Ultrasmall targeted nanoparticles with engineered antibody fragments for imaging detection of HER2-overexpressing breast cancer

Controlling the biodistribution of nanoparticles upon intravenous injection is the key to achieving target specificity. One of the impediments in nanoparticle-based tumor targeting is the inability to limit the trafficking of nanoparticles to liver and other organs leading to smaller accumulated amounts in tumor tissues, particularly via passive targeting. Here we overcome both these challenges by designing nanoparticles that combine the specificity of antibodies with favorable particle biodistribution profiles, while not exceeding the threshold for renal filtration as a combined vehicle. To that end, ultrasmall silica nanoparticles are functionalized with anti-human epidermal growth factor receptor 2 (HER2) single-chain variable fragments to exhibit high tumor-targeting efficiency and efficient renal clearance. This ultrasmall targeted nanotheranostics/nanotherapeutic platform has broad utility, both for imaging a variety of tumor tissues by suitably adopting the targeting fragment and as a potentially useful drug delivery vehicle.

activity of 5.28-13.43 mCi/μg (470-1195 Ci/mmol) of zirconium. 4 Activity measurements were performed using a CRC-15R Dose Calibrator (Capintec). For the quantification of activities, experimental samples were counted on an Automatic Wizard 2 γ-Counter (PerkinElmer). All in vivo experiments were performed according to protocols approved by the Memorial Sloan Kettering Institutional Animal Care and Use Committee. A purity of greater than 95% was confirmed using radio-TLC for all of the 89 Zrlabeled DFO-scFv-PEG-Cy5-C' dots.
In vivo radio-stability studies. For in vivo radio-stability, healthy mice were injected with ~200 μCi (~7.4 MBq) of 89 Zr-DFO-scFv-PEG-Cy5-C' dots. Whole blood was collected at 2, 24 and 48 h postinjection, and the plasma fraction was isolated from red blood cells by centrifugation at 8000 rpm for 10 min. The percentage of the intact 89 Zr-DFO-scFv-PEG-Cy5-C' dots was then measured by using Radio-TLC with the plates analyzed on a Bioscan AR-2000 radio-TLC plate reader using Winscan Radio-TLC software (Bioscan Inc., Washington, DC).
Dosimetry. Time-activity curves derived for each tissue were analytically integrated, accounting for radioactive decay, to yield the corresponding cumulative activity. Organ absorbed doses were then calculated by multiplying the cumulative activity by the 89 Zr equilibrium dose constant for nonpenetrating radiations (positrons), assuming complete local absorption of such radiations and ignoring the contribution of penetrating radiations (i.e., γ-rays). Mouse normal organ cumulated activities were converted to human normal organ cumulated activities by taking into account differences in total-body and organ masses between mice and humans (assuming 70-kg standard human). Calculated human normal-organ cumulated activities were entered into the OLINDA dosimetry program to compute standard human organ absorbed doses using formalism of the Medical Internal Dosimetry Committee of the Society of Nuclear Medicine. 5 This human dosimetry model is a "normal" (i.e., tumor-free) anatomic model. at 10X magnification. Whole tumor section images were obtained by acquiring multiple fields that were then stitched together using MicroSuite Biologic Suite (version 2.7, Olympus). Brightfield (H&E) and fluorescence (Cy5) images were acquired using the appropriate filter cube sets. Processing of fluorescence images was carried out in Adobe Photoshop (CS6), as described previously. 6 HER2 immunofluorescence staining. Frozen sections were air dried for 30 min at room temperature, followed by fixation in 3% paraformaldehyde for 15 min at room temperature in a humidified chamber.
Sections were washed twice with buffer (1X TBS/0.1% Tween-20) for 5 min. Endogenous peroxides were blocked using BLOXALL (SP-6000, Vector Labs, Burlingame, CA) for 10 min, and tissue sections were subsequently washed twice with buffer for 5 min, prior to the addition of the blocking solution (5% Normal Goat Serum/wash buffer). Blocking solution was incubated with sections for 1 hour at room temperature, followed by washing and the addition of HER2 (#2165S, Cell Signaling, Danvers, MA) primary antibody (1:400 dilution) in blocking solution. Sections were then incubated overnight at 4°C in a humidified chamber, followed by multiple washes and addition of FITC conjugated anti-rabbit secondary antibody (Cell Signaling, #4412) at a dilution of 1:300 in blocking solution. Tissue sections were incubated at room temperature for 30 min, and washed in triplicate for 5 min. Nuclear counterstaining was performed with DAPI (0.1 mg/ml). Sections were washed twice for 5 min, followed by coverslip application using Prolong Gold Antifade (ThermoFisher, #P10144) mounting media.
Images were captured using identical acquisition settings and subjected to synonymous postprocessing procedures across all specimens. Image acquisition was performed in an analogous manner to the methods described above.
Digital Autoradiography. 10 μm tumor tissue sections were exposed to a phosphor-imaging plate (Fujifilm BAS-MS2325, Fuji Photo Film, Japan) at -20°C. Following exposure, the phosphor-imaging plate was read at a resolution of 25 μm using a Typhoon 7000 IP (GE Life Sciences, Pittsburgh, PA) plate reader. Images were saved as TIFF and GEL formats for future analysis and processed using ImageJ software (https://imagej.nih.gov/ij/).

Section 2. Supplementary Discussion
2.1. Purification of anti-HER2 scFv fragments demonstrates high monomeric content. All affinity purified scFv fragments, except C-terminal scFv, showed a discrete band by SDS-PAGE and Coomassie staining (Fig. 1c). The scFv containing the nnAA at the C-terminus showed several higher molecular mass bands migrating at regular intervals. Resolution of this protein under reducing conditions eliminated these bands, suggesting formation of disulfide-linked multimers in this sample (data not shown). The doublet bands in the C-terminal sample represent 6xHis tagged and truncated versions of the scFv fragments that migrate at slightly different molar masses. Size exclusion chromatography (SEC) analysis of non-reduced products was performed to examine the monomeric state of the samples before the aggregate removal described above (Supplementary Fig. 1

Efficient PEG Conjugation and Product Analysis.
To confirm successful incorporation of N6-((2-azidoethoxy)carbonyl)-L-lysine (AzK) into the scFv fragments, nnAA-bearing scFvs were conjugated to 20-kDa PEG molecules functionalized with DBCO. The DBCO moiety provides a strained alkyne group that reacts with an azide to form a triazole linkage. The resulting adduct increases the mass of the molecule, resulting in a retardation of gel mobility in SDS-PAGE analyses. As shown in Supplementary Fig. 2, scFv fragments and scFv-PEG conjugates were resolved by SDS-PAGE and stained with Coomassie Blue; a clear increase in molecular mass was observed in the PEGylated samples. To further confirm the specificity of conjugate formation, mass spectrometric analysis of modified scFvs was conducted. The scFv fragments of the HC and N-terminal constructs were reacted with a five-fold molar excess of DBCO-PEG 4 increasing the mass of the adducts by 838 Da. Unmodified and PEGylated samples were analyzed by liquid chromatography-mass spectrometry (LC-MS) and the intact ion mass of the proteins measured (Supplementary Fig. 3). In both cases, we observed an 838 Da increase in the overall mass of the respective PEGylated scFv fragments. No unreacted material was observed in any of the PEGylated samples, indicating very efficient conjugation.
Another consideration for selecting a conjugation site is whether scFv binding function would be impaired by formation of the adduct. Both free and PEGylated scFv fragments were assessed for antigen-binding capability. While PEG chains cannot reproduce the possible influence a conjugated nanoparticle might have on the binding affinity of scFvs, the addition of a 20-kDa flexible PEG identified sites that might interfere with the antigen binding domains. Kinetic binding analyses were performed to assess on-and off-rates of the scFvs and PEGylated scFvs to the ECD of HER2 (Supplementary Fig.   4). The data showed that the N-terminal and HC44 sites resulted in a slight reduction of the equilibrium dissociation constant (K D ) relative to the unmodified scFv (Supplementary Table 1). This is likely due to steric inhibition from the long flexible PEG molecule that interferes with the antigen binding domains of the scFv. The dynamics of the PEG conjugate in the context of a nanoparticle will be different due to the constrained nature of the PEG in the C-dots. The data presented indicate that the site of conjugation offers a viable attachment site that preserves the function of the scFv. The C-terminal site scFv, which contains a 6xHis tag, was not measurable in this assay as NiNTA sensors were used for the HER2 receptor capture.

Surface DBCO and scFv density optimization.
Post surface engineering plays a vital role in developing multifunctional nanoplatforms for cancer diagnosis and treatment. 2,7 Although PEGylation may prevent aggregation of ultrasmall C' dots during surface modification 2,8 and transport in the bloodstream -a key step to ensure bulk renal clearance in animal models and human subjects 9, 10endowing the less than 7 nm sized fluorescent PEG-NH 2 -Cy5-C' dots with two additional functions (PET imaging and targeting) via four additional chemical moieties (i.e., DBCO, DFO, scFv and 89 Zr) of varying molar mass (i.e., 0.6 kDa to 25 kDa) to improve active tumor targeting while maintaining bulk renal clearance and low RES uptake is challenging. Different reaction ratios (i.e., 1:25, 1:50 and 1:100) between PEG-NH 2 -Cy5-C' dot and DBCO-PEG4-NHS ester were explored resulting in accessible DBCO numbers per particle in the range of 6 to 12, as shown in Supplementary Fig. 8. Quantification of number of accessible DBCOs per particle was achieved by reacting DFO-DBCO-PEG-Cy5-C' dots with an excess amount of azide-functionalized green dye, i.e., carboxyrhodamine 110 alkyne (or azide-488) as schematically shown in number of accessible DBCO molecules per C' dot should equal that of azide-488 dye per C' dot due to high yields of the specific strain-promoted azide-alkyne cycloaddition (SPAAC) reaction between DBCO and azide groups. Free unreacted azide-488 was separated from the azide-488-DFO-DBCO-PEG-Cy5-C' dot as demonstrated in Supplementary Fig. 10, which shows the two distinguishable elution profiles of free azide-488 dye and PEG-NH 2 -Cy5-C' dots. Increasing numbers of azide-488/DBCO per dot were further evidenced by the increased absorbance of azide-488 at 501 nm ( Supplementary Fig. 11).

FCS measurement results of dye-conjugated scFv (i.e., scFv-488) and C' dot
immunoconjugates. The hydrodynamic size of scFv was measured by FCS (Supplementary Fig.   15b), and results compared to DLS measurement results (Supplementary Fig. 15a). In order to endow scFv with fluorescence signal for FCS measurements, a green fluorescent dye, i.e. carboxyrhodamine hydrodynamic diameter of the dye conjugated scFv-488 was determined to be around 3.8 nm by FCS ( Supplementary Fig. 15b), and slightly larger than results from DLS measurements without the dye ( Supplementary Fig. 15a).
In order to rationalize the "apparent" smaller than expected increase of particle size upon fragment attachment, it is important to emphasize that DLS and FCS techniques only provide "effective" particle HDs that correspond to the diffusive properties of an assumed perfectly spherical particle.
Considering that the HD of the base aminated particle, onto which the scFvs were attached, was 6.6 nm, a diameter increase of 0.7 nm after scFv attachment corresponds to a more than 30% increase in particle volume. Since the particle molar mass is roughly around 100 kDa, 11 this substantial particle volume increase is consistent with the molar mass of the scFv, even though the particle size increase seems minor. Another way of thinking about this result is to picture the diffusive properties of the particle-fragment conjugate as that of an anisotropic ellipsoidal object (rather than a perfect sphere).
Even though this object is elongated in one direction (e.g. the particle-fragment axis), its resistance to flow is most likely governed by it cross-sectional area along the ellipsoidal short axis which essentially still is that of the base particle, consistent with our experimental measurements.

GPC Purification of DFO-scFv-PEG-Cy5-C' dots.
We originally tried to use a PD-10 column to remove unreacted scFv-azidefragments from DFO-scFv-PEG-Cy5-C' dots after the fragment attachment step. However, purification was not successful, maybe due to the fact that both samples were eluted from 2.5 mL to 4.0 mL fractions. To illustrate this result, a green dye conjugated scFv fragment was prepared by reacting scFv-azide with carboxyrhodamine 110 DBCO (or DBCO-488) to form scFv-488 (Supplementary Figs. 16a-b). The elution profile, shown in Supplementary Fig. 16c, confirmed a clean separation of scFv-488 from the free un-reacted DBCO-488. In contrast, overlapping fractions (i.e., from 2.5 mL to 4.0 mL) were found between free scFv-488 fragment and PEG-NH 2 -Cy5-C' dots ( Supplementary Fig. 17), resulting in inefficient purification of DFO-scFv-PEG-Cy5-C' dots using a PD-10 column.
To ensure clean separation, a GPC column purification process was used. 11 The GPC elugram of unpurified DFO-scFv-PEG-Cy5-C' dots exhibited three peaks (Supplementary Fig. 6c). A peak at around 9 mins corresponded to the main particle product (Supplementary Fig. 6c), while a peak at about 5 min could be attributed to impurities with large molar mass that can usually be removed by GPC. A peak at around 12 to 13 min elution time could be shown to correspond to free azidefunctionalized scFv (Supplementary Fig. 6c bottom), revealing unreacted azide-scFv. By collecting the elution volume between about 8 and 11 min, all impurities could be successfully removed, resulting in a GPC elugram with a single peak (Supplementary Fig. 6d) and suggesting high product purity.

Quantification of surface functional group density by UV-Vis.
The numbers of DBCO and scFv per C' dot were further quantified using an indirect approach based on the conjugation efficacy of DBCO groups. As compared to the quantification based on UV-Vis spectrophotometry, which detects all the DBCO moieties on C' dot surface, the conjugation efficacy approach is only sensitive to the DBCO groups that are reactive. As a result, while 22 DBCO groups were identified using the spectrophotometry approach, the number of accessible DBCO moieties per C' dot was estimated to be about 7 using the conjugation efficacy approach (Supplementary Table 2). This difference suggests that a considerable fraction of the DBCO groups on the C' dot surface is not accessible, and therefore cannot react with azide-functionalized molecules. This may be due to DBCO groups folding back onto the (microporous) silica surface under the PEG layer. In contrast, measurements of the number of scFv fragments per particle exhibited high consistency between the two qualification methods Table 2). This is expected as scFv could only react with accessible DBCO groups, and the quantification of scFv is therefore not affected by the percentage of non-accessible DBCO.

(Supplementary
Considering that the number of accessible DBCO ligands is more meaningful than the overall ligand number, in the remainder of this paper we will exclusively refer to the former obtained using the conjugation efficacy approach when describing DBCO ligand numbers.
To estimate mean organ absorbed doses and the effective dose of 89 Zr-DFO-scFv-PEG-C' dots in a 70-kg standard man, dosimetric analyses (Supplementary Table 3) 5 were performed on the basis of the biodistribution data shown in Supplementary Fig. 27. Overall, favorable dosimetry was found for this radioimmunoconjugate in terms of total-body dose equivalent (~0.2 mSv/MBq) and effective dose (~0.2 mSv/MBq). An dose equivalent of ~0.4 mSv/MBq was estimated for the liver, which was only one-fourth of that reported for 89 Zr-DFO-trastuzumab. In the latter case, the liver was found to be the dose-limiting organ, with an uptake of ~12 %ID, and an average estimated absorbed dose of 1.54 mSv/MBq. 19 These clinically promising results led to our subsequent rigorous evaluation of targeted uptake in breast cancer models with 89 Zr-DFO-scFv-PEG-Cy5-C' dots.

2.8.
Additional in vivo studies were performed using newly-developed, smaller-sized targeted particles (i.e., 6.4 nm) having a comparable scFv surface density to assess size-dependent changes in PK profiles and/or HER2-specific targeting and retention in (i) BT-474 xenografted mice (beyond the 72-hour window) and in (ii) an orthotopic BT-474 model. In the former case, prolonged targeted particle retention was observed up to 10 days, with T/M and T/L ratios of >20 and nearly 3, respectively ( Supplementary Fig. 31). Enhanced target-specific uptake (i.e., greater than 2-fold) was also observed in orthotopic tumors using 89 Zr-DFO-scFv-PEG-Cy5-C' dots, in comparison to that found for control group (i.e., mice injected with isotype control scFv fragments conjugated C' dots, 89 Zr-DFO-Ctr/scFv-PEG-Cy5-C' dots, Supplementary Fig. 32). Importantly, while the use of a smaller diameter particle modestly lowered liver uptake, it also led to alterations in other biological properties, including faster blood clearance and mildly reduced absolute tumor uptake and T/L ratios (Supplementary Fig. 33).

In vivo
Growth of the tumor was monitored by bioluminescence imaging (BLI) using an IVIS Spectrum small  Figs. 32j-k). Reduced tumor uptake was found for the smaller sized 89 Zr-DFO-scFv-PEG-Cy5-C' dots (controls), as against that seen with the larger-sized particles (6.4±1.1 %ID/g vs 13.2±2.9 %ID/g). However, orthotopic BT-474 tumor sizes were also seen to be smaller, on average, for the former group (i.e., ~4 mm), as against those measured for the latter group (i.e., ~6-7 mm). As such, reduced tumor uptake may be attributed to less neovascularization of the relatively smaller orthotopic BT-474 tumors or, perhaps reflects the use of a smaller-size particle probe that clears more rapidly (Supplementary Fig. 33a) and demonstrates a shorter blood circulation half-life.
2.9. Impact of smaller sized probe on in vivo PK. It has been generally accepted that, for renally clearable nanoparticles (e.g. quantum dots 20 ), a decrease in particle size usually leads to a shorter blood circulation half-life and faster renal clearance rate. In this section, we compared differences in tumor and liver uptake for probes having two different hydrodynamic diameters. As expected, and as shown in Supplementary Fig. 33, decreasing the HD of 89 Zr-DFO-Ctr/scFv-PEG-Cy5-C' dots by about 1 nm could lead to lower blood retention times, lower liver uptake, lower tumor uptake, as well as a lower tumor-to-liver ratio. For example, the blood activity-concentration of 6.4 nm 89 Zr-DFO-Ctr/scFv-PEG-Cy5-C' dots was on average 25-30% less than that of the 7.3 nm sized probes. The reduction in non-specific liver uptake was not as obvious as that seen for the blood. Larger sized 89 Zr-DFO-Ctr/scFv-PEG-Cy5-C' dots showed 10-20% higher BT-474 uptake and tumor-to-liver ratios. Taken together, our results suggested that the current design of the "target or clear" ~7.3 nm 89 Zr-DFO-scFv-PEG-Cy5-C' dots, bearing about 1.4 scFv fragments per particle, achieves the best balance among the biological properties measured, that is, it yields the highest HER2-specific targeting efficacy, while maintaining dominant renal clearance and low non-specific RES accumulations.    While liver and spleen uptake values were found to be less than 5 %ID/g for 89 Zr-DFO-scFv-PEG-Cy5-C' dots with scFv ligand numbers per dot equal to or less than 1.3, significantly reduced renal clearance rates, accompanied by increased liver and spleen uptake values, were observed for nanoconjugates bearing 2 scFv fragments per particle due to an increased HD of ~8 nm. All 89 Zr-DFO-scFv-PEG-Cy5-C' dots were synthesized using DFO-DBCO-PEG-Cy5-C'dots bearing 6-7 accessible DBCO per C' dot. Inset shows the corresponding hydrodynamic diameters of the tested samples.

Morphology
WBC morphology is within normal limits.
A manual differential was not performed due to low WBC count (<2.00 K/uL).
A manual differential was not performed due to low WBC count (<2.00 K/uL).