Prototyping Trastuzumab Docetaxel Immunoliposomes with a New FCM-Based Method to Quantify Optimal Antibody Density on Nanoparticles

Developing targeted nanoparticles is a rising strategy to improve drug delivery in oncology. Antibodies are the most commonly used targeting agents. However, determination of their optimal number at the surface remains a challenging issue, mainly due to the difficulties in measuring precisely surface coating levels when prototyping nanoparticles. We developed an original quantitative assay to measure the exact number of coated antibodies per nanoparticle. Using flow cytometry optimized for submicron particle analysis and beads covered with known amounts of human IgG-kappa mimicking various amounts of antibodies, this new method was tested as part of the prototyping of docetaxel liposomes coated with trastuzumab against Her2+ breast cancer. This quantification method allowed to discriminate various batches of immunoliposomes depending on their trastuzumab density on nanoparticle surface (i.e., 330 (Immunoliposome-1), 480 (Immunoliposome-2) and 690 (Immunoliposome-3), p = 0.004, One-way ANOVA). Here we showed that optimal number of grafted antibodies on nanoparticles should be finely tuned and highest density of targeting agent is not necessarily associated with highest efficacy. Overall, this new method should help to better prototype third generation nanoparticles.


Results
FCM method: development of a quantification assay for submicron particles. As described in detail in the Supplemental data section, the 10 µm QIFIkit calibrator, was scaled-down to generate a series of prototype 1 µm-sized "µ-QIFIkit" calibrator beads covering an approximate range of ~20 to ~20,000 mouse IgG/ bead. The correlation coefficient r² of the prototype was equal to 0.9869 (Fig. 1A).

Detection of immunoliposomes and quantification of coated trastuzumab on fcM.
As explained in the method section, the triggering parameter for FCM analysis of liposomes was based on DiD fluorescence. The positivity threshold was defined using unstained immunoliposomes ( Fig. 2A). Except from electronic and fluidic background, no DiD+ elements were detected. Using tagged liposomes, the SSC/DiD dot plot allowed the definition of the DiD+ (immuno)liposome gate. DiD bright elements were also excluded from the gate due to the possible generation of doublets or multiplets of immunoliposomes. DiD immunoliposomes were successfully detected (Fig. 2B). DiD+ Immunoliposome numeration provided using FCM represented 65% (see Supplemental data section for qNano: measurement of liposome concentration) of total Immunoliposomes counted using TRPS technology (qNano). Moreover, considering the PE quantification channel, the positivity threshold was adjusted using DiD+ uncoated liposomes incubated with the same anti-Human IgG-k PE MAb IgHk calibrated beads using µ-QIFIkit as reference. IgHk beads have been first saturated with unlabeled L1C1 Mab, washed and fluorescently stained with PE-labeled anti-mouse IgG sencondary reagent in parallel with the µ-QIFIkit beads used as reference. (C) Superposition of histograms from the prototyped "IgHk calibrator beads" as seen on the PE channel of CytoFLEX S under yellow laser excitation. A limited series of 5 beads regularly distributed over the measuring range were chosen from the larger series of all (n = 13) calibrated beads to be later used in direct IF experiments with L1C1-PE as "IgHk calibrator beads". (D) Calibration line of the prototyped "IgHk calibrator beads" generated for this study. immunoliposome characterization. No significant difference in size, PDI and entrapment efficiency was observed between immunoliposome batches. According to DLS analysis, immunoliposome population was unimodal in size (i.e., PDI = 0.1 ± 0.01) with a mean diameter of 140 ± 3.4 nm. Docetaxel entrapment efficiency was >90%.
Stability studies. No significant difference in stability was observed between immunoliposome batches. As previously published 25 , after 45 days, immunoliposome was steady, PDI increase was not significant (i.e., from 0.1 to 0.153 in 45 days) and mean docetaxel leakage was of 17 ± 13% per week. Evolution in time of coated trastuzumab is illustrated in Fig. 4. After an initial loss of maximum 20% in the first 2 weeks (i.e., 19% and 12%, p > 0.05, t-student) for Immunoliposome-1 and Immunoliposome-2, respectively, trastuzumab density remained stable for 45 days (p < 0.05, t-student). On the contrary, Immunoliposome-3 trastuzumab density presented strong variations with an apparent increase of 87% at day 45 (p = 0.011, One-way ANOVA).

Discussion
In the present study for the first time we used QFCM approach for characterizing third generation of drug-loaded liposomes, a.k.a. immunoliposomes. Here, we worked on prototyping docetaxel-trastuzumab immunoliposome which previously demonstrated in vitro 7 and in vivo 26 anti-tumor efficacy. However, two major hurdles had to be overcome: first, QFCM had never been applied yet on submicrometer-sized particles and second, FCM analysis of nanoparticles was still a major challenge 27 . Schematically, two main QFCM strategies are still in use for absolute quantification of molecules on individual particles, using either direct (e.g., Quantibrite, BD Biosciences, San Jose, CA, USA) or indirect immuno-fluorescence (i.e., QIFIkit, Dako, Glostrup, D and CellQuant Calibrator, BioCytex, Marseille, France) 13,16,17 . A third new, more specific and possibly more robust approach, is to create calibrator beads directly coated with the antigen of interest in various known amounts 28 . Such calibrator beads can be tailored to the appropriate size and antigen density range to mimic the particles of interest in the effective staining protocol. This was our present strategy, with kappa light chains of human IgG (so-called "IgHk") as the target antigen mimicking trastuzumab molecules coated on immunoliposomes. Using transitional calibration systems, we progressively reduced beads size and antigen density from 10 µm to 1 µm and from a few dozens to a few thousands' molecules/ bead, respectively, leading us to select a set of 5 calibrator beads, so-called "IgHk calibrator beads", exhibiting from 36 to 14900 IgHk molecules/bead.
However, a few special technological tricks still had to be developed to achieve our final goals: first, fluorescence tagging to sort-out liposomes from non-specific events (i.e., both electronical and optical background) and second, the best sensitivity as possible in terms of fluorescent detection.
Indeed, because of their low refractive index and detection as compared to polystyrene beads of similar size range, liposomes generate very faint scatter signals. To overcome this major limitation, we followed other specialists' suggestion to use a fluorescence-based rather than a scatter-based triggering parameter [29][30][31] . Thus, immunoliposomes were labeled during synthesis using a lipophilic fluorescent dye (i.e. DiD) with a different color than the reporting immunological reagents. As a result, immunoliposomes were defined as dye positive, green fluorescent particles on which human IgG should be quantified.
Then, to benefit from the highest sensitivity and allow a clear resolution, using last generation cytometer Cytoflex S (Beckman-Coulter, Villepinte France) we excited phycoerythrin (PE), one of the brightest fluorochrome available, at its optimal absorption peak (i.e. 561 nm instead of 488 nm for several benchtop instruments).
Together these two tricks allowed us to detect 65% of immunoliposome population and to consider this FCM analysis as representative of all immunoliposomes.
Thus, this QFCM method provides the effective number of IgG per particle on an absolute, rather than relative, basis and allows independent measurements along time. This may be quite useful for initial and long-term quality control of the MAb coating, as illustrated here in our 6 weeks long stability study. Associated with other absolute size measurement techniques such as DLS, it may also allow calculation of true surface density, expressed as molecules per surface unit, which may become useful for comparison of immunoliposomes with different sizes when several prototypes have to be tested.
To our knowledge, this is the first description of a FCM-based quantitative analysis of antigens/receptors on submicron particles and can be easily adapted to any other target molecule that could be biotinylated. Noteworthy, QFCM can not only provide a median value for the number of molecules per particle but can also inform about the dispersion of the immunoliposome contains, as it was illustrated in Fig. 3. Thus, both median density and homogeneity of coating comes at hands, even on very small particles such as liposomes.
Finally, this new method was used to evaluate the influence of median trastuzumab density on the cytotoxic effect of immunoliposomes. To test our hypothesis, three different batches of immunoliposomes were synthesized and exhibited a 330-690 range of coated trastuzumab per liposome, respectively. As previously published 7 , Immunoliposomes displayed a steady size but mild docetaxel leakage over time. Similarly, for Immunoliposome-1 and Immunoliposome-2, coated trastuzumab slightly decreased weekly. Oppositely, coated trastuzumab on Immunliposome-3 seemed to increase over time (i.e., +190% after 45 days) raising numerous questions. It is likely that this increase in coating was related to immunoliposome DiD staining. Indeed, DiD staining is a key point in our quantification technique since only fluorescent immunoliposomes can be accurately analyzed on Cytoflex S (Fig. 2). However, as docetaxel, DiD is a lipophilic agent and is expected to leak from liposome bilayers over time. Thus, the smaller liposomes which are considered less stable 32,33 , loose rapidly their fluorescence and become invisible within a few weeks to CMF analysis. Since smaller liposomes are expected to exhibit lower number of coated trastuzumab, this loss in DiD staining could be responsible for an analysis of larger liposomes only, after a few weeks, thus resulting in increased coated trastuzumab density. Although it appeared as a limitation of our new quantification assay when applied to stored liposomes, it could be overcome using extemporaneously-synthesized liposomes or steadier nanoparticles such as inorganic or polymeric ones.
The three batches of immunoliposomes were then tested in vitro on bidimensional (2D) and three-dimensional (3D) models. We found that similarly to standard MTT assays 26,34 , apoptosis studies using 2D models were not suitable to discriminate the efficacy of the various immunoliposomes as compared with free docetaxel or liposomal docetaxel. Conversely when using 3D spheroids, we found that all immunoliposomes performed better than free docetaxel and its liposomal form. Although not significant, differences between Immunoliposome batches were observed because higher antiproliferative efficacy was achieved with Immunoliposome-2 and Immunoliposome-3 as compared with Immunoliposome-1. Reducing the coating probably led to a loss in cellular  www.nature.com/scientificreports www.nature.com/scientificreports/ uptake, thus impairing trastuzumab and docetaxel cytotoxic effects. Conversely, increasing, trastuzumab coating proved to be at best equally effective but not better, and was then nothing but a waste in trastuzumab during synthesis. Indeed, although 45% more molecules of trastuzumab were coated on Immunoliposome-3, it resulted in similar cellular uptake and antiproliferative efficacy, possibly because of steric hindrance and binding-site barrier issues 35 . Thus, Immunoliposome-2, corresponding to 480 ± 110 molecules of coated trastuzumab per liposome, could be the optimal number of trastuzumab to be coated to ensure a maximal efficacy in this breast cancer model. Interestingly, preliminary in vivo efficacy studies have already shown its benefit over free docetaxel + free trastuzumab and antibody drug conjugate T-DM1 26 .
Although, the present technology-oriented study needs more investigations, for instance about cell Her2 expression influence 36,37 , our data already suggest that maximal density of the targeting agent is not a major requirement for achieving maximal efficacy, thus highlighting how bio-physical parameters must be finely tuned and how critical is the need for an accurate quantitation of the targeting agent on NP's surface. Confirmation on more sophisticated in vitro 38,39 and in vivo models may also be of major interest to better illustrate the exquisite specificity of immunoliposome-based treatments.
outlook. For the first time, we have developed a sensitive FCM-based quantitative method to measure the number of coated antibodies on nanoparticles. This new quantification assay can help to characterize targeting nanoparticles as part of early prototyping steps. In addition, it can also be used as a quality control tool to control batch-to-batch variations when the optimal NP has been selected. Applied to docetaxel -trastuzumab immunoliposomes in a model of Her2+ breast cancer cell line, QFCM was successfully used to precise the optimal number of trastuzumab molecules to coat to achieve a maximal efficacy. Beyond this first application, we believe that this original method could be useful to researchers looking for a rapid, simple and precise method to quantify any monoclonal antibodies used as targeting agents when developing smart nanoparticles.  France). Docetaxel was purchased from VWR (Fontenay sous bois, France) and DiD fluorescent tag from Thermo Fisher (Illkirch, France). Trastuzumab (Herceptin) was kindly given by Genentech (South San Francisco, CA, USA). Anti-human IgG PE (clone L1C1) and a prototyped human IgG kappa light chains quantification kit for submicron particles (e.g., immunoliposomes) were provided by Biocytex (Marseille, France) as described below. All other reagents were of analytical grade.

QFCM method: development of a quantification assay for human IgG on submicron particles.
For QFCM analysis of submicron particles both the size and measuring range of the official calibration systems (i.e. 10 µm diameter and 3,300 to 783,000 IgG/bead for QIFIkit, DAKO, Glostrup, Denmark) are far too high for the specific needs. Thus, we had to reduce both parameters towards the more acceptable diameter of 1 µm and range of a few dozens to a few thousands' molecules/bead. This was done in successive steps as summarized in Table 2 and detailed in the Supplemental data section 13,14,17,19,28,[40][41][42][43][44][45][46] depicting the generation of the prototype "µ-QIFIkit" calibrator.
Since liposomes (submicron particles in general) are inappropriate to afford washing steps, a direct IF no-wash protocol was then adapted for their staining. L1C1 anti-IgHk Mab was directly conjugated to PE and purified according to BioCytex internal procedures. Titration was done in the final assay with high concentration of the highest level IgHk-expressing beads to ensure saturating conditions without excess of PE-Mab. With this Mab of rather high affinity, the saturating PE-conjugate initial concentration during staining was not more than 5 µg/mL and thus less than 2 µg/mL in the final diluted suspension during QFCM analysis. immunoliposome preparation. As previously described, liposomes were synthesized using the Thin-Film method 7 . PC, Chol, PG, docetaxel and Mal-PEG were mixed in a 50:19:15:1.7:1 molar ratio. Briefly, lipids were dissolved in methanol. Lipid solution was further mixed with DiD as lipophilic membrane insertable fluorescent reporter when required for liposome-oriented FCM analysis. Methanol was then removed by rotary evaporation (Laborota 4003, Heidolph Instruments, Schwabach, Germany) at 38 °C. After 30 minutes a thin lipid film was obtained. To remove the remaining solvent, this lipid film was dried under a stream of nitrogen for two hours at room temperature. The film was then hydrated with a 5% vol/vol glucose solution and large liposomes were obtained. Reduction and homogenization in size was thus achieved by two cycles of extrusion through 100 nm and 80 nm polycarbonate pore membranes (Nucleopore, Whatman) using a LipoFast LF-50 extruder. Trastuzumab was then coated using a maleimide linker 8,10 requiring a preliminary step of trastuzumab thiolation. Trastuzumab was first dissolved in a 0.1 M sodium phosphate buffer (PBS) pH 8.0 containing 5 mM EDTA and mixed under constant shaking, for two hours at room temperature with a Traut's reagent solution at 1:10 molar ratio (Traut's:trastuzumab). Thiolated trastuzumab was then directly mixed with the pegylated liposomes at 1:508 (Immunoliposome-1), 1:127 (Immunoliposome-2) and 1:16 (Immunoliposome-3) trastuzumab:Mal-PEG molar ratio. The mixture was kept under constant shaking at 4 °C overnight. Unbound trastuzumab and free docetaxel were removed using 6 000 g centrifugation on MWCO 300KDa Vivaspins (VWR,Fontenay sous bois, France) followed by size exclusion chromatography on qEV columns (IZON Science, Lyon, France).
Size and polydispersity study. Size and polydispersity index (PDI) were measured by Dynamic light scattering (DLS). Liposomes and immunoliposomes were diluted in a PBS solution and then analyzed by a Zeta sizer Nano S (Malvern instruments, UK). Liposomal preparations were considered unimodal for a PDI < 0.2 47 . Docetaxel entrapment efficiency. Docetaxel concentrations were measured using a validated HPLC-UV method 48 after liquid/liquid extraction using a C18 column (25 cm × 4.6 mm, 5 µm). The mobile phase was composed of 53% of ammonium acetate buffer (35 nM, pH 5) and 47% of acetonitrile. Samples were eluted at a constant flow rate of 1.8 ml/min with UV detection (227 nm). Data were acquired and analyzed using Chemstation software (Agilent, France). Docetaxel and paclitaxel typical retention times were respectively 11 minutes and 13.5 minutes. Docetaxel entrapment efficiency was calculated using the following formula: flow cytometric detection and quantitative analysis of trastuzumab-coated immunoliposomes.
QFCM analysis was performed on highly sensitive flow cytometer, CytoFLEX S (Beckman-Coulter, Villepinte, France) using PE-conjugated polyclonal or monoclonal antibodies and 561 nm excitation laser.
To avoid unspecific events mixing with liposomes in the FCM analysis, fluorescent triggering was involved, using a generic red fluorescent signal issued from the red laser excitation point, totally independent from the PE-MAb staining which provides orange fluorescence from the yellow laser. To apply such a generic staining, liposomes were tagged with DiD fluorescent lipophilic molecules, which encapsulated in the phospholipid bilayer of all liposomes. Consequently, the triggering parameter for their analysis on the CytoFLEX S was the red fluorescence of DiD taken from the red laser.
To determine the percentage of immunoliposome detected with flow cytometry, an absolute quantification of immunoliposomes was performed using TRPS technology (qNano, IZON, Lyon France).
Stability studies. Stability studies were performed in PBS at 4 °C, protected from light. Immunoliposome size, PDI, docetaxel leakage and persistence of coated trastuzumab were evaluated weekly for a month, then bimonthly for up to 45 days, using differential centrifugation to separate NPs from soluble material. cell lines. In vitro experiments were carried on HER2+ human breast cancer cell line MDA-MB-453. Cells were purchased from the American Type Culture Cell (Molsheim, France) cultured in RPMI (Thermo Fisher, Illkirch, France) supplemented with 10% FBS, 1% penicillin and 0.16% kanamycin and grown in a humidified CO2 incubator at 37 °C 26 . Cells were regularly checked for cell viability, morphology and doubling time. Cells were stably transfected with dTomato lentivirus developed and kindly provided by Pr Jacques Robert (Institut Bergonié, Bordeaux, France) and selected with blasticidin to allow fluorescence imaging 26 . (2020) 10:4147 | https://doi.org/10.1038/s41598-020-60856-z www.nature.com/scientificreports www.nature.com/scientificreports/ In vitro efficacy studies: 2D model. To evaluate cell apoptosis, we used flow cytometry with an AnnexinV/ PI kit (Sigma Aldrich,St Quentin Fallavier, France). MDA-MB-453 were seeded at a density of 7 × 10 5 cells per well in 6-well plates. After overnight attachment, cells were exposed to free docetaxel + free trastuzumab (i.e., free drugs), docetaxel liposomes + free trastuzumab (i.e., Liposome), and immunoliposomes for 12, 36 and 72 hours. For all conditions, docetaxel concentration was 2 µM and trastuzumab concentrations were 3 nM, 4 nM and 6 nM for batches 1, 2 and 3, respectively, whether trastuzumab was free or coated. Cells were then stained with 10 µl of Annexin V FITC that targets Phosphatidylserine (PS) molecules present on the outside layer of apoptotic cell membranes and 10 µl of Propidium Iodide (PI), a fluorescent DNA intercalant molecule that stains nuclear DNA when cells undergo late apoptosis/necrosis. Cells were incubated for 20 minutes à 4 °C and then washed in 2 ml of Binding buffer (i.e., Ca2+ stabilizing buffer). Cells were then centrifuged 10 min, 900 g at 4 °C and cell pellet was resuspended in 500 µl of binding buffer prior to FCM acquisition on Gallios FCMr (Beckman Coulter, Villepinte, France). Live cells were considered as AnnexinV neg/PI neg events. Cells in early apoptosis were considered as AnnexinV+/PI−. Cells in late apoptosis were considered as AnnexinV+/PI+ and finally, necrotic cells were considered as AnnexinV−/PI+. In vitro efficacy studies: 3D model. MDA-MB-453 were seeded with 20% methylcellulose solution on U-bottom 96-well plate for 24 hours before the experiment begins. Cell density was 4000 cells/well. To evaluate treatment antiproliferative efficacy on spheroids, the following conditions were tested: free docetaxel + free trastuzumab (i.e., free drugs), liposomes + free trastuzumab (Liposome), Immunoliposome-1, Immunoliposome-2 and Immunoliposome-3. Treatments were incubated 3 days after seeding. Cells were exposed continuously to treatments for a week, then treatment was repeated, and drugs incubated until day 14. For all conditions concentration of docetaxel was 8 nM. Cell viability was determined using CellTiter-Glo (Promega, Charbonnières-les-Bains), following manufacter's guidelines and luminescent spectrophotometric reading on PHERAstar FSX (BMG Labtech, Heathfiel, UK). Spheroids were also monitored daily using a fluorescence microscope (Nikon, Eclipse TS100), coupled to digital camera.

Statistical analysis.
Similarly to all our studies 7,26 , in vitro experiments were performed at least in triplicate and data were represented as mean ± standard deviation (SD) or ± standard error of the mean (SEM). Statistical analyses were performed on SigmaStat (San Jose, USA). Differences between treatments were analyzed by One-Way Anova with Multiple Comparison testing or Student's t-test according to data distribution and sample size.