of the bladder during treatment (instillation buffer in the presence of urine and absence of bovine serum), fluorescently labeled human epidermal growth factor (hEGF) was capable of targeting human (T24), mouse (MB49) bladder cancer cell lines as well as dog bladder cancer cells (isolated from canine spontaneous bladder tumors).

During a typical treatment instillation (i.e., after bladder voiding and rinsing) urine is expected to be present at less than 50% of the total bladder content; however, we exaggerated its contribution in the assays to maximize putative detrimental effects of its components (e.g., urea) on receptor binding.

After 30 min of incubation of cells with EGF-TMR in saline solution + 50% of homologous urine at 37°C, we proceeded to eliminate unbound ligand by rinsing, and to remove the bound but noninternalized fraction using an acidic wash.²⁴ Our results indicate that soluble EGF was able to bind its receptor in the presence of saline and urine, and it was also internalized by the different bladder cancer cells (see Fig. 1c, upper panels for representative images). Accordingly, the EGF-toxin was highly efficient for elimination of human, mouse and dog cancer cells under the same conditions (Fig. 1c, lower panel). It should be noted that the EGF-toxin was highly effective against dog tumor and T24 cells even when these were Her2-positive (Fig. 1d, Supporting Information Fig. S1). This is important as Her2 is known to interfere EGFR internalization^21,22 and RNA-seq data from 28 dog bladder tumors (vs. 4 normal urothelial layer) indicated that Her2 was enriched in dog bladder tumors in these animals (Supporting Information Fig. S2).

EGF-toxin action on bladder cancer cells

Next, we utilized a stepwise approach to characterize three different aspects of toxin action on bladder cancer cells: toxin binding to EGFR (Fig. 2), and the time-courses of toxin oligomer assembly/internalization/LF_N-DTA translocation (Fig. 3), and toxin-induced cell death (Fig. 4).

Figure 2.

Binding of the EGF–toxin to human bladder cancer cells. (a) Left: Cartoon depicting binding of the EGF-PA′ to cells (Fig. 1b for symbol/scheme meaning). Right: EGF-toxin binding and specificity: drug–response relationship. The stepwise experimental setup described in materials and methods was performed incubating serum-“starved” T24 cells with different concentrations of EGF-PA′ for 45 min on ice; after that, unbound ligand was removed by washing. Incubation with LF_N-DTA at 37°C was conducted for 30 min (followed by elimination of noninternalized ligand), and MTT assays were performed 48 hr after toxin exposure. (b) In order to test the EGFR specificity of the EGF-toxin, a 100× excess of unmodified EGF was added to compete off EGF-PA′ binding to the receptor.

Figure 3.

Kinetic of toxin assembly and Internalization. Left panel: After EGF-PA′ clustering, LF_N-DTA is bound and the complex internalized allowing the latter to be translocated into the cytosol (see Fig. 1b for symbol/scheme meaning). Right panel: binding of EGF-PA′ at either LC₅₀ (top) or LC₁₀₀ (bottom) concentrations was conducted on ice, and allowed complex assembly and internalization for different times before stripping off noninternalized protein and adding complete media. MTT assays were performed 48 hr later.

Figure 4.

Kinetics of toxin action/cell death. (a) Left panel: LF_N-DTA (green triangle) ADP-ribosylates eEF2 promoting apoptosis in bladder cancer cells. Right panel: 10⁴ cells were incubated with LC₁₀₀ concentration of EGF-PA′ on ice, followed by addition of 10 nM LF_N-DTA at 37°C for 30 min to allow assembly and uptake of the toxin. MTT assays were performed at the indicated times. (b) Cells were seeded at 100% confluence on glass coverslips within six-well plates. At the indicated times after EGF-toxin treatment the cells were fixed, stained and imaged (upper panel). The cell occupancy (fraction of area cover by cells) was determined using ImageJ software. Average occupancy (± standard deviation) of three independent experiments is shown/indicated.

Binding: establishment of the dose–response relationship for the EGF-toxin system.

We exposed the cells to a range of EGF-PA′ concentrations at 4°C for 45 min (Binding step, Fig. 2a, left) followed by removal of the unbound fraction, while maintaining constant 37°C incubation time at 30 min (prepore assembly, complex internalization and LF_N-DTA translocation). After that, we eliminated bound, but noninternalized ligand by acidic wash (the efficiency of this wash was controlled by monitoring the removal of receptor-bound EGF-TMR in a parallel sample—see Materials and methods). Then the cells were incubated at 37°C in complete media for 48 hr and subjected to MTT cell viability assays or cell imaging. Using this experimental set up with different bladder cancer cells, we determined their corresponding EGF-toxin lethal concentration required to kill 50% of the population (LC₅₀; Fig. 2a, right, Table 1). Specifically, human T24, mouse MB49 and cells isolated from several spontaneous canine tumors showed similar sensitivity to the EGF-toxin with a LC₅₀ ranging from 0.21 to 0.54 nM (Table 1). Importantly, addition of excess of WT EGF competed-off EGF-PA′ binding to EGFR, suppressing the EGF-toxin cytotoxic effect (Fig. 2b). These results clearly indicate that the EGF-toxin activity depends on the specific recognition of EGFR. Further supporting the idea that sensitivity to the EGF-toxin depends on the levels of EGFR expressed by the targeted cell, the MB49^LE cell line variant (shown to have low levels of EGFR—Supporting Information Fig. S3A) had a LC₅₀ two orders of magnitude above the one from the MB49 parental cell line (Table 1). Importantly, transfection of MB49^LE cells with a construct for the expression of EGFR-GFP from a cytomegalovirus promoter greatly enhanced the cell line sensitivity to the EGF-toxin (Supporting Information Fig. S3B). Real-time quantitative reverse transcription-polymerase chain reaction (QRT-PCR) results indicated that EGFR messenger RNA levels of MB49^LE cells were comparable to mouse normal bladder epithelial cells (Supporting Information Fig. S3A).

Table 1.

EGF-toxin lethal concentrations for different bladder cancer cells

Bladder cancer cells	LC50 (nM)	LC100 (nM)
Human T24	0.31 ± 0.05	2.2 ± 0.5
Mouse MB49WT	0.21 ± 0.02	1.9 ± 0.2
MB49LE1	20 ± 6	205 ± 29
Spontaneous canine tumor cells2	0.33 ± 0.04	2.6 ± 0.4
	0.54 ± 0.07	3.6 ± 0.6
	0.34 ± 0.073	2.5 ± 0.73

¹MB49^LE: variant expressing very low levels of EGFR.

²Cells obtained from disaggregation of tumor biopsies. All samples except one (see note 3) were immortalized.

³Primary culture obtained from one of the dogs treated with the EGF-toxin.

In addition, our results indicate that Her2 is expressed in T24, but not in MB49 cells (Supporting Information Fig. S1); however, both cell lines presented similar sensitivity to the EGF-toxin (Table 1).

Toxin oligomer assembly and internalization time-course.

Next, we maintained constant the EGF-PA′ concentration to its LC₅₀ value during binding at 4°C while we varied the duration of the incubation at 37°C (after unbound ligand elimination; Fig. 3). This experimental design was used to determine the time-dependence for the assembly/internalization/translocation phase of the EGF-toxin action (Fig. 3, left). Specifically, we established that using EGF-PA′ LC₅₀, 15–30 min were needed to achieve 50% cell elimination (Fig. 3 right, upper panel).

Importantly, under the same experimental conditions, but using EGF-PA′ at its absolute lethal concentration (LC₁₀₀), we established that exposure times to the toxin as short as 3 min assured maximal tumor cell killing by the toxin (Fig. 3, right, lower panel). These results suggest that at this dose, the amount of pore assembled and the fraction of LF_N-DTA released into the cytosol during this short exposure time is sufficient to assure tumor cell elimination.

Toxin-induced cell death time-course.

After incubating cells with EGF-PA′ LC₁₀₀ for 45 min at 4°C, unbound toxin was eliminated, and the samples were incubated at 37°C for 30 min (Fig. 4). Next, the bound noninternalized ligand was stripped off using acidic wash, and the toxin effects on the cells were monitored by MTT and apoptosis assays at different times (Fig. 4 and Supporting Information Fig. S4, respectively).

Four hours after removal of the toxin, we detected fluorescent annexin-V binding on treated cells, a phenomenon indicative of membrane alterations (exposure of phosphatidylserine in outer leaflet) compatible with apoptosis (Supporting Information Fig. S4). However, by MTT assays, maximal cell death was observed between 24 and 48 hr after exposure to the toxin (Fig. 4a, right panel). In addition, we also conducted experiments in which confluent T24 cell monolayers were exposed to complete/reconstituted EGF-toxin and cell elimination was monitored periodically and quantitatively by microscopy (Fig. 4b). These experiments, in agreement with the toxin-induced cell death time-courses, also indicated sustained cell loss as a function of time, with maximal cell elimination by 24–48 hr (Fig. 4b).

Incubation of human tumor cells with reconstituted EGF-toxin

Exposure of T24 cells to EGF-PA′/LF_N-DTA in saline/urine, at 37°C and without elimination of unbound or noninternalized proteins did not result in any significant change in EGF-toxin sensitivity by the studied cells as compared to the stepwise approach results (Supporting Information Fig. S5A). In addition, using EGF-PA′ at its LC₅₀, we titrated the LF_N-DTA concentration needed to obtain the expected 50% cell population elimination. Our results showed that LF_N-DTA concentrations below 10 nM were suboptimal for supporting 50% cell death by the EGF-PA′ LC₅₀ (Supporting Information Fig. S5B).

Importantly, primary cultures obtained from freshly resected human bladder tumors were tested for EGF-TMR binding/internalization and cell elimination by the EGF-toxin. Our results showed that similar to other bladder cancer cells, primary tumor cells from eight different patients were able to bind and internalize EGF-TMR in the presence of PBS and urine (Figs. 5a and 5b shows two representative examples). In contrast to other cell sources, we could not verify EGFR overexpression a priori for these samples. In fact, our data indicated that there was some degree of EGFR expression level variation among patients, and even some variability within tumor cells from a single patient (Figs. 5a and 5b). However, our results showed that overall patient cancer cells displayed sensitivity to the EGF-toxin yielding an 85% median elimination rate of tumor cells using 2 nM EGF-PA′/10 nM LF_N-DTA concentration (Fig. 5c).

Figure 5.

Determination of the effect of the EGF-toxin on human bladder cancer patient cells. Human bladder cancer patient cells bind fluorescent TMR-EGF as detected by epifluorescence microscopy (a) and flow-cytometry (b). Note the EGF-binding heterogeneities within Patient 1 tumor cells: panel a-left shows a cell with substantial EGF-TMR binding (asterisk) accompanied by two others with lower binding capacity. A similar result was obtained by FACS analysis, Patient 1 exhibited two populations of cells: one with substantial EGF-TMR binding capacity than the other (red line peaks). Patient 4 cells appear more homogeneous with high levels of EGF-TMR binding. Scale bar: 20 μm. (c) In total, 2 × 10⁴ cells from eight patients were treated with reconstituted EGF-toxin (2 nM PA′-EGF/10 nM LF_N-DTA). Viability was measured by MTT assays 48 hr after and showed a significant decrease in cells treated with the reconstituted EGF-toxin as compared to the control (**p < 0.05; Wilcoxon test). EGF-toxin induces bladder tumor reduction in dogs afflicted with spontaneous bladder cancer. The photomicrograph in d is a paired negative control for EGFR immunohistochemistry. In e, note the striking membrane immunoreactivity (brown color) in the cancer sample. Some immunoreactivity was noted in normal urothelium as previously reported.⁴⁴ (f) Ultrasound images captured from a dog with invasive urothelial carcinoma before (PRE, top two images) and after (POST, bottom two images) treatment with EGF toxin. The estimated tumor volume (sagittal area × transverse dorsal–ventral dimension) was 5.7 cm³ pretreatment and 3.9 cm³ posttreatment, indicating ~30% reduction in tumor volume. This change was noted after one 5-day course of treatment.

The EGF-toxin displays antitumor activity in vivo in dogs with spontaneous, terminal bladder cancer

As the first step toward in vivo studies, we tested the EGF-toxin for potential adverse effects in tumor-free animals. Specifically, the toxin was instilled into the bladder of control animals: six mice and four dogs. No toxicity was detected in the animals by any analysis made (including daily observation and physical exam, urinalyses, complete blood counts and serum biochemical profiles for over a month—data not shown). Therefore, we proceeded to treat six dogs with bulky (in some cases partially blocking the exit of urine), naturally occurring invasive urothelial carcinomas, resistant to conventional therapies with 10 × LC₁₀₀ dose of EGF-toxin. The overexpression of EGFR (as compared to controls) was verified in these dogs by immunohistochemistry (Figs. 5d and 5e).

In all cases, dogs treated with the EGF-toxin had some reduction in tumor volume (even when multiple tumors were present) with an average of ~30% decrease in size after a single treatment cycle. In some cases, the tumor size was reduced by 10%/day during the 5 days of treatment (see example in Fig. 5f and Table 2). This response after just one treatment cycle is encouraging because invasive urothelial carcinoma in dogs closely mimics human invasive bladder cancer in behavior and treatment response.³² Importantly, the EGF-toxin treatment was very well tolerated. Only a mild bladder irritation (slight increase in amount of blood in the urine and increase in frequency of urination) was noted by treatment cycle Day 5. These resolved within 2 days after treatment ended.

Table 2.

EGF-toxin treatment of dogs affected by bladder cancer

Dog #	Tumor volume (cm3)1			Number of doses2	Age, sex, breed	Weight (kg)
	Before	After	Change (%)
1	36.6	22	40	5	12 years, Male, Mixed	23
2	101.2	81.1	20	3	12 years, Male, Collie	28
3	4.9	3.2	35	5	9 years, Male, Sheltie	17
4	6	5.3	12	5	12 years, Female, Mixed	31
5	5.7	3.9	31	5	11 years, Male, Beagle	14
6	3.9	2.5	36	4	12 years, Male Beagle	14

¹Measured by ultrasound (see Materials and methods).

²Within a single treatment cycle.

Discussion

The results presented in our study set the basis for the development of a therapeutic strategy against bladder cancer considered transformative due to its (i) high efficiency for targeting bladder cancer cells; (ii) fast action, that is, having the potential of drastically decreasing patient treatment time from hours (current therapies) to minutes; (iii) its ability to eliminate Her2-positive cells; (iv) demonstrated in vivo efficacy and (v) its enhanced safety.

The constant influx of urine into the lumen of the bladder and the necessary voiding of the bladder content are serious challenges to the instillation of nontargeted therapeutics against bladder cancer.¹⁹ This makes the devising of targeted approaches a must for efficacious antibladder cancer strategies.

Since EGFR is overexpressed in both superficial and invasive bladder carcinomas,²⁰ this receptor emerges as a suitable target against bladder cancer. However, use of the anti-EGFR monoclonal antibody cetuximab against bladder cancer have yielded disappointing results, failed to prevent disease progression and developed drug resistance/patient toxicity.^35–37

Since internalization of EGFR complexes plays an important role in the success of these targeted agents, the presence of EGFR internalization-impairing mutations or upregulation of Her2/neu (known to interfere with EGFR endocytosis²¹) are among the factors known to impair the efficacy of EGF-based strategies.^21–23

Therefore, and based on previous developments from our lab²⁴ we decided to use an agent that while targeting EGFR, would be able to induce its own internalization through a clustering-dependent mechanism. Specifically, we used a chimeric protein resulting from the fusion of EGF (targeting ligand) and the binary anthrax toxin (cytotoxic component and microclustering active element). Very importantly, endocytosis of this agent is triggered by oligomerization of the PA subunit of the anthrax toxin (Fig. 1),²⁸ that is, contrary to other toxins (e.g., diphtheria toxin^25–27) anthrax assembles oligomers at the plasma membrane inducing its own uptake.²⁸Therefore, as opposed to other EGF-based technologies, the EGF-toxin is able to bypass Her2s negative effect on EGFR internalization (Figs. 1c and 1d; Table 1 shows similar sensitivity by the Her2⁺ T24 cells and the Her2⁻ MB49 line) and is predicted to be insensitive to EGFR mutations that inhibit its endocytosis.

This agent not only fills-in an important need for better, more efficient antibladder cancer therapeutics, but we humbly think that it will be transformative for bladder cancer treatment for the following reasons.

1. This is an approach of very high efficacy (with subnanomolar LC₅₀) against bladder tumor cells—Figures 1 and 2, Table 1. It is known that more than 90% of the EGFR in cells display an affinity (as measured by EGF-EGFR dissociation constant, K_D) for EGF of ~2 nM.³⁸ As expected, since our data reflects biological activity, the LC₅₀ range value for EGFR-expressing bladder cancer cells (~0.2–0.5 nM; Table 1) is substantially lower than the KD. This is often seen with pharmacologically active ligands^39,40 and suggests that is not necessary to bind a substantial number of receptors on the cell to trigger maximal biological activity, this is expected for the EGF-toxin considering the enzymatic nature of LF_N-DTA.

   Importantly, the EGF-toxin approach is very specific for cells overexpressing EGFR; that is, cancer cells (Fig. 2b, Table 1 and Supporting Information Fig. S3B). Indeed, we found that the EGFR-low level expressing MB49^LE cells are two orders of magnitude less sensitive than the MB49 original cell line (Table 1). However, transfection/expression of GFP-EGFR in these cells substantially increased their sensitivity for the EGF-toxin (Supporting Information Fig. S3B).

2. The EGF-toxin exhibits very fast action (min vs. hr of current treatments). The time-course of toxin action (Fig. 3) indicates that the kinetics of PA-clustering mediated uptake is not substantially different from the rate of other ligand-receptor complexes. Indeed, pulse-chase experiments indicate that 5 min after binding, a substantial fraction of EGF can be found in the early endosomal compartment.^41,42 This is very important because it points out to the potential of this agent to drastically reduce patient treatment time from hours to minutes.

3. The EGF-toxin is capable of eliminating tumor cells even in the presence of Her2. The EGFR-related protein Her2 is a poor prognosis marker, but also an EGFR internalization-interfering factor²¹; however, our results clearly demonstrate that the EGF-toxin is highly effective (Figs. 1c and 1d, Table 1) independently of Her2 expression. These observations indicate that the microcluster-induced endocytosis mechanism^24,28 overrides Her2-mediated inhibition of EGFR internalization. In fact, while uptake of conventional EGF-derived agents²⁷ occurs by a mechanism initiated by EGFR dimerization affected by the formation of EGFR-Her2 dimers, EGF-PA′ triggers a PA-mediated microclustering internalization event.²⁸ Therefore, we also predict that this agent would be able to eliminate cells displaying EGFR-internalization mutants.

4. This approach is efficacious in vivo, even when used in patient dogs refractory to all other available treatments (Figs. 5d–5f, Table 2). In addition to be readily applicable to superficial bladder cancer, we speculate that its high efficacy may turn this strategy into a substitute for bladder resection in some invasive bladder cancer cases.

   Cell killing experiments performed on T24 cell monolayers in which we monitored toxin action by direct observation of cell elimination (creation of “holes” in the cell monolayer), suggested that as early as 4–8 hr after treatment, the outer layer of a putative tumor would be breached to an extent that a second toxin application at this time would allow access to the underlying layers of the tumor. Furthermore, the looser tumor cell junctions would allow deeper tumor penetration of this molecular-sized agent as compared to nanoparticulated entities or bacteria (e.g., BCG). We speculate that these observations should be taken into account for the design of future treatment strategies using this agent.

5. The EGF–toxin is a safe approach. This approach is safer than any other toxin-based approach reported in the literature.²⁷ Among the factors that assure the safety of this approach: (a) the toxin is instilled in the lumen of the bladder (NOT in the bloodstream); therefore, as discussed above, only the exposed tumor cells are accessed by the agent. (b) Even in the event of a toxin leak into circulation, our data indicate that due to their heightened sensitivity (see item #1 above) the dose required for bladder cancer cell elimination is substantially lower than the one required for toxin intoxication (e.g., bladder cancer cell LC₁₀₀ ~ 2 nM vs. mice killing dose ~ 1 μM⁴³). Even if the entire volume of the mouse bladder instillate (80 μl) would to leak into the bloodstream (~1.5 ml), the resulting toxin concentration would be 10,000 times below the lethal dose. (c) In contrast to one-component agents (e.g., diphtheria toxin²⁷), binary toxins (such as anthrax) are safer as dilution greatly decreases the probability of both components reconstituting on normal cells. In addition, when the potential toxicity of the bladder-instilled agent was tested in control animals (mice and dogs) it yielded negative results; furthermore, administration of the toxin to dogs with spontaneous bladder cancer led to substantial tumor reduction without discernible toxic side-effects.

To summarize, the approach presented in this article can be transformative for the treatment of superficial bladder cancer due to its high efficacy both in vitro and in vivo and its fast action. We also speculate that its high efficiency in vivo could make it a viable substitute for cystectomy in early stages of invasive bladder cancer.

What’s new?

Bladder cancer is hard to treat and after surgery has a notoriously high recurrence rate (70%). Here the authors propose a new therapeutic strategy combining epidermal growth factor (EGF) with anthrax toxin. Since bladder cancer cells are exposed to urine and carry high levels of EGF receptor, the toxin after intravesical application was specifically taken up by cancer cells and induced rapid apoptosis, regardless of whether cancer cells expressed another EGF receptor, Her2, or not. This underscores the broad potential of the approach for treatment of bladder tumors in the future.

Abbreviations:

ADP

adenosine diphosphate

BCG

Bacillus Calmette–Guerin

CMV

cytomegalovirus

DMEM

Dulbecco’s modified Eagle medium

EF2

elongation factor 2

EGF

epidermal growth factor

EGFR

epidermal growth factor receptor

FBS

fetal bovine serum

GAG

glycosaminoglycan

GFP

green fluorescent protein

hEGF

human epidermal growth factor

K_D

dissociation constant

LC₁₀₀

lethal concentration required to kill 100% of the population

LC₅₀

lethal concentration required to kill 50% of the population

LF_N-DTA

lethal factor N-terminus fused to the catalytic domain of diphtheria toxin A

MB49^LE

MB49 low EGFR-expressing

mRNA

messenger RNA

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide

nM

nanomolar

PA

protective antigen

PBS

phosphate-buffered saline

QRTPCR

real-time quantitative reverse transcription-polymerase chain reaction

RNA-seq

ribonucleic acid-sequencing

TMR

tetra-methyl rhodamine