PE38-based gene therapy of HER2-positive breast cancer stem cells via VHH-redirected polyamidoamine dendrimers

Breast cancer stem cells (BCSCs) resist conventional treatments and cause tumor recurrence. Almost 25% of breast cancers overexpress human epidermal growth factor receptor-2 (HER2). Here we developed a novel multi-targeted nanosystem to specifically eradicate HER2+ BCSCs. Plasmids containing CXCR1 promoter, PE38 toxin, and 5′UTR of the basic fibroblast growth factor-2 (bFGF 5'UTR) were constructed. Polyamidoamine (PAMAM) dendrimers functionalized with anti-HER2 VHHs were used for plasmid delivery. Stem cell proportion of MDA-MB-231, MDA-MB-231/HER2+ and MCF-10A were evaluated by mammosphere formation assay. Hanging drop technique was used to produce spheroids. The uptake, gene expression, and killing efficacy of the multi-targeted nanosystem were evaluated in both monolayer and spheroid culture. MDA-MB-231/HER2+ had higher ability to form mammosphere compared to MCF-10A. Our multi-targeted nanosystem efficiently inhibited the mammosphere formation of MDA-MB-231 and MDA-MB-231/HER2+ cells, while it was unable to prevent the mammosphere formation of MCF-10A. In the hanging drop culture, MDA-MB-231/HER+ generated compact well-rounded spheroids, while MCF-10A failed to form compact cellular masses. The multi-targeted nanosystem showed much better uptake, higher PE38 expression, and subsequent cell death in MDA-MB-231/HER2+ compared to MCF-10A. However, the efficacy of our targeted toxin gene therapy was lower in MDA-MB-231/HER2+ spheroids compared with that in the monolayer culture. the combination of the cell surface, transcriptional, and translational targeting increased the stringency of the treatment.

Gene constructs. The methods of making the gene constructs have been described in detail previously 22 .

Preparation of VHH-PEG-PAMAM and Trastuzumab-PEG-PAMAM conjugates. G5 PAMAM
was reacted with NHS-PEG3500-Mal in 2 mL of degassed phosphate buffer (pH 7.5) at a ratio of 1:10 (mol/ mol) under N2 atmosphere at room temperature with gentle shaking for 2 h. The removal of unreacted PEG molecules was carried out by ultrafiltration using Amicon ultrafilters (MWCO 10 kDa; Millipore, Schwalbach, Germany). The number of PEG groups introduced to PAMAM was determined based on TNBSA assay. For the conjugation of the anti-HER2 VHHs to maleimide at the distal end of the PEGylated PAMAM dendrimers, the anti-HER2 VHHs were first thiolated using Traut's reagent. Traut's reagent was added to the anti-HER2 VHHs in borate buffer (50 mM sodium borate, 0.1 M EDTA, pH 8.3) to produce a ratio of 1:10 (mol/mol) and the reaction was performed under N2 atmosphere at room temperature with gentle shaking for 1 h. The removal of the unbound Traut molecules and replacing the buffer with phosphate buffer were carried out by ultrafiltration using Amicon ultrafilters (MWCO 30 kDa). Moreover, the thiolated VHHs were added to the PEG-PAMAM to produce a ratio of 2:1 (mol/mol), and the reaction was performed under N2 atmosphere at room temperature with gentle shaking for 18 h. The final conjugates (VHH-PEG-PAMAM) were purified using Amicon ultrafilters (MWCO 50 kDa), lyophilized, and stored at − 80 °C. All of the procedures used for the preparaton of the www.nature.com/scientificreports/ Trastuzumab-PEG-PAMAM conjugates were the same as those of the VHH-PEG-PAMAM conjugates except that the molar ratio of Traut's reagent to Trastuzumab was 20:1 and the removal of the unbound Traut molecules were carried out by Amicon ultrafilters with a bigger pore size (MWCO 100 kDa). From now on, PEG-PAMAM, VHH-PEG-PAMAM, and Trastuzumab-PEG-PAMAM will be referred to as PG-PAM, VHH-PG-PAM, and Tra-PG-PAM, respectively. Dendriplex characterization. Gel retardation assay. The binding of the pG-CX-bF-PE to PAMAM, PG-PAM and VHH-PG-PAM were corroborated by agarose gel retardation assay. PAMAM or its conjugates were complexed with 1 μg plasmid at the N/P ratio of 10, and an amount of 1 μg free plasmid as a control, were run in 1% agarose gel (prepared in 1 M Tris-acetate-EDTA (TAE) buffer solution and stained simultaneously with Ethidium Bromide and Bromophenol Blue). Electrophoresis was performed at 100 constant volts for 30 min.

Characterization of PG-PAM and VHH-PG
Dynamic light scattering and zeta potential measurements. PAM/pG-CX-bF-PE, PG-PAM/pG-CX-bF-PE, and VHH-PG-PAM/pG-CX-bF-PE dendriplexes were prepared according to the method mentioned in Dendriplex formation section and diluted by 800 μL of Milli-Q water. Their size and surface charge were measured by Zetasizer Nano ZS (Malvern Instruments, Malvern, UK).
Atomic force microscopy. The shapes and particle sizes of VHH-PG-PAM/pG-CX-bF-PE were analyzed using atomic force microscopy. Dendriplexes were prepared using VHH-PG-PAMAM complexed with pG-CX-bF-PE at an N/P ratio of 10 and incubated for 15 min at RT. The dendriplex solution was then diluted (1:20) with MilliQ-water. After that, a volume of 5 µL of the solution was placed on a freshly cleaved untreated mica surface and allowed to stick for 5 min. Then, the excess of the sample was removed by careful absorption onto filter paper and the mica surface were further dried under a gentle stream of air at RT. Sample were examined with a JPK AFM (JPK Instruments Co., Germany) using contact mode, HYDRA6V-100N cantilever with pyramidal shape tip, force constant of 0.292 N/m, and resonance frequency of 66 kHz.
Transmission electron microscopy. Transmission electron microscopy (TEM) images were obtained from CEM 902A ZEISS (Jena, Germany) transmission electron microscope with an accelerating voltage of 80 kV to investigate the size and morphology of VHH-PG-PAM/pG-CX-bF-PE at an N/P ratio of 10. Mammosphere cultivation. Mammospheres were generated from 2 × 10 4 single cells seeded in 6-well tissue culture plates which were coated with 1.5% agarose and contained 2 mL DMEM/F12 (Gibco) without serum, supplemented with B27 (1:50, Invitrogen), 5 mg/mL insulin (Sigma), 20 ng/mL bFGF (R&D), and 20 ng/mL EGF (R&D). The cultures were incubated in a humidified atmosphere (5% CO 2 ) at 37 °C for 3 days without any disturbance. After 3 days, 400 µL of fresh media was added to each well (without removing the old media). In the next step, the mammospheres were imaged on day 7. Moreover, the sizes and number of the mammospheres were quantitated using ImageJ software (NIH, USA) and mammosphere formation efficiency (MFE) was calculated as the number of mammospheres per well (diameter > 50 μm) divided by the original number of single cells seeded per well × 100. Experiments were performed in duplicates.

Monolayer cell transfection.
For cell transfection, 6 × 10 4 cells/well were seeded and incubated until they reached ∼ 80% confluency. Immediately before the transfection process, the complete medium was removed, cells were washed with PBS and then fresh medium without serum and antibiotics was added to each well. Moreover, the dendriplexes were prepared freshly and prior to use and then 100 μL of the dendriplex solution at a final transfection DNA concentration of 2 µg/mL −1 was applied to the cultured cells. After 6 h of incubation, the transfection solution was removed, the cells were washed with PBS, and then cell-specific complete medium was added and the culture was further incubated for the time period required for each experiment.
Hanging drop culture. MCF-10A and MDA-MB-231/HER + cell lines were cultured as spheroids in their appropriate media by the hanging drop method. In detail, the cells were dissociated from the monolayer cell culture, counted, and resuspended in complete medium at a concentration of 3 × 10 4 cells/mL. 13 µL aliquots of the suspension (containing 1000 cells each) were deposited on the underside of a 10 cm petri dish lid. The lid was then inverted over the dish filled with 10 mL of PBS (as a method to keep the cells hydrated). The dish was maintained at 37 °C in a humidified incubator (5% CO 2 ) for 7 days. The growth media were exchanged every other day by up-righting the lid, taking 10 µL media from each drop, and adding 14 µL fresh media into it. The cells were imaged daily and the formed spheroids' size (average of the major and minor axis length) was measured using the NIH ImageJ software.

Spheroid transfection.
On the seventh day following spheroid formation, the complete medium was replaced with PBS and subsequently with fresh medium without serum and antibiotics using the sequential pipetting method. The dendriplexes were prepared freshly prior to use and were carefully introduced to each of the spheroids by taking 14 µL media from and adding 14 µL of dendriplex solution to them. After 6 h of incubation, the medium was replaced with the complete medium using the sequential pipetting method. The hanging drop culture was further incubated for the time period required for each experiment. Then, the transfected spheroids were collected and dissociated to single cells by pipetting through a 200 µL pipet tip. In the next step, the obtained single cells were used for cellular uptake, PE38 mRNA expression, or cytotoxicity assays as described in subsequent Methods sections. For each assay, the experiments were performed in triplicate and 20 spheroids were analyzed per condition. PE38 mRNA expression analysis. Sixteen hours after the transfection process with different dendriplexes, the cells were collected, the total RNA was isolated with the High Pure RNA Isolation kit (Roche), and subsequently reverse transcribed using M-MuLV reverse transcriptase and oligo-dT (Fermentas www.nature.com/scientificreports/ β-actin. Also, the primers for amplifying PE38 and β-actin were as follows: PE38 for: 5ʹ AGG ACC TCG ACG CGA TCT G, PE38 Rev: 5ʹ TCA GGC TGG TGC GGT AGA AG, β-actin for: 5ʹ TCC CTG GAG AAG AGC TACG, and β-actin Rev: 5ʹ GTA GTT TCG TGG ATG CCA CA. Moreover, the relative PE38 mRNA expression in different samples were calculated using Pfaffl method and the experiments were repeated three times. Cytotoxicity study. To evaluate PE38 cytotoxicity, the cell viability of the transfected cells vs. non-transfected cells was determined by MTT assay. Briefly, the cells were seeded and transfected as described for monolayer or spheroid culture. 6 h after transfection with different dendriplexes, the transfection solution was removed and then the cells were washed with PBS and later complete medium was added to each well. After 48 h of incubation, the complete medium was replaced by fresh medium without serum and antibiotics and then 1 mg/mL MTT reagent (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; Sigma) was added to each well according to the manufacturer's protocols. The assay was repeated three times.
Inhibition of mammosphere formation. We  Statistical analyses. The statistical analyses were performed using GraphPad PRISM v.6.01 (GraphPad Software Inc, CA, USA). All data subjected to statistical analysis were obtained from at least three parallel experiments. Unpaired Student's t-test was used to determine significant differences between each group. Two-way ANOVA with Tukey's multiple comparison post hoc test was used to determine significant differences between different groups. A p-value ≤ 0.05 was considered statistically significant.

Characterization of PG-PAM and VHH-PG-PAM conjugates. Measuring primary amines of PAMAM
and PEGylated PAMAM. TNBSA assay confirmed that PEGylation of dendrimers with molar ratios of 10 yielded conjugates containing an average of 19 PEG chains per PAMAM molecule.
Characterization of PG-PAM by FTIR. The formation of the PG-PAM conjugates was further confirmed by FTIR spectra. In FTIR spectra (Fig. 1a), the peak at 3300 cm -1 due to -NH-groups of PAMAM stretching vibration is evident. The characteristic peaks of methyl C=O stretching of carbonyl groups of PAMAM internal amides or PG-PAM amide bonds is found at 1650 cm -1 . Besides, the peak at 1110 cm -1 due to stretching vibration of -CH2-O-CH2-etheric bonds of PEG molecules, indicating that PEG chains are successfully attached to PAMAM dendrimers.
Dendriplex characterization. Gel retardation assay. Agarose electrophoretic mobility retardation assay was performed using PAM/pG-CX-bF-PE, PG-PAM/pG-CX-bF-PE and VHH-PG-PAM/pG-CX-bF-PE, at N/P ratio of 10. The purified pG-CX-bF-PE plasmid moved toward anode (Fig. 2a, lane 1). In contrast, Fig. 2a, lanes 2 shows that, not only PAMAM reversed the plasmid electrophoresis pattern, but also the negatively charged Bromophenol Blue dye were also electrophoresed to the opposite direction, cathode. This result suggests that the overall charge of PAM/pG-CX-bF-PE were comparatively positive. Figure 2a, lanes 3 and 4 show that, PG-PAM and VHH-PG-PAM interacted sufficiently with the pG-CX-bF-PE and neutralized its negative charge so the plasmid were retarded in the well( Full-length gel image is provided in supplementary Fig. S1).
Dynamic light scattering and zeta potential measurements. The size and zeta potential of the dendriplexes at N/P ratio of 10 were measured by zetasizer Nano ZS. According to results presented in Table 2, all the dendriplexes showed the appropriate size and charge density. Low polydispersity indexes indicate the formation of homogeneous and aggregate free dendriplexes. It seems that by conjugating PEG and VHH to the PAMAM dendrimers, the size of dendriplexes increased, while the zeta potential was decreased. These findings imply that both PEG and VHH were successfully attached to the dendrimers and shielded the positive charges of the free amines on the PAMAM surface. However, the size of dendriplexes ranged between 105.6 and 154.7 nm and the zeta potential ranged between 26.8 and 5.2 mV which are considered to be appropriate for endocytosis.
Atomic force microscopy. The morphology of VHH-PG-PAM/pG-CX-bF-PE at an N/P ratio of 10, were observed by AFM (Fig. 2b)  www.nature.com/scientificreports/ Transmission electron microscopy. Transmission electron microscopy was performed to further investigate the size and morphology of VHH-PG-PAM/pG-CX-bF-PE dendriplexes. TEM result was in accordance with the AFM data, showing spherical structures with an average particle size of 118 ± 7 nm (Fig. 2c). However, the particle size visualized by TEM and AFM were smaller than those determined by DLS. The most probable explanation would be that TEM and AFM determine the dry particle size, whereas DLS reflects the hydrodynamic size.
There are discernible differences in BCSC pools between MCF-10A and MDA-MB-231/ HER2 + . Mammosphere assay was used to evaluate the stem property of the cell lines. According to Fig. 4a, both   Mammosphere formation assay identifies BCSCs based on their resistance to anoikis in non-adherent/serumfree conditions. This technique is considered to be a reliable technique since conventional BCSC specific markers occasionally vary infinitely from one breast cancer cell to another 24 The data of our previous work also agreed with this concept 22 , thus in the present study we based our work on the ability of BCSCs to form mammospheres." Morphology of spheroids in hanging drops. We utilized the hanging drop technique to generate spheroids with uniform shape and size. MCF-10A grown in hanging drop culture formed thoroughly loose aggregates exposing free areas between the cells. While MDA-MB-231/HER + generated tightly compact well-rounded spheroids within 24 h (Fig. 4b). Based on these results, MCF-10A aggregates were excluded from further investigation and MDA-MB-231/HER + spheroids were selected for subsequent studies. MDA-MB-231/HER + spheroids were maintained in hanging drop culture for 7 days and imaged daily. As illustrated in Fig. 5, spheroids of different drops were consistent in both size and shape. On the first day spheroids measured approximately 203 μm. From first to sixth day the spheroids diameter increased slightly. On the sixth day their average size reached to 484 μm. However, by day 7 it decreased to 439 μm, indicating spheroids started to become denser. Several researchers suggest that spheroids of diameter ranging between 300 and 500 μm accurately mimic real tumors [25][26][27] . As it can be seen in Fig. 5, with increasing length of hanging drop culture, spheroids became a little darker in color, which is an indicative of gradual compactness 28 . On the seventh day following hanging drop culture, spheroids were transfected with the previously mentioned dendriplexes. It is noticeable that, spheroids were washed and transfected while they were in the drops, which to the best of our knowledge has not been previously   www.nature.com/scientificreports/ for the lower uptake rate in spheroids might be the penetration resistance inherent in their structure. Spheroid is a three-dimensional (3D) model that closely resembles small avascular tumors and micrometastases in that they contain a proliferative outer shell, a relatively large zone of hypoxic quiescent cells and a relatively small necrotic area at the center. This 3D structure exhibit cell-to-cell and cell-to-matrix interactions as well as nutrient, pH and oxygen gradients. These characteristics make them proportional to penetration so that they are not efficiently transfected. This kind of resistance might not be observed when cells are cultured as monolayer 31,32 . Error bars represent the standard deviation. The statistical significance of the groups was calculated using t-test, two-way ANOVA and Tukey's test for multiple comparisons. A p-value ≤ 0.05 was considered statistically significant (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). www.nature.com/scientificreports/ tion process and as shown in Fig. 8b, the viability was about 80% in the three cell lines. In contrast, following treatment with PE38 encoding vectors, extensive cell death was detected. This result reveals the clear cytotoxic potential of this toxin. However, having transfected with PAM/pG-CM-PE (non-targeted), all the three cell lines exhibited about 13% viability. In the case of VHH-PG-PAM/pG-CM-PE(targeted on cell surface level), 28% and 24% cell viability for MCF-10A and MDA-MB-231 were detected, respectively; but it was 7% for MDA-MB-231/HER2 + . These results suggest that VHH-PG-PAM/pG-CM-PE exerts potent cytotoxicity in HER2-overexpressing cells due to the specific cell surface targeting. VHH-PG-PAM/pG-CX-PE exhibited higher viability rate, as compared to VHH-PG-PAM/pG-CM-PE in all the cell lines, showing that CXCR1 promoter is weaker than the commonly used strong CMV promoter. Cell viability of MCF-10A and MDA-MB-231 transfected with the VHH-PG-PAM/pG-CX-PE (targeted on cell surface and transcriptional levels) were 41% and 29%, respectively, while in the MDA-MB-231/HER2 + it was 11%. However, VHH-PG-PAM/pG-CX-PE demonstrated more viability variation between the cell lines compared to the non-specific CMV promoter, and the lowest viability belonged to the MDA-MB-231/HER2 + , indicating CXCR1 promoter specificity for BSCSs and HER2 expressing breast cancer cells. These results also demonstrate that, combination of the cell surface and transcriptional targeting, increased the stringency of the treatment. When transfected with VHH-PG-PAM/pG-CX-bF-PE (targeted on cell surface, transcriptional and translational levels), MCF-10A and MDA-MB-231 exhibited 61% and 38% viability, respectively. But MDA-MB-231/HER2 + showed 15% recovery of the cells. When comparing the effect of treatments with dendriplexes targeted on "cell surface", "cell surface and transcriptional" and "cell surface, transcriptional and translational" levels, as it can be seen in Fig. 8b, by the addition of each targeting element to the others, viability variation between different cell lines were increased. This result, indicates that the stringency of the treatment increases when these targeting elements are combined together. It is noticeable that, viability of VHH-PG-PAM/pG-CX-bF-PE treated groups had a tendency to increase as compared to that of VHH-PG-PAM/pG-CX-PE treated groups, probably because the highly structured, GC rich 5' UTR of the bFGF hinders efficient translation. Translation of mRNAs with highly structured bFGF 5′UTR is particularly dependent on the unwinding activity of eIF4E. Higher levels of eIF4E in breast cancer cell lines, relative to non-malignant MCF10A cells have been reported previously [33][34][35] . Thus, it can be concluded that overexpression of eIF4E in tumorigenic MDA-MB-231 and MDA-MB-231/HER2 + might be responsible for efficient toxin translation and subsequent massive cell death. Nonetheless, low levels of eIF4E in normal MCF-10A repressed toxin translation, resulting in elevated viability. Our study findings are consistent with other reports showing the efficient targeting efficacy of bFGF 5′UTR toward cancerous cells 9,33,36 .
Comparison of MDA-MB-231/HER2 + grown as monolayer vs spheroid revealed that MDA-MB-231/HER2 + spheroid followed the same viability pattern as its monolayer, upon all the treatments, except the viability were significantly higher upon all the treatments (p < 0.01). These results further validated those obtained by cellular uptake and real-time PCR experiments. Indeed, previous studies have demonstrated a differential response to compounds when cells grown as spheroid compared to monolayer 37,38 . A good illustration would be the study of Carver et al. 39 . They demonstrated that the delivery of oligonucleotides with Lipofectamine lipoplex and PEI polyplex was significantly attenuated in spheroid models compared to monolayer cultures. Further, their results showed that only cells located at the periphery of the spheroid received the oligonucleotides.
Our multi-targeted nanosystem efficiently inhibited mammosphere formation of MDA-MB-231 and MDA-MB-231/HER2 + cells, while it was unable to prevent mammosphere formation of MCF-10A. To confirm the data obtained with MTT, we examined the effect of different dendriplexes, on the formation of mammospheres. The outcome revealed that, when the cells were transfected at the seeding step, the addition of PAM/pG-CM-PE resulted in the complete proliferation inhibition of all the three cell lines. Also, all of the dendriplexes strongly inhibited mammosphere formation of MDA-MB-231 and MDA-MB-231/HER2 + . However, when they were transfected after five days growth in mammosphere culture conditions, the existing mammospheres were completely dissociated by Day 7. These results can be explained by the high cytotoxicity potential of PE38 protein. It is believed that expression of only one PE38 molecule efficiently kills the host cell l40 Interestingly, VHH-PG-PAM/pG-CM-PE, VHH-PG-PAM/pG-CX-PE and VHH-PG-PAM/pG-CX-bF-PE were unable to prevent the mammosphere formation from MCF-10A, which were transfected either at seeding or after five days of growth in mammosphere culture conditions. However, the above mentioned dendriplexes led to a reduction in MCF-10A MFE % by Day 7 (from 0.5 ± 0.11% for non-treated to 0.2 ± 0.62%, 0.3 ± 0.17%, and 0.3 ± 0.55% in case of seeding stage transfection; and to 0.3 ± 0.05%, 0.3 ± 0.62%, and 0.4 ± 0.15% in case of transfecting the existing mammospheres, respectively).

Conclusion
In summary, we successfully developed a well-designed multi-targeted nanosystem using anti-HER2 VHH functionalized PAMAM, CXCR1 promoter, PE38 toxin and bFGF 5′UTR. This nanosystem was selectively internalized, specifically transcribed, efficiently translated and caused a selective cytotoxicity in HER2-positive BCSCs. Our data suggest that this novel multi-targeted nanosystem would be a potent strategy for selective cancer killer gene therapy. We further demonstrated that the efficacy of our targeted gene therapy was lower in spheroid models compared to monolayer cultures. Since tumor spheroid models are a more similar candidate for bulky tumor lesions and their microenvironment (in patients and/or preclinical anima models) in comparison with conventional monolayer cell cultures, it might be reasonable to anticipate lower antitumor efficacy for our nanosystem in preclinical animal models of human breast cancer in comparison with that achieved from in vitro experiments.
However, in vivo validity of the proposed idea remains to be assessed.

Data availability
All data generated or analyzed during this study are included in this published article.