A quantitative NIR molecular imaging analysis of nanoliposom

ISSN 1998-0124 CN 11-5974/O4 2 https://doi.org/10.1007/s12274-020-3178-x

Res

earc

h Ar

ticle

Is tumor cell specificity distinct from tumor selectivity in vivo?: A

quantitative NIR molecular imaging analysis of nanoliposome targeting Girgis Obaid1,2,†,§, Kimberley Samkoe3,§, Kenneth Tichauer4, Shazia Bano1, Yeonjae Park3, Zachary Silber1, Sassan Hodge3, Susan Callaghan1, Mina Guirguis,2 Srivalleesha Mallidi1,‡, Brian Pogue3, and Tayyaba Hasan1,5 ()

Nano Res., Just Accepted Manuscript • https://doi.org/10.1007/s12274-020-3178-x http://www.thenanoresearch.com on Oct. 13, 2020 © Tsinghua University Press 2020 Just Accepted

This is a “Just Accepted” manuscript, which has been examined by the peer-review process and has been accepted for publication. A “Just Accepted” manuscript is published online shortly after its acceptance, which is prior to technical editing and formatting and author proofing. Tsinghua University Press (TUP) provides “Just Accepted” as an optional and free service which allows authors to make their results available to the research community as soon as possible after acceptance. After a manuscript has been technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Please note that technical editing may introduce minor changes to the manuscript text and/or graphics which may affect the content, and all legal disclaimers that apply to the journal pertain. In no event shall TUP be held responsible for errors or consequences arising from the use of any information contained in these “Just Accepted” manuscripts. To cite this manuscript please use its Digital Object Identifier (DOI®), which is identical for all formats of publication.

Address correspondence to First A. Firstauthor, email1; Third C. Thirdauthor, email2

1

Is Tumor Cell Specificity Distinct from Tumor Selectivity In Vivo?: A Quantitative NIR Molecular Imaging Analysis of Nanoliposome Targeting Girgis Obaid*, Kimberley Samkoe*, Kenneth Tichauer, Shazia Bano, Yeonjae Park, Zachary Silber, Sassan Hodge, Susan Callaghan, Mina Guirguis, Srivalleesha Mallidi, Brian Pogue, and Tayyaba Hasan* Massachusetts General Hospital and Harvard Medical School, United States

Page Numbers. The font is ArialMT 16 (automatically inserted by the publisher)

Biomolecular interactions of targeted nanoliposomes (tNLs) with solid tumor receptors in vivo are elusive. Here, for the first time, tNLs specific for EGFR are used with NIR molecular imaging for non-invasive quantitation of their in vivo molecular specificity towards tumor EGFR. We also show that tumor specificity is independent of tumor selective delivery. The purpose of targeting is thus emphasized and tNL development is encouraged, given that the principal goal for targeting is in vivo molecular specificity.

2

Is Tumor Cell Specificity Distinct from Tumor Selectivity In Vivo?: A Quantitative NIR Molecular Imaging Analysis of Nanoliposome Targeting

Girgis Obaid1,2†*, Kimberley Samkoe3*, Kenneth Tichauer4, Shazia Bano1, Yeonjae Park3, Zachary Silber1, Sassan Hodge3, Susan Callaghan1, Mina Guirguis,2 Srivalleesha Mallidi1†, Brian Pogue3, and Tayyaba Hasan1,5 () 1 Wellman Center for Photomedicine, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts 02114, U.S. 2 Department of Bioengineering, University of Texas at Dallas, Richardson, Texas 75080, U.S. 3 Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire 037551, U.S. 4 Armour College of Engineering, Illinois Institute of Technology, Chicago, Illinois 60616, U.S. 5 Division of Health Sciences and Technology, Harvard University and Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, U.S. † Girgis Obaid present address: Department of Bioengineering, University of Texas at Dallas, Richardson, Texas 75080, U.S. † Srivalleesha Mallidi present address: Department of Biomedical Engineering, Tufts University, Medford, Massachusetts 02155, U.S. * Both authors contributed equally to the study Received: day month year / Revised: day month year / Accepted: day month year (automatically inserted by the publisher) ©The Author(s) 2010. This article is published with open access at Springerlink.com

ABSTRACT The significance and ability for receptor targeted nanoliposomes (tNLs) to bind to their molecular targets in solid tumors in vivo has been questioned, particularly as the efficiency of their tumor accumulation and selectivity is not always predictive of their efficacy or molecular specificity. This study presents, for the first time, in situ NIR molecular imaging-based quantitation of the in vivo specificity of tNLs for their target receptors, as opposed to tumor selectivity, which includes influences of enhanced tumor permeability and retention. Results show that neither tumor delivery nor selectivity (tumor-to-normal ratio) of cetuximab and IRDye conjugated tNLs correlate with EGFR expression in U251, U87 and 9L tumors, and in fact underrepresent their imaging-derived molecular specificity by up to 94.2%. Conversely, their in vivo specificity, which we quantify as the concentration of tNL-reported tumor EGFR provided by NIR molecular imaging, correlates positively with EGFR expression levels in vitro and ex vivo (Pearson’s r= 0.92 and 0.96, respectively). This study provides a unique opportunity to address the problematic disconnect between tNL synthesis and in vivo specificity. The findings encourage their continued adoption as platforms for precision medicine, and facilitates intelligent synthesis and patient customization in order to improve safety profiles and therapeutic outcomes.

KEYWORDS Molecular Recognition, Receptors, Nanoparticles, Specificity, Cancer

Nano Res (automatically inserted by the publisher) DOI (automatically inserted by the publisher) Research Article

———————————— Address correspondence to Tayyaba Hasan, [email protected]

3

Introduction Throughout the past two decades of pre-clinical

and clinical development of targeted nanoliposomes (tNLs) as platforms for precision medicine, the value of functionalizing nanoliposomes with tumor-specific ligands has been the focus of significant controversy. Tumor delivery of larger targeted macromolecular nanosized constructs, such as tNLs, is heavily influenced by their pharmacokinetic behavior and their enhanced permeability and retention in tumors [1-5]. Thus, the tumor delivery and selectivity of tNLs becomes less and less influenced by specific ligand-receptor binding as their size increases. This relationship between targeted nanoparticle size, affinity for target receptors and tumor diffusivity was elegantly modelled by Schmidt and Wittrup [6]. The model concluded that as the diameter of ligand functionalized nanoparticles exceed 50 nm, the improvements in bulk tumor delivery and retention that were initially provided by ligand targeting diminish, and nanoparticle size becomes increasingly more influential [6]. Thus, from a pharmacokinetic perspective, the advantages of ligand targeting of nanoliposomes specifically in the 80-150 nm size range, which would justify their complex fabrication and optimization can become ambiguous. However, in certain tNL systems, ligand targeting has been repeatedly shown to markedly improve the therapeutic outcomes of nanoliposomes and other nanomedicines, even without an improvement in tumor delivery or tumor selectivity [7-9]. Herein lies the critical distinction between tNL selectivity (tumor-to-normal ratio) that is largely influenced by the preferential accumulation in tumors, and specificity, which is the discrete molecular affinity that tNLs exhibit for tumor cell receptors. The significance of this distinction is that a tNL can only serve as a platform for precision medicine (targeted drug delivery, imaging, diagnostics, and guided surgery) if it exhibits the desired molecular specificity in vivo. Evidence in the literature suggests that a tNL’s tumor selectivity alone is not sufficient to reliably and consistently report on its molecular specificity for target tumor receptors, which are the crux of precision medicine. Although the tumor delivery and tumor selectivity of tNLs and targeted nanoparticles in general, have been traditionally

used to evaluate their specificity, the growing body of literature strongly suggests that this approach is inconclusive. Thus, in the absence of non-invasive and quantitative approaches for establishing and measuring the in vivo specificity of tNLs, it remains largely elusive without arduous tissue sampling and processing.

A number of preclinical studies investigating nanoliposomes and polymeric nanoparticles targeted with antibodies[7], antibody fragments[8] and natural ligands[9] have found that ligand targeting did not increase tumor delivery or tumor selectivity of the nanoconstructs, but did improve treatment efficacy [7-9]. In other studies, ex vivo tumor analysis concluded that targeting nanoliposomes with scFv antibody fragments increased their intracellular uptake in cancer cells by up to six-fold, which also markedly improved their therapeutic indices [10, 11]. This improvement in therapeutic indices was independent of the fact that overall delivery and tumor selectivity was not improved by ligand targeting of the nanoliposomes. These reports, however, are not consistent for all nanoparticle systems. A recent study found that the delivery of gold nanoparticles to SKOV-3 was enhanced by trastuzumab (anti-HER-2 mAb) targeting, however, it was largely a result of increased macrophage uptake and non-specific binding to the extracellular matrix [12]. These findings, amongst others, emphasize that tNL delivery and selectivity in tumors are not always a reliable indicator for tNL specificity, nor do they consistently predict an improvement in efficacy.

Currently, non-invasive and robust method for measuring tNL specificity in vivo exists. Furthermore, quantifying targeted tumor cell internalization ex vivo using electron microscopy and flow cytometry, amongst other techniques, is invasive, arduous and subject to false-positives due to non-specific phagocytosis. Thus, there remains a critical need to non-invasively measure tNL specificity in order to be able to conclusively determine the conditions (i.e. nanosystem, nature of ligand, nature of molecular target, conjugation strategy etc.) under which ligand targeting may prove to be beneficial. Molecular specificity of tNLs is especially important with regard to NIR-activatable photodynamic therapy (PDT), which is a particular focus of our research

4

efforts.[2-5] For PDT, targeted binding, internalization and photoactivation kinetics are critical for tumor cell selective photodestruction [2, 13-17]. The short-lived (ns-µs half-lives) therapeutic reactive molecular species generated by NIR PDT, such as singlet oxygen, hydroxyl radicals and superoxide anions, have a short diffusion range (nm

- µm range), which necessitates cellular, subcellular

and molecular precision during photoactivation [3, 4, 18-23]. Recent advances in sophisticated NIR-responsive nanoparticles have also been reported demonstrating a prominent role in NIR photoactivatable cancer immunotherapy[24] and semi-conducting nanopolymers for hypoxia-activated PDT,[25] amongst many others

[26].

Scheme 1. Conceptual representation of how the in vivo specificity of NIR-active targeted nanoliposomes (nano-Cet-680; NIRA-tNLs) for solid tumor receptors is measured using NIR molecular imaging. IRDye 680RD and Cetuximab functionalized NIRA-tNLs are intravenously administered to mice bearing tumors overexpressing EGFR. A non-specific spectrally distinct nanoconstruct functionalized with IRDye 800CW and IgG (nano-IgG-800) is co-administered as a sham NIRA-tNL-mimetic reference for in situ quantitative NIR molecular imaging and binding potential derivation. The in situ specificity of the NIRA-tNL for EGFR is calculated from the derived binding potentials and presented as the concentration of EGFR it reports in vivo.

In the clinic, antibody-based tNLs are of the most developed classes of anti-cancer targeted nanomedicines. These predominantly include liposomal carriers of chemotherapy and oligonucleotides that are targeted with various antibody formats [27-30]. To address the long-standing controversy regarding tNL specificity, we present here for the first time a prototypic and clinically relevant antibody-directed tNL in conjunction with a robust in situ analysis of its in vivo molecular specificity towards its target receptor. Given that EGFR is a tumor-associated receptor that is overexpressed in 97.5% of glioblastoma (GBM) patients with EGFR gene amplifications, and is also overexpressed in a diverse range of cancers, such as colon cancer, head and neck cancer, skin cancers and

breast cancer, we use EGFR as a prototypic target receptor for the evaluation of the in vivo molecular specificity of tNLs in GBM xenografts [31-34]. We prepare an NIR-active (NIRA)-tNL formed of a cetuximab (anti-EGFR mAb)-directed nanoliposome modified with IRDye 680 (nano-Cet-680) in addition to a spectrally distinct sham mimetic (nano-IgG-800) for quantitative NIR molecular imaging (Scheme 1). By quantifying tNL interactions with EGFR in situ in U251 (high EGFR), U87 (medium EGFR) and 9L (EGFR-null) GBM xenografts, we address the question of whether or not tNLs actually exhibit tumor cell specificity in vivo, and if it is in fact quantifiable and distinct from tumor selectivity. As outlined in Scheme 1, the general working principle for the non-invasive measurement of the in vivo specificity of tNLs is as follows: 1) synthesis of a

5

molecular targeted NIRA-tNL functionalized with a specific NIR fluorophore, 2) fabrication of a structurally and biologically mimicking sham construct with a second spectrally distinct NIR fluorophore, 3) the co-administration of both the NIRA-tNL and the sham mimetic into animals bearing human tumor xenografts with known tumor EGFR expression levels, 4) longitudinal tracing of both constructs, 5) quantitation of binding potentials, and 6) calculation of the tNL-reported EGFR concentrations. Both the parenterally administered NIRA-tNL and the sham mimetic enter into the vascular system, extravasate and passively accumulate at the tumor site. The NIRA-tNL then interacts with the ectodomain of tumor cell EGFR, while the interstitial sham remains unbound. The calculated binding potentials are converted into in vivo concentrations of NIRA-tNL-reported EGFR. For the purpose of this study using the molecular imaging approach, the concentrations of NIRA-tNL-reported EGFR are presented as a quantitative measure of their in vivo specificity – the concentration of EGFR that the NIRA-tNLs are able to report in vivo correspond to the degree of their specificity for EGFR in vivo.

The findings we present here are the first demonstration of quantitative and non-invasive in situ measurements of the molecular specificity of ligand directed tNLs, addressing a significant controversy in the field of molecular targeted nanoscale therapeutics.

We present previously unattainable insights into the molecular interactions between larger, clinically-relevant tNLs and their target tumor receptors in vivo following intravenous administration. This study proposes a paradigm shift in understanding of how tumor specificity of tNLs can be defined, validated and iteratively tuned. This work also serves to deepen the understanding of the in vivo behavior of ligand targeted nanotherapeutics, in order to minimize failures that often occur after significant investments in time and resources. Such an approach based on rigorous in vivo molecular specificity also holds significant potential for expediting the clinical translation of tNLs to improve the specificity, and potentially the tolerability and efficacy of current and emerging anti-cancer therapeutics.

1. Experimental

1.1. Synthesis and characterization of the NIRA-tNL and the sham tNL mimetic.

Anionic DOPG-containing DPPC liposomes were prepared with 5 mol % total DSPE-PEG2000 content. Of that total 5 mol % of DSPE-PEG2000, 0.2 mol % was amine-derivatized (DSPE-PEG2000-NH2) to mediate covalent amide linkage with the IRDye NHS esters and 0.5 mol % was azadibenzocyclooctyne derivatized (DSPE-PEG2000-DBCO) to allow for copper-free click chemistry with the azido-derivatized Cetuximab (Cet) or the irrelevant human IgG isotype control (IgG). The lipid mixture consisting of DPPC, DOPG, cholesterol and DSPE-mPEG2000, DSPE-PEG2000-DBCO and DSPE-PEG2000-NH2 were combined at mol % values of 58.2, 7.9, 28.9, 4.3, 0.5 and 0.2, respectively.

The lipids dissolved in chloroform were mixed and dried using a nitrogen flow then stored in a vacuum overnight to remove residual chloroform. The dry lipid films were hydrated in 1x DPBS and subject to five freeze-thaw-vortex cycles in ice and in a 42 oC water bath. The hydrated lipid mixture was then extruded 11 times through 0.1 μm polycarbonate membranes (Whatman®). Extruded liposomes were reacted with either IRDye 680RD-NHS or IRDye 800CW-NHS esters (Li-COR Biosciences, Inc.) at a 5-fold molar excess to DSPE-PEG2000-NH2 for 24 h at room temperature. Unreacted IRDye was removed by dialysis in 100 kDa Float-A-Lyzer (Spectrum Labs) against 1x PBS at 4 oC for 48 h in the dark. Liposomes conjugated to IRDye 680RD or IRDye 800CW will be referred to as nano-680 and nano-800 hereon, respectively. The hydrodynamic diameters, polydispersity indices and ζ-potentials of both nanoconstructs were characterized using a Malvern Zetasizer Nano Dynamic Light Scattering instrument. UV Visible absorption spectrophotometry was used to determine the concentration of IRDye 680RD (ε 672 nm = 165,000 M-1.cm-1) and IRDye 800CW (ε 778 nm = 240,000 M-1.cm-1) using a ThermoFisher Evolution 300 UV-Visible Absorption Spectrophotometer.

Cetuximab (Cet) and IgG (2 mg/ml in 1x DPBS) were reacted with a 2.5-fold molar excess of NHS-PEG4-N3 (Thermo Fisher Scientific) for 24h at 4

oC. Derivatized antibodies were purified with PD10

6

desalting columns (GE Healthcare) equilibrated with 1x DPBS and concentrated to 4 mg/ml using Amicon Ultrafiltration Tubes (30 kDa MWCO, EMD Milipore) centrifuged at 2,000 xg for 30 minutes.

The nano-680 and nano-800 nanoconstructs were conjugated to the azido derivatized Cet and IgG, respectively, using copper-free click chemistry by reacting at a ratio of 50 antibodies per liposomes for 24 h at room temperature. The nano-Cet-680 NIRA-tNL and the nano-IgG-800 sham mimetic were purified from unreacted antibodies using Sepharose CL-4B (Sigma-Aldrich) size exclusion columns equilibrated with 1x DPBS. The hydrodynamic diameters, polydispersity indices and ζ-potentials of both the nano-Cet-680 NIRA-tNL and the nano-IgG-800 sham mimetic were measured using a Malvern Zetasizer Nano. The concentrations of IRDye 680RD and IRDye 800CW were determined as described earlier. All nanoconstructs were stored at 4

oC in the dark. In order to quantify the conjugation

efficiency of Cet and IgG to the nano-680 and nano-800 constructs, antibody solutions were reacted with 2.5-fold molar excess of Alexa Fluor 488-NHS ester and 2.5-fold molar excess of NHS-PEG4-N3 and purified using size exclusion chromatography as described earlier. Following click chemistry conjugation of the antibodies to the nanoliposomes, the fluorescence of Alexa Fluor 488 was measured using an Excitation of 480 nm and an emission wavelength of 517 nm. Antibody concentrations were back-calculated from standard curves of Alexa Fluor 488–antibody conjugates to calculate conjugation efficiencies.

Fluorescence measurements of the free IRDye molecules, the IRDye conjugated nanoliposomes, the nano-Cet-680 NIRA-tNL, and the nano-IgG-800 sham mimetic were all performed in 1x DPBS using a Horiba Fluoromax Fluorimeter. Excitation at 660 nm was used for IRDye 680RD-containing samples and an excitation at 760 nm was used for IRDye 800CW-containing samples.

1.2. Tissue culture of tumor cells

U251, U87 and 9L tumor cells were purchased from ATCC as authenticated mycoplasma-free cell lines that have all been handled separately. All three cell lines also tested negative for mycoplasma using a

standard PCR testing kit. The cells have not been listed as cross-contaminated or misidentified in the ICLAC database. The cells were cultured in media containing 10% fetal bovine serum and 1x penicillin/streptomycin at 37oC and 5% CO2.

1.3. NIRA-tNL binding specificity in vitro U87 cells were cultured to 80% confluence, trypsinized, and 5x104 cells were incubated with varying concentrations of Cet and IgG conjugated to Alexa Fluor 488 and NHS-PEG4-N3 at 37 oC for 30 min. Cells were then pelleted at 1,000 xg for 5 min, the antibody containing supernatant was removed and the cells were redispersed in PBS. The cells were then run on a BD FACS Aria flow cytometer using the Alexa Fluor 488 fluorescence and analyzed with FlowJo. A Cet-conjugated NIRA-tNL and an IgG-conjugated sham mimetic were prepared as before but included 0.1 mol % 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (16:0 Liss Rhod PE), as opposed to the DSPE-PEG2000-NH2 used for IR dye coupling. The nano-Cet-Rhod NIRA-tNL and the nano-IgG-Rhod sham mimetic were incubated with 5x104 cells for 30 min at 37 oC. Cells were then pelleted at 1,000 xg for 5 min, the supernatant was disposed of, the cells were redispersed in PBS and run on a BD FACSAria flow cytometer using the Liss Rhod PE fluorescence. The data was also analyzed with FlowJo.

1.4. Quantitative EGFR flow cytometry A thorough description of receptor quantification using flow cytometry has been reported previously [35, 36]. Briefly, U251, U87 and 9L cells were cultured to 80% confluence, trypsinized, and incubated with EGF-biotin, then with Cy5-streptavidin with appropriate washing steps. The cells were run on a Miltenyi MACSQuant flow cytometer and analyzed with FlowJo. In addition to the cells, QuantumTM MESF beads (Bangs Laboratories, Inc.) were run and analyzed for each trial such that a standard curve could be made to validate the number of fluorophores per receptor, assuming 3 Cy5 fluorophore molecules are bound to each streptavidin molecule as per the manufacturer’s characterization. All experiments were performed in triplicate.

7

1.5. In vivo quantitative NIR molecular

imaging All procedures were performed according to an approved protocol from the Institutional Animal Care and Use Committee (IACUC) at Dartmouth College. Female athymic nude mice (Charles River Laboratories, CT) were implanted subcutaneously in the right flank with U251, U87, or 9L cell lines at a concentration of 1×106 cells per 50 μL mixed in 1:1 Matrigel: culture medium. When tumors reached ~200-300 mm3 the mice were anesthetized (1-2% isoflurane, 1 L/min O2) and placed in the Pearl Impulse (LI-COR Biosciences, Inc., Lincoln, NE) imaging system. Pre-injection autofluorescence images (t=0) were collected in both the 700 (IRDye 680RD) and 800 nm (IRDye 800CW) channels prior to injecting 200 μL of a 1:1 (0.1 nmol) dye equivalent mixture of nano-Cet-680: nano-IgG-800 administered intravenously through tail vein injections. Images were collected immediately after injection (t=1) and then every five minutes afterward for 15 minutes. The mice were allowed to wake up and were imaged again at 24, 48- and 72-hours post-administration. After the 72-hour time point, the U87 mice had the skin removed and the tumor was imaged again to validate the binding potential calculations derived from imaging through the skin. After sacrifice, all tumors were excised and fixed in formalin for Hematoxylin and Eosin (H&E) staining and EGFR immunohistochemistry using horseradish peroxidase secondary antibodies from the Pathology Translational Research Laboratory as described previously [36]. Quantitative NIR molecular imaging was performed on 4 U251-bearing mice, 3 U87-bearing mice, and 5 9L-bearing mice, with three individual timepoints (24 h, 48 h, 72 h) being analyzed comparatively throughout the different tumor-bearing mice.

1.6. Biodistribution of the NIRA-tNL Mice were implanted with the U251, U87 and 9L tumors as described earlier. Once tumors reached ~200-300 mm3 in volume, mice were anesthetized and injected with a 200 μL solution of sterile PBS containing 0.1 nmol IRDye equivalent of the NIRA-tNL. 24 h following administration, mice were euthanized, and the tissue was harvested. Tissue

uptake of the NIRA-tNL was quantified using fluorimetry measurements with the Pearl Impulse (LI-COR Biosciences, Inc., Lincoln, NE) imaging system using respective reference standards of the construct.

1.7. Cross-sectional microscale tumor imaging U251, U87 and 9L tumors harvested from mice 24 h following administration of 0.1 nmol IRDye equivalent of the NIRA-tNL were bisected and imaged using the LI-COR Odyssey Imaging System at 21-μm resolution. Bisected tumors were also fixed in formalin and stained for EGFR using peroxidase immunohistochemistry.

1.8. In vivo NIR photoacoustic imaging Photoacoustic imaging of the nano-Cet-680 NIRA-tNL, and the nano-IgG-800 sham mimetic was performed in vivo in the U87 murine xenograft model using a VisualSonics Vevo LAZR Imaging System before and 24 h following intravenous administration of an equimolar cocktail of the two constructs (2.5 nmol IRDye equivalent for each construct) in PBS.

1.9. Quantitative NIR molecular imaging analysis

All image analysis was performed using ImageJ imaging software, available from the NIH [37]. The average mean fluorescence intensity was collected for each tumor region and normal skin in both the 700 and 800 nm channel images for each time point. Single-time point binding potential values for the tumor and skin were calculated at each time point using the general form of Equation 1,[38] where 𝑅𝑂𝐼𝑐𝑒𝑡𝑐𝑜𝑟𝑟 and 𝑅𝑂𝐼𝐼𝑔𝐺𝑐𝑜𝑟𝑟 represent the corrected signal contribution in the region of interest (ROI) for each tissue from the nano-Cet-680 and nano-IgG-800, respectively. Equations 2 - 5 describe how the corrected signal contribution for nano-Cet-680 and nano-IgG-800 was determined in each tissue at each time point ranging from 0-72 hours, where t=0 is the autofluorescence value and t=1 is the first time point after injection. The corrected regions-of-interest take into consideration that the tumor regions involved mixed signal from both the tumor and skin in vivo and the average intensities of the sham nano-IgG-800 were normalized to that of the nano-Cet-680.

8

(Equation 1) 𝐵𝑃 =𝑅𝑂𝐼𝑐𝑒𝑡

𝑐𝑜𝑟𝑟−𝑅𝑂𝐼𝐼𝑔𝐺𝑐𝑜𝑟𝑟

𝑅𝑂𝐼𝐼𝑔𝐺𝑐𝑜𝑟𝑟

(Equation 2) 𝑅𝑂𝐼(𝑡𝑢𝑚𝑜𝑟)𝐶𝑒𝑡𝑐𝑜𝑟𝑟 = 𝑅𝑂𝐼(𝑡𝑢𝑚𝑜𝑟 + 𝑠𝑘𝑖𝑛)𝐶𝑒𝑡𝑡=𝑖 − 𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐶𝑒𝑡𝑡=𝑖

(Equation 3) 𝑅𝑂𝐼(𝑡𝑢𝑚𝑜𝑟)𝐼𝑔𝐺𝑐𝑜𝑟𝑟 = �𝑅𝑂𝐼(𝑡𝑢𝑚𝑜𝑟 + 𝑠𝑘𝑖𝑛)𝐼𝑔𝐺𝑡=𝑖 − 𝑅𝑂𝐼(𝑡𝑢𝑚𝑜𝑟)𝐼𝑔𝐺𝑡=0� �𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐶𝑒𝑡𝑡=1

𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐼𝑔𝐺𝑡=1�

(Equation 4) 𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐶𝑒𝑡𝑐𝑜𝑟𝑟 = 𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐶𝑒𝑡𝑡=𝑖 − 𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐶𝑒𝑡𝑡=0

(Equation 5) 𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐼𝑔𝐺𝑐𝑜𝑟𝑟 = �𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐼𝑔𝐺𝑡=𝑖 − 𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐼𝑔𝐺𝑡=0� �𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐶𝑒𝑡𝑡=1

𝑅𝑂𝐼(𝑠𝑘𝑖𝑛)𝐼𝑔𝐺𝑡=1�

(Equation 6) [𝑡𝑁𝐿−𝑏𝑜𝑢𝑛𝑑 𝑡𝑢𝑚𝑜𝑟 𝐸𝐺𝐹𝑅](𝑛𝑀) = 𝐵𝑖𝑛𝑑𝑖𝑛𝑔 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 × 𝑘𝐷(𝑛𝑀)

2. Results and Discussion 2.1. Modular synthesis and

characterization of the NIRA-tNL and the sham mimetic

Nanoliposomes were doped with 0.2 mol% of DSPE-PEG-NH2 for amide coupling with NHS esters of IRDye 680RD and IRDye 800CW. Following purification, the DSPE-PEG-IRDye lipid content was 0.013 ± 0.02% (IRDye 680RD constructs) and 0.018 ± 0.003% (IRDye 800CW constructs); (Table 1). The constructs were also doped with 0.5 mol% of DSPE-PEG-DBCO for copper-free click chemistry coupling to azido- Cet and sham IgG, yielding the nano-Cet-680 NIRA-tNL construct and the nano-IgG-800 sham mimetic. The antibody conjugation efficiency was matched for the NIRA-tNL (47.5%; 23.8 Cet/construct) and the sham mimetic (47.3%; 23.6 IgG/construct); (Table 1). This

degree of ligand surface density is rationally-designed by literature reports of Cet-liposomes, [39, 40] whereby increasing the Cet density from 24 to 40 per construct reduces the tumor uptake by 3.7-fold [40]. Given that tNL circulation half-lives and liver clearance are also impacted by ligand surface density, it was critical to match Cet and IgG densities in the two constructs [40, 41]. Matching the surface density and molecular weight of PEG, in addition to the hydrodynamic diameter and ζ-potential of the NIRA-tNL (133.2 ± 1.6 nm, -18.9 ± 0.8 mV, respectively) with the sham mimetic (134.5 ± 2.4 nm, -19.1 ± 0.5 mV; Table 1) was also critical. The shelf life for both constructs extended beyond 6 weeks in storage at 4°C, as the constructs remained monodisperse with polydispersity indices below 0.1 (Fig. 1(e), S-1(g,h) in the Electronic Supplementary Material (ESM)).

Table 1. Physical characterization of nanoliposomal formulations throughout each step of the preparation of the nano-Cet-680 NIRA-tNL and the nano-IgG-800 sham mimetic. Values are mean (± S.D.).

Nanoconstruct Formulation

Hydrodynamic Diameter (nm)

Polydispersity Index

ζ-potential (mV)

mol % IRDye-lipid of total liposomal nanoconstruct

Ab per construct

Ab Conjugation Efficiency (%)

nano-IRDye800CW 129.2 (± 0.9) 0.02 (± 0.01) -20.5 (± 2.6) 0.018 (± 0.003) - -

nano-IgG-800 134.5 (± 2.4) 0.03 (± 0.01) -19.1 (± 0.5) 0.018 (± 0.003) 23.6 (± 0.5) 47.3 (± 1.0)

9

(sham mimetic)

nano-IRDye680RD 128.8 (± 1.1) 0.030 (±0.02)

-20.9 (± 0.8) 0.013 (± 0.02) - -

nano-Cet-680 (NIRA-tNL) 133.2 (± 1.6)

0.036 (± 0.007) -18.9 (±0.8) 0.013 (± 0.02) 23.8 (± 0.4) 47.5 (± 0.8)

The optical properties of the free and nanoconstruct IRDye 800CW and IRDye 680RD were measured in PBS (Table 2, Fig. S-1 in the ESM). Importantly for co-imaging in vivo, no change in the absorption maxima of the IRDyes in the NIRA-tNL and the sham mimetic were observed when mixed together (Fig. 1b). There were also no significant redshifts in the absorption maxima or emission maxima of IRDye 680RD and IRDye 800CW following nanoconstruct incorporation (Table 2, Fig. S-1(a-e) in the ESM). The ca. two-fold quenching of

IRDye 680RD following incorporation in the NIRA-tNL is attributed to partial dye stacking and static quenching at the nanoconstruct surface (Fig. S-1(e) in the ESM). This is most likely due to the higher degree of hydrophobicity and lower topological polar surface area of IRDye 680RD (calculated logP = -2.63; polar surface area = 196.69), as compared to the IRDye 800CW which does not quench (calculated logP = -10.00; polar surface area = 256.75). Calculations were performed using ChemDraw Professional 17.0.

Figure 1. a) Modular synthesis of NIRA-tNLs starting with amide coupling of IRDye-680RD to nanoliposomal PEG-NH2, followed by strain-promoted cycloaddition through copper-free click chemistry conjugation of nanoliposomal PEG-DBCO to azido-derivatized Cet. The sham mimetic is fabricated using the same strategy with a non-specific human IgG molecules and IRDye 800CW. b) Absorption spectra of the nano-Cet-680 NIRA-tNL, the nano-IgG-800 sham mimetic, and an equimolar mix of both constructs in PBS (1 μM dye equivalent). Nano-Cet-Rhod tNLs exhibited a greater degree of binding to U87 cells than the nano-IgG-Rhod sham mimetic (c) with up to 14-fold selectivity at lipid equivalent concentrations of 25 µM and 50 µM (d). (e) The shelf life of both the NIRA-tNL and the sham mimetic extends beyond 6 weeks in storage at 4°C. (Mean ± S.E.M., n = 3 biological replicates; Analysis Of Variance (ANOVA) with a Tukey post-test, *** = P<0.0001).

The 42.9% reduction in Φf of the NIRA-tNL resulted in a close matching of the fluorescence

brightness (ε x Φf) of the NIRA-tNL (19,800) and the sham mimetic (23,520), although differences in dye

10

labeling efficiency, optical imaging systems and light collection efficiencies can change this ratio during in vivo imaging. These differences were accounted for by normalizing the nano-IgG-800 signals over time to the ratio of nano-Cet-680 and nano-IgG-800 at t=0h, as described in Equation s 3 and 5. Administering 0.1 nmol dye equivalent of the construct containing 0.018 mol% IRDye 680RD equivalent (NIRA-tNL) or 0.013 mol% IRDye 800CW equivalent (sham mimetic) correspond to a total mouse phospholipid dose of 0.0082 mg/kg and 0.011 mg/kg, respectively. These are considered only microdoses, which are only 0.08% and 0.1% of the typical Doxil lipid load used for treatment, respectively [42].

Tumor Hypoxia, which varies by tumor type, size, vascularity and anatomical site, results in tumor acidosis (pH <6.4 – 7.0) [43]. Changes in tumor pH levels can alter the fluorescence properties of optical imaging contrast agents. As such, we have selected Li-Cor IRDye molecules for this study which are widely used for molecular imaging of solid tumors in pre-clinical studies and in more than a dozen clinical trials. We have further characterized the pH dependence of the fluorescence emission of the IRDye 680RD nanoconstructs and IRDye 800CW nanoconstructs used in this study. Figure S-1(i-j) in the ESM show that the fluorescence emission of the IRDye 680RD and IRDye 800CW are unaltered as the pH drops from 7.0 to 4.0, which spans the range of

the pH changes during tumor acidosis in vivo. Although not relevant to physiological changes in pH, the fluorescence of both the IRDye 680RD nanoconstructs and IRDye 800CW nanoconstructs decreases moderately as the pH increases from 7.0 to 10.0.

Conjugation of Alexa Fluor 488-NHS and the NHS-PEG4-N3 linker to Cet did not impact its specificity for U87 cells, as a dissociation constant (Kd) of 0.106 nM (Fig. S-1(f) in the ESM) was derived, which closely resembles its reported Kd for the isolated EGFR protein (0.38 nM) [44]. Modified IgG was also confirmed to have no binding specificity. The in vitro binding specificity of 16:0 Liss Rhod PE-doped NIRA-tNLs was found to be highest in U87 cells (ca. 14-fold) with respect to the sham mimetic at lipid concentrations of 25 µM and 50 µM (Fig. 1(c,d)).

Given that human IgG is a dysopsonin while Cet is not, we confirm that this specific human IgG isolate is incapable of forming a corona on IRDye nanoliposomes (~1.19 IgG molecule per construct) after 24 h incubation at a serum-relevant IgG concentration of 1 mg/ml [45, 46]. Concerns regarding IgG’s ability to influence pharmacokinetics are also mitigated by the observation that 9L tumor accumulation of the sham mimetic and the NIRA-tNL are not significantly different (Fig. S-5(a) in the ESM).

Table 2. Optical characterization of the IRDye molecules free in PBS, and when conjugated to the nanoconstruct in PBS.

Dye or Nanoconstruct λ Abs max λ Emi max Fluorescence Quantum Yield (Φf)

IRDye 800CW 774 nm [47] 789 nm 0.10 [48]

nano-IgG-800 778 nm 791 nm 0.098

IRDye680RD 672 nm [49] 691 nm 0.21 [50]

nano-Cet-680 672 nm 694 nm 0.12

2.2. Measuring in vitro and ex vivo EGFR expression levels

Flow cytometry was used in order to quantify the EGFR expression per cell in U251, U87 and 9L GBM cell lines [35, 36]. We found that U251 cells expressed 4x104 ± 1x104 EGFR per cell, U87 cells expressed 2x104

± 3x103 EGFR per cell and 9L cells were confirmed to be EGFR-null. Ex vivo EGFR expression levels in

subcutaneous U251, U87 and 9L tumors were evaluated using immunohistochemistry, confirming that a linear relationship exists between in vitro and ex vivo EGFR expression (Fig. S-2 in the ESM).

2.3. In vivo quantitative NIR molecular imaging

11

Quantitative NIR molecular imaging of NIRA-tNL specificity in vivo was performed using an equimolar mix with the sham mimetic in U251, U87 and 9L tumor-bearing mice. Specificity in this study is defined as the in vivo tumor EGFR concentration (receptor density in three-dimensional space) that the NIRA-tNL is capable of reporting. EGFR concentration was derived from binding potentials, which were calculated using the general form of Equation 1 and 6.

Representative images of raw NIRA-tNL tumor delivery are shown in Fig. 2a. We have previously shown that subtracting the signal of a targeted probe from a sham probe increases the contrast for image guided surgery of pancreatic ductal adenocarcinoma,[51]. This enhanced contrast is only

relative to the differences between tumor delivery of the two probes. In this study, quantitative NIR molecular imaging was performed to generate internally-normalized binding potential images, which correspond to NIRA-tNL specificity that are independent of NIRA-tNLs signal strength (efficiency of delivery; Fig. 2a). A visual comparison of NIRA-tNL delivery (nano-Cet-680 contrast) and specificity (binding potential contrast) demonstrates the inadequacy of using NIRA-tNL delivery alone to evaluate molecular specificity. Additionally, quantitative NIR molecular imaging of NIRA-tNL specificity in situ provides opportunities for tumor optical biopsies, informed iterative tNL synthesis, patient customization, and spatially-resolved image-guided surgery and phototherapy [2, 52].

Figure 2. a) Representative images of mice bearing U251 (high EGFR), U87 (medium EGFR) and 9L (EGFR-null) tumors. White light images are matched with optically-corrected NIRA-tNL fluorescence images that represent tumor delivery, and NIRA-tNL binding potential contrast images which correspond to specificity for EGFR (concentration of tNL-reported EGFR). Comparative images split between the NIRA-tNL contrast alone and NIRA-tNL EGFR specificity emphasize that imaging of the tumor accumulation of NIRA-tNLs alone cannot not inform of their in vivo specificity. b) Corresponding quantitation of the in vivo specificity of the NIRA-tNLs in the U251, U87 and 9L tumors, and the combined skin at 24, 48h and 72 following intravenous administration. (Mean ± S.E.M., biological replicates n = 4 (U251), 3 (U87) and 5 (9L); ANOVA with a Tukey post-test; * = P< 0.05, ** = P<0.01, *** = P<0.0005, **** = P< 0.0001.)

During longitudinal imaging, all fluorescence signals were optically corrected and normalized at

time 0 h to account for blood content and accumulation in the tumor interstitium and skin (Fig.

12

S-3 in the ESM). Single timepoint binding potentials were calculated at 24, 48- and 72-h timepoints (Fig. S-4(a,b) in the ESM) [38]. The skin binding potentials in all mice (2.5-2.6) correspond to EGFR expression in the epidermis and basal cell layer of normal skin (Fig. S-4(a,d) in the ESM) [53]. No significant differences were observed between U87 tumor binding potentials derived by optical correction of the skin and physical removal of the skin, confirming

the robustness of the non-invasive imaging we perform in this study (Fig. S-4(c) in the ESM).

As is consistent with the specificity images (Fig. 2a), binding potentials were highest for U251 tumors, followed by U87 tumors, and lowest for 9L tumors. Tumor binding potentials plotted against the respective EGFR receptors/cell (flow cytometry) demonstrated a linear relationship (r2 = 0.8535) with a statistically significant (P≤ 0.0004) Pearson’s r correlation of 0.9239 (Fig. S-4(b) in the ESM).

Table 3. Optical biopsies of U251 tumors derived using a fluorescent variant of the EGF ligand [35] and a fluorescent EGFR affibody [36].

Tumor Specific Tracer Dissociation Constant (Kd; nM)

Binding Potential In vivo EGFR Concentrations Calculated from Binding Potentials (nM)

EGF [35] 1 1.96 1.96

EGFR-specific Affibody [36] 2.1 1.18 2.48

13

Figure 3. a) The imaging-derived in vivo specificity of the NIRA-tNLs (concentration of reported EGFR) in U251, U87,

and 9L tumors correlates positively with the EGFR expression of each individual tumor line at the cellular level in vitro (EGFR/cell) and with the relative tumor EGFR expression levels from ex vivo immunohistochemistry analyses (Pearson’s r =0.9225*** and 0.9591****, respectively). However, neither the in vivo tumor delivery (% I.D./g tumor; b) nor tumor selectivity of the NIRA-tNLs (tumor-to-normal brain ratio; c) at 24 h correlated with EGFR expression ex vivo. While the NIRA-tNL specificity at 24h normalized to that of 9L EGFR-null tumors exhibited a linear trend, both the raw NIRA-tNLs tumor delivery (% I.D./g tumor) normalized to 9L tumors (d) and the NIRA-tNLs tumor selectivity (tumor-to-normal ratio) normalized to 9L tumors (e) grossly underrepresented specificity in U87 and U251 tumors. (Values are mean ± S.E.M., statistical significance was calculated using ANOVA with a Tukey post-test; n = 4 biological replicates (U251), 3 biological replicates (U87) and 5 biological replicates (9L); * = P< 0.05, *** = P<0.0005, **** = P< 0.0001.)

The in vivo NIRA-tNL specificity (concentration of reported EGFR, nM) was then calculated (Equation 6) in all tumors using the binding potentials and the dissociation constant of Cet for EGFR (0.38 nM)[44]). NIRA-tNLs exhibited the highest specificity (reported EGFR concentration) in U251 tumors, followed by U87 tumors, then lowest for 9L tumors (Fig. 2b). As expected, the specificity

for skin EGFR was greater than that for EGFR-null 9L tumors. As a demonstration of the robustness of using molecular imaging to quantify the in vivo molecular specificity of tNLs, we observed statistically significant (P≤ 0.0004) Pearson’s r correlations between the in vivo NIRA-tNL specificity and the tumor EGFR expression in vitro and ex vivo (Fig. 3a). Interestingly, neither tumor delivery (% I.D./g tumor; Fig. 3b) nor tumor selectivity (tumor-to-normal brain ratio; Fig. 3c) correlated with

14

either in vitro or ex vivo tumor EGFR expression levels. Furthermore, NIRA-tNL tumor delivery (%ID/g) grossly under-represented specificity by 63.6% (U87) and 77.5% (U251) (Fig. 3d), while NIRA-tNL tumor selectivity (tumor-to-normal brain ratio) grossly under-represented specificity by 94.2% (U87) and 69.8% (U251) (Fig. 3e). Although optically-corrected NIRA-tNL delivery better matched their molecular specificity in U251 tumors, it still under-represented specificity by 48% in U87 tumors (Fig. S-5(f) in the ESM). It important to reiterate that the efficiency of tumor delivery of nanomedicines is universally advantageous; however, the literature suggests that the desired function of ligand-directed tNLs is their molecular specificity at the cellular level, which is attributed to their oftentimes improved therapeutic indices. Collectively, our findings stress the importance of establishing NIRA-tNL specificity in vivo as the benchmark for tNL interactions with tumor receptors in vivo, as opposed to inferences derived from the evaluation of tumor delivery alone, which we show to be inconclusive for our NIRA-tNL system. NIRA-tNL specificity for U251 tumors was found to be 2.15 ± 0.62 nM EGFR. This is of particular importance, as it closely matches the in vivo EGFR concentrations in U251 tumors previously reported by fluorescently labeled EGF ligand[35] and EGFR-specific affibodies [36] (Table 3). This agreement suggests that our NIRA-tNL exhibits the optimal degree of in vivo specificity for EGFR, as it is evidently capable of reporting the same concentration of tumor EGFR as the EGF ligand [35] and EGFR-specific affibodies. The implications of these findings are of paramount importance, as we hold the tumor EGFR expression levels reported by EGF [35] and EGFR-specific affibodies as the benchmark for the degree of in vivo molecular specificity (concentration of reported EGFR) that an optimal NIRA-tNL should exhibit.

Tumor vascular endothelial cells are also known to express EGFR, and EGFR signaling is also implicated in angiogenesis.[54-56] As such, all pre-clinical and clinical EGFR-targeted treatments, such as antibodies, antibody conjugates and targeted nanoconjugates would likely exhibit specificity for both tumor vascular EGFR, as well as tumor cell EGFR, depending on the tumor. This suggests that

tumor tissue specificity of all such pre-clinical and clinical EGFR-targeted treatments involves a combination of specificity for angiogenic endothelial cells and tumor cells. While it would be important to de-couple NIRA-tNL specificity for angiogenic endothelial cells from that of the tumor cells in this study, it is important to note here that no EGFR was reported by the NIRA-tNLs in the EGFR-null tumors, which also likely contained EGFR-expressing vascular endothelial cells. As such, the contribution of vascular EGFR with respect to total tumor EGFR is not expected to be significant. This is further confounded by the fact that the degree of cross-species reactivity of Cetuximab for mouse EGFR is not known.[57-59] The extent to which an EGFR-targeted entity, such as the NIRA-tNL, exhibits specificity for angiogenic endothelial cells as opposed to the tumor cells themselves is worth exploring in future studies, although the lack of a widely accepted and robust anti-mouse EGFR monoclonal antibody could limit studies in murine xenograft models EGFR over-expressing tumors.[60] This suggests that future studies would likely require a proteomic and transcriptomic analysis to quantify the amount of vascular mouse EGFR with respect to total human tumor EGFR.

It is well-established that the most prevalent types of nanoliposomes used in the clinic (~100 nm in diameter) exhibit size-dependent restrictions in tumor tissue penetration, yet they remain attractive due to their prolonged circulation times and their amenability for high-payload multi-agent entrapment [61, 62]. These restrictions can be further heightened by molecular specificity for tumor receptors through the binding site barrier effect [63, 64]. As expected for receptor targeted nanoliposomes, the NIRA-tNLs exhibited a greater affinity for the tumor periphery in U251 and U87 tumors, and penetrated the tumor core most efficiently in EGFR-null 9L tumors (Fig. S-6 in the ESM). The significance of the NIRA-tNL platform we present here is its amenability for incorporating lipidated photosensitizing molecules[65] for NIR-activable photodynamic therapy (PDT)-based combinations, whereby photochemistry in the tumor can modulate the vasculature and stroma to promote tissue permeation of tNLs [2, 66-68]. Owing to the intrinsic photoacoustic properties of both IRDye 680RD and

15

IRDye 800CW, Figure S-7a in the ESM provides further in vivo evaluation of the tumor distribution of the NIRA-tNLs and sham mimetic using in situ multiwavelength photoacoustic imaging. The U87 tumor uptake of both the NIRA-tNL and the sham mimetic can be visualized cross-sectionally deep within the tumor using photoacoustic imaging, as opposed to the surface visualization of the fluorescence imaging. Furthermore, Figures S-7b,c in the ESM are visual representations of the intra-tumoral and inter-tumoral heterogeneity of the uptake of the non-sham mimetic within the three tumor types 24 h following administration. This heterogeneity underscores the necessity for a sham mimetic construct to serve as an internal reference for each tumor while measuring NIRA-tNL specificity.

These findings presented here provide evidence that tNL specificity is in fact distinct from tumor selectivity, and is both discrete and quantifiable in situ using non-invasive and NIR quantitative molecular imaging. As such, we propose that iterative imaging of tNL specificity should be combined with emerging strategies to promote tumor delivery (e.g. biomimetic and proteolipid coatings), and tumor-permeating technologies (e.g. PDT) to engineer cutting-edge tNLs with enhanced tumor selectivity, delivery, and distribution, while maintaining optimal molecular specificity for tumor tissue. 3. Conclusions

Macromolecular tumor targeted systems, such as tNLs, have been long-associated with questions surrounding their value as tumor receptor-specific entities. There is a school of thought which suggests that tNL selectivity in tumors is enhanced by their biomolecular target recognition (specificity), while another school of thought suggests tumor selectivity dominates through enhanced permeability and retention effects. The true molecular specificity of larger targeted constructs, such as clinically-relevant antibody functionalized tNLs, therefore becomes ambiguous in a complex in vivo environment. Thus, the physiological and therapeutic role of molecular specificity is unclear. However, there are substantive and rigorous studies that have reported a marked increase in the therapeutic indices of nanotherapeutic drug formulations by ligand

targeting [7-10, 69]. This contradiction is particularly complex, given that ligand targeting of nanomedicines has not consistently increased their tumor delivery and selectivity, as defined by the conventional tumor-to-normal ratios. Furthermore, not all reports suggest an improvement in therapeutic indices with targeting. With no known non-invasive and robust approach for evaluating the tumor specificity of tNLs at the molecular level, irrespective of delivery or selectivity, the controversy remains.

This study thus offers a potential resolution for the controversy by presenting an adaptable framework for evaluating tNL specificity in vivo using quantitative NIR molecular imaging. We show this by demonstrating and quantifying, for the first time, the degree of molecular specificity that a tNL exhibits for tumor cell receptors in vivo, which in this case is tumor EGFR. The data provides previously inaccessible insights into the interaction of tNLs with solid tumor cell surface receptors. Specificity in this study, is presented as the capacity for the tNL to accurately report the EGFR expression status in vivo. Critically, this assessment of specificity is internally normalized to each tumor and is not influenced by differences in tumor tNL delivery and selectivity at the macrophysiological level, which can lead to a gross under-representation of specificity as we have shown can be up to 94.2%.

An important implication of the findings we present here, is the capacity for the informed iterative synthesis of tNLs guided by their ultimate molecular specificity in vivo. This quantitative approach for measuring specificity promises to increase the accuracy of tNL development. It does so by helping avoid false positives that ultimately fail to deliver improvements in therapeutic outcomes, and also avoid false negatives that wrongly exclude nanoconstructs with the capacity for molecular targeted cellular delivery, even though they may not increase overall tumor delivery. The power of this quantitation of tNL specificity can also be extended to additional nanoconstructs targeted to various receptor combinations and non-receptor targets, such as disease-associated glycans,[3] and for developing increasingly critical, and likewise complex, photoactivatable anticancer tNLs. These developments can thus push the boundaries of

16

current nanotechnologies, whilst retaining the intended tNL function of specific tumor biomolecular recognition in vivo.

The field of tNLs as nanomedicines is rapidly expanding, with serious considerations being made to the multiparametric modulation of nanotherapeutics, in order to extend circulation half-lives, reduce clearance by the mononuclear phagocyte system, increase the efficiency of tumor delivery, enhance tumor permeability and address the caveat of biomarker heterogeneity, amongst several other advances [70-73]. By fully acknowledging the importance of all these directions, specifically within the context of molecular targeted nanotherapeutics, we propose a conceptual paradigm shift in NIRA-tNL development by using a rational and modular approach for the nanofabrication of complex, multimodal tNLs as detailed in Fig. 4.

By taking into consideration multiple interdependent factors that influence molecular specificity of NIRA-tNLs, such as ligand orientation and density, we propose a logical iterative synthesis process that directs and focuses the development of molecular targeted nanoconstructs towards the goal of obtaining true molecular specificity in vivo. Modulation of these various nanoparticle parameters should always work synergistically with attempts to achieve optimal in vivo specificity. As discussed earlier, optimal in vivo specificity can be defined as the concentration of in vivo receptor expression reported by a NIRA-tNL, which closely matches the actual concentration of the target receptor expressed by the same tumor. In the case of our EGFR-directed NIRA-tNLs, optimal in vivo specificity is defined as the in vivo concentration of EGFR that the tNL reports, which matches the in vivo concentration of EGFR that is reported by the benchmark standards, the natural ligand EGF and an EGFR affibody.

Our findings presented in this study using robust quantitative NIR molecular imaging show, for the first time, that large ligand-directed tNLs are capable of exhibiting true molecular specificity in vivo, which is quantifiable and distinct from tumor selectivity. We show that this can be done without the need for arduous, complex and unreliable tissue processing. The findings warrant a plethora of investigative avenues into the molecular specificity

of more complex, multimodal nanotherapeutics to improve treatment specificity. Such advancements in nanoparticle development promise to further reduce the adverse effects that conventional cancer treatments have on patient quality of life and could transform the strategies leveraged for the synthesis and evaluation of receptor-specific tNLs.

Figure 4. Conceptual overview of our proposed paradigm shift in the informed and iterative synthesis of receptor targeted nanoliposomes. This paradigm shift combines the multifaceted modulation of key and emerging parameters that are important for NIRA-tNL fabrication, with quantitation of in vivo molecular specificity of the constructs. Iterative engineering processes will ultimately implement rigorous pre-clinical development routines that enable the intelligent and informed development of NIRA-tNLs, minimizing failures and providing valuable modular insights for achieving in vivo molecular specificity.

In light of the persisting enigma regarding tNL receptor interactions in vivo, our findings strongly encourage the continued adoption of ligand targeted nanoliposomes, given that quantifiable in vivo specificity is redefined as the principal endpoint for their engineering, while continuously striving to

17

improve the efficiency of their delivery to tumors. This study thus provides a pivotal framework to an already rapidly advancing field and offers key insights into quantifying and evaluating the specificity of molecular targeted nanoliposomes in vivo.

Acknowledgements

We thank Drs. Akilan Palanisami and Mans Broekgaarden for insightful discussions and Drs. Jie Zhao and Danian Cao for their excellent technical expertise. This work was supported by the National Institutes of Health (K99CA215301 and R00CA215301 to G.O.; R37CA212187 to K.S.; and P01CA084203, R01CA156177, R01CA160998, R21CA220143 to T.H.); the Bullock-Wellman Fellowship (G.O.), Science Foundation Ireland and the Irish Research Council (S.C.), and the American Society of Lasers in Surgery and Medicine Research Grant (S.M.).

Electronic Supplementary Material: Supplementary material (supporting data) is available in the online version of this article at http://dx.doi.org/10.1007/10.1007/s12274-***-****-* (automatically inserted by the publisher) and is accessible free of charge

References [1] Maeda, H.;Nakamura, H.; Fang, J. The EPR effect for

macromolecular drug delivery to solid tumors: Improvement of tumor uptake, lowering of systemic toxicity, and distinct tumor imaging in vivo. Adv. Drug Deliv. Rev. 2013, 65, 71-79.

[2] Obaid, G.;Bano, S.;Mallidi, S.;Broekgaarden, M.;Kuriakose, J.;Silber, Z.;Bulin, A. L.;Wang, Y.;Mai, Z.;Jin, W.;Simeone, D.; Hasan, T. Impacting Pancreatic Cancer Therapy in Heterotypic in Vitro Organoids and in Vivo Tumors with Specificity-Tuned, NIR-Activable Photoimmunonanoconjugates: Towards Conquering Desmoplasia? Nano Lett. 2019, 10.1021/acs.nanolett.9b00859.

[3] Obaid, G.;Chambrier, I.;Cook, M. J.; Russell, D. A. Targeting the oncofetal Thomsen-Friedenreich disaccharide using jacalin-PEG phthalocyanine gold nanoparticles for photodynamic cancer therapy. Angew. Chem. Int. Ed. Engl. 2012, 51, 6158-6162.

[4] Obaid, G.;Chambrier, I.;Cook, M. J.; Russell, D. A. Cancer targeting with biomolecules: a comparative study of photodynamic therapy efficacy using antibody or lectin conjugated phthalocyanine-PEG gold nanoparticles. Photochem. Photobiol. Sci. 2015, 14, 737-747.

[5] Bano, S.;Obaid, G.;Swain, J. W.;Yamada, M.;Pogue, B. W.;Wang, K.; Hasan, T. NIR Photodynamic Destruction of PDAC and HNSCC Nodules Using Triple-Receptor-Targeted

Photoimmuno-Nanoconjugates: Targeting Heterogeneity in Cancer. Journal of Clinical Medicine 2020, 9, 2390.

[6] Schmidt, M. M.; Wittrup, K. D. A modeling analysis of the effects of molecular size and binding affinity on tumor targeting. Mol. Cancer Ther. 2009, 8, 2861-2871.

[7] Han, H.; Davis, M. E. Single-antibody, targeted nanoparticle delivery of camptothecin. Mol. Pharmaceutics 2013, 10, 2558-2567.

[8] Mamot, C.;Drummond, D. C.;Noble, C. O.;Kallab, V.;Guo, Z.;Hong, K.;Kirpotin, D. B.; Park, J. W. Epidermal growth factor receptor–targeted immunoliposomes significantly enhance the efficacy of multiple anticancer drugs in vivo. Cancer Res. 2005, 65, 11631-11638.

[9] Bartlett, D. W.;Su, H.;Hildebrandt, I. J.;Weber, W. A.; Davis, M. E. Impact of tumor-specific targeting on the biodistribution and efficacy of siRNA nanoparticles measured by multimodality in vivo imaging. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 15549-15554.

[10] Kirpotin, D. B.;Drummond, D. C.;Shao, Y.;Shalaby, M. R.;Hong, K.;Nielsen, U. B.;Marks, J. D.;Benz, C. C.; Park, J. W. Antibody targeting of long-circulating lipidic nanoparticles does not increase tumor localization but does increase internalization in animal models. Cancer Res. 2006, 66, 6732-6740.

[11] Kotagiri, N.;Sudlow, G. P.;Akers, W. J.; Achilefu, S. Breaking the depth dependency of phototherapy with Cerenkov radiation and low-radiance-responsive nanophotosensitizers. Nat. Nanotechnol. 2015, 10, 370-379.

[12] Dai, Q.;Wilhelm, S.;Ding, D.;Syed, A. M.;Sindhwani, S.;Zhang, Y.;Chen, Y. Y.;MacMillan, P.; Chan, W. C. Quantifying the Ligand-Coated Nanoparticle Delivery to Cancer Cells in Solid Tumors. ACS nano 2018, 12, 8423-8435.

[13] Obaid, G.;Broekgaarden, M.;Bulin, A.-L.;Huang, H.-C.;Kuriakose, J.;Liu, J.; Hasan, T. Photonanomedicine: a convergence of photodynamic therapy and nanotechnology. Nanoscale 2016, 8, 12471-12503.

[14] Heukers, R.;en Henegouwen, P. M. v. B.; Oliveira, S. Nanobody–photosensitizer conjugates for targeted photodynamic therapy. Nanomedicine: NBM 2014, 10, 1441-1451.

[15] Nagaya, T.;Nakamura, Y.;Sato, K.;Harada, T.;Choyke, P. L.;Hodge, J. W.;Schlom, J.; Kobayashi, H. Near infrared photoimmunotherapy with avelumab, an anti-programmed death-ligand 1 (PD-L1) antibody. Oncotarget 2017, 8, 8807.

[16] Soukos, N. S.;Hamblin, M. R.;Keel, S.;Fabian, R. L.;Deutsch, T. F.; Hasan, T. Epidermal growth factor receptor-targeted immunophotodiagnosis and photoimmunotherapy of oral precancer in vivo. Cancer Res. 2001, 61, 4490-4496.

[17] Kotagiri, N.;Cooper, M. L.;Rettig, M.;Egbulefu, C.;Prior, J.;Cui, G.;Karmakar, P.;Zhou, M.;Yang, X.;Sudlow, G.;Marsala, L.;Chanswangphuwana, C.;Lu, L.;Habimana-Griffin, L.;Shokeen, M.;Xu, X.;Weilbaecher, K.;Tomasson, M.;Lanza, G.;DiPersio, J. F.; Achilefu, S. Radionuclides transform chemotherapeutics into phototherapeutics for precise

18

treatment of disseminated cancer. Nat. Commun. 2018, 9, 275.

[18] Skovsen, E.;Snyder, J. W.;Lambert, J. D.; Ogilby, P. R. Lifetime and diffusion of singlet oxygen in a cell. J. Phys. Chem. B 2005, 109, 8570-8573.

[19] Kessel, D. Photodynamic therapy: Promotion of efficacy by a sequential protocol. J. Porphyr. Phthalocyanines 2016, 20, 302-306.

[20] Rizvi, I.;Nath, S.;Obaid, G.;Ruhi, M. K.;Moore, K.;Bano, S.;Kessel, D.; Hasan, T. A Combination of Visudyne and a Lipid-anchored Liposomal Formulation of Benzoporphyrin Derivative Enhances Photodynamic Therapy Efficacy in a 3D Model for Ovarian Cancer. Photochem. Photobiol. 2019, 95, 419-429.

[21] Hudson, R.;Carcenac, M.;Smith, K.;Madden, L.;Clarke, O. J.;Pelegrin, A.;Greenman, J.; Boyle, R. W. The development and characterisation of porphyrin isothiocyanate-monoclonal antibody conjugates for photoimmunotherapy. Br. J. Cancer 2005, 92, 1442-1449.

[22] Heukers, R.;van Bergen en Henegouwen, P. M.; Oliveira, S. Nanobody-photosensitizer conjugates for targeted photodynamic therapy. Nanomedicine 2014, 10, 1441-1451.

[23] Rizvi, I.;Obaid, G.;Bano, S.;Hasan, T.; Kessel, D. Photodynamic therapy: Promoting in vitro efficacy of photodynamic therapy by liposomal formulations of a photosensitizing agent. Lasers Surg. Med. 2018, 50, 499-505.

[24] Li, J.;Cui, D.;Huang, J.;He, S.;Yang, Z.;Zhang, Y.;Luo, Y.; Pu, K. Organic Semiconducting Pro-nanostimulants for Near-Infrared Photoactivatable Cancer Immunotherapy. Angew Chem Int Ed Engl 2019, 58, 12680-12687.

[25] Cui, D.;Huang, J.;Zhen, X.;Li, J.;Jiang, Y.; Pu, K. A Semiconducting Polymer Nano-prodrug for Hypoxia-Activated Photodynamic Cancer Therapy. Angew Chem Int Ed Engl 2019, 58, 5920-5924.

[26] Obaid, G.;Broekgaarden, M.;Bulin, A. L.;Huang, H. C.;Kuriakose, J.;Liu, J.; Hasan, T. Photonanomedicine: a convergence of photodynamic therapy and nanotechnology. Nanoscale 2016, 8, 12471-12503.

[27] ClinicalTrials.gov Identifier: NCT03076372 Accessed July 2020.

[28] ClinicalTrials.gov Identifier: NCT02369198 Accessed July 2020.

[29] ClinicalTrials.gov Identifier: NCT01304797 Accessed July 2020.

[30] ClinicalTrials.gov Identifier: NCT01517464 Accessed July 2020.

[31] Harari, P. M.; Huang, S.-M. Head and neck cancer as a clinical model for molecular targeting of therapy: combining EGFR blockade with radiation. International Journal of Radiation Oncology* Biology* Physics 2001, 49, 427-433.

[32] Lee, H. J.;Seo, A. N.;Kim, E. J.;Jang, M. H.;Kim, Y. J.;Kim, J. H.;Kim, S. W.;Ryu, H. S.;Park, I. A.;Im, S. A.;Gong, G.;Jung, K. H.;Kim, H. J.; Park, S. Y. Prognostic and predictive values of EGFR overexpression and EGFR copy number alteration in HER2-positive breast cancer. Br J Cancer 2015, 112, 103-111.

[33] Hashmi, A. A.;Naz, S.;Hashmi, S. K.;Irfan, M.;Hussain, Z. F.;Khan, E. Y.;Asif, H.; Faridi, N. Epidermal growth factor receptor (EGFR) overexpression in triple-negative breast cancer: association with clinicopathologic features and prognostic parameters. Surgical and Experimental Pathology 2019, 2, 6.

[34] Shinojima, N.;Tada, K.;Shiraishi, S.;Kamiryo, T.;Kochi, M.;Nakamura, H.;Makino, K.;Saya, H.;Hirano, H.; Kuratsu, J.-i. Prognostic value of epidermal growth factor receptor in patients with glioblastoma multiforme. Cancer research 2003, 63, 6962-6970.

[35] Tichauer, K. M.;Samkoe, K. S.;Sexton, K. J.;Hextrum, S. K.;Yang, H. H.;Klubben, W. S.;Gunn, J. R.;Hasan, T.; Pogue, B. W. In vivo quantification of tumor receptor binding potential with dual-reporter molecular imaging. Mol. Imaging Biol. 2012, 14, 584-592.

[36] Samkoe, K. S.;Tichauer, K. M.;Gunn, J. R.;Wells, W. A.;Hasan, T.; Pogue, B. W. Quantitative in vivo immunohistochemistry of epidermal growth factor receptor using a receptor concentration imaging approach. Cancer Res. 2014, 74, 7465-7474.

[37] Rasband, W. S. ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA, 1997-2016, https://imagej.nih.gov/ij/,.

[38] Tichauer, K. M.;Samkoe, K. S.;Sexton, K. J.;Gunn, J. R.;Hasan, T.; Pogue, B. W. Improved tumor contrast achieved by single time point dual-reporter fluorescence imaging. J. Biomed. Opt. 2012, 17, 066001.

[39] Maruyama, K.;Takizawa, T.;Yuda, T.;Kennel, S. J.;Huang, L.; Iwatsuru, M. Targetability of novel immunoliposomes modified with amphipathic poly(ethylene glycol)s conjugated at their distal terminals to monoclonal antibodies. Biochim. Biophys. Acta. 1995, 1234, 74-80.

[40] Harding, J. A.;Engbers, C. M.;Newman, M. S.;Goldstein, N. I.; Zalipsky, S. Immunogenicity and pharmacokinetic attributes of poly (ethylene glycol)-grafted immunoliposomes. Biochim. Biophys. Acta, Biomembr. 1997, 1327, 181-192.

[41] Shokeen, M.;Pressly, E. D.;Hagooly, A.;Zheleznyak, A.;Ramos, N.;Fiamengo, A. L.;Welch, M. J.;Hawker, C. J.; Anderson, C. J. Evaluation of multivalent, functional polymeric nanoparticles for imaging applications. ACS Nano 2011, 5, 738-747.

[42] Gabizon, A.;Shmeeda, H.; Barenholz, Y. Pharmacokinetics of pegylated liposomal Doxorubicin: review of animal and human studies. Clin. Pharmacokinet. 2003, 42, 419-436.

[43] Corbet, C.; Feron, O. Tumour acidosis: from the passenger to the driver's seat. Nat Rev Cancer 2017, 17, 577-593.

[44] Patel, D.;Lahiji, A.;Patel, S.;Franklin, M.;Jimenez, X.;Hicklin, D. J.; Kang, X. Monoclonal antibody cetuximab binds to and down-regulates constitutively activated epidermal growth factor receptor vIII on the cell surface. Anticancer Res. 2007, 27, 3355-3366.

[45] Semple, S. C.;Chonn, A.; Cullis, P. R. Interactions of liposomes and lipid-based carrier systems with blood proteins: Relation to clearance behaviour in vivo. Adv. Drug Delivery Rev. 1998, 32, 3-17.

19

[46] Gonzalez‐Quintela, A.;Alende, R.;Gude, F.;Campos, J.;Rey, J.;Meijide, L.;Fernandez‐Merino, C.; Vidal, C. Serum levels of immunoglobulins (IgG, IgA, IgM) in a general adult population and their relationship with alcohol consumption, smoking and common metabolic abnormalities. Clin. Exp. Immunol. 2008, 151, 42-50.

[47] Dyes, I. I. Doc # 988-15067 12/14. [48] Peng, X.;Chen, H.;Draney, D. R.;Volcheck,

W.;Schutz-Geschwender, A.; Olive, D. M. A nonfluorescent, broad-range quencher dye for Forster resonance energy transfer assays. Anal. Biochem. 2009, 388, 220-228.

[49] Microscale, I. R. P. L. K.-. Doc # 988-15067 12/14. [50] Lin, H.;Zhang, R.;Gunn, J. R.;Esipova, T.

V.;Vinogradov, S.;Gladstone, D. J.;Jarvis, L. A.; Pogue, B. W. Comparison of Cherenkov excited fluorescence and phosphorescence molecular sensing from tissue with external beam irradiation. Phys. Med. Biol. 2016, 61, 3955-3968.

[51] Obaid, G.;Spring, B. Q.;Bano, S.; Hasan, T. Activatable clinical fluorophore-quencher antibody pairs as dual molecular probes for the enhanced specificity of image-guided surgery. J.Biomed. Opt. 2017, 22, 1-6.

[52] Spring, B. Q.;Bryan Sears, R.;Zheng, L. Z.;Mai, Z.;Watanabe, R.;Sherwood, M. E.;Schoenfeld, D. A.;Pogue, B. W.;Pereira, S. P.;Villa, E.; Hasan, T. A photoactivable multi-inhibitor nanoliposome for tumour control and simultaneous inhibition of treatment escape pathways. Nat. Nanotechnol. 2016, 11, 378-387.

[53] Pastore, S.;Mascia, F.;Mariani, V.; Girolomoni, G. The epidermal growth factor receptor system in skin repair and inflammation. J. Invest. Dermatol. 2008, 128, 1365-1374.

[54] Minder, P.;Zajac, E.;Quigley, J. P.; Deryugina, E. I. EGFR regulates the development and microarchitecture of intratumoral angiogenic vasculature capable of sustaining cancer cell intravasation. Neoplasia 2015, 17, 634-649.

[55] van Cruijsen, H.;Giaccone, G.; Hoekman, K. Epidermal growth factor receptor and angiogenesis: Opportunities for combined anticancer strategies. Int. J. Cancer 2005, 117, 883-888.

[56] Amin, D. N.;Hida, K.;Bielenberg, D. R.; Klagsbrun, M. Tumor endothelial cells express epidermal growth factor receptor (EGFR) but not ErbB3 and are responsive to EGF and to EGFR kinase inhibitors. Cancer research 2006, 66, 2173-2180.

[57] Tundidor, Y.;García-Hernández, C. P.;Pupo, A.;Infante, Y. C.; Rojas, G. Delineating the functional map of the interaction between nimotuzumab and the epidermal growth factor receptor. In mabs; Taylor & Francis, 2014; pp 1013-1025.

[58] Sato, J. D.;Kawamoto, T.;Le, A. D.;Mendelsohn, J.;Polikoff, J.; Sato, G. H. Biological effects in vitro of monoclonal antibodies to human epidermal growth factor receptors. Mol Biol Med 1983, 1, 511-529.

[59] Hoeben, B. A.;Molkenboer‐Kuenen, J. D.;Oyen, W. J.;Peeters, W. J.;Kaanders, J. H.;Bussink, J.; Boerman, O. C. Radiolabeled cetuximab: Dose optimization for epidermal growth factor receptor imaging in a head‐

and ‐ neck squamous cell carcinoma model. International journal of cancer 2011, 129, 870-878.

[60] He, X.;Cruz, J. L.;Joseph, S.;Pett, N.;Chew, H. Y.;Tuong, Z. K.;Okano, S.;Kelly, G.;Veitch, M.; Simpson, F. Characterization of 7A7, an anti-mouse EGFR monoclonal antibody proposed to be the mouse equivalent of cetuximab. Oncotarget 2018, 9, 12250.

[61] Yuan, F.;Leunig, M.;Huang, S. K.;Berk, D. A.;Papahadjopoulos, D.; Jain, R. K. Mirovascular permeability and interstitial penetration of sterically stabilized (stealth) liposomes in a human tumor xenograft. Cancer Res. 1994, 54, 3352-3356.

[62] Drummond, D. C.;Meyer, O.;Hong, K.;Kirpotin, D. B.; Papahadjopoulos, D. Optimizing liposomes for delivery of chemotherapeutic agents to solid tumors. Pharmacol. Rev. 1999, 51, 691-744.

[63] Juweid, M.;Neumann, R.;Paik, C.;Perez-Bacete, M. J.;Sato, J.;van Osdol, W.; Weinstein, J. N. Micropharmacology of monoclonal antibodies in solid tumors: direct experimental evidence for a binding site barrier. Cancer Res. 1992, 52, 5144-5153.

[64] Baker, J. H.;Lindquist, K. E.;Huxham, L. A.;Kyle, A. H.;Sy, J. T.; Minchinton, A. I. Direct visualization of heterogeneous extravascular distribution of trastuzumab in human epidermal growth factor receptor type 2 overexpressing xenografts. Clin. Cancer Res. 2008, 14, 2171-2179.

[65] Obaid, G.;Jin, W.;Bano, S.;Kessel, D.; Hasan, T. Nanolipid Formulations of Benzoporphyrin Derivative: Exploring the Dependence of Nanoconstruct Photophysics and Photochemistry on Their Therapeutic Index in Ovarian Cancer Cells. Photochem. Photobiol. 2019, 95, 364-377.

[66] Snyder, J. W.;Greco, W. R.;Bellnier, D. A.;Vaughan, L.; Henderson, B. W. Photodynamic therapy: a means to enhanced drug delivery to tumors. Cancer Res. 2003, 63, 8126-8131.

[67] Huang, H. C.;Rizvi, I.;Liu, J.;Anbil, S.;Kalra, A.;Lee, H.;Baglo, Y.;Paz, N.;Hayden, D.;Pereira, S.;Pogue, B. W.;Fitzgerald, J.; Hasan, T. Photodynamic Priming Mitigates Chemotherapeutic Selection Pressures and Improves Drug Delivery. Cancer Res. 2018, 78, 558-571.

[68] Overchuk, M.; Zheng, G. Overcoming obstacles in the tumor microenvironment: Recent advancements in nanoparticle delivery for cancer theranostics. Biomaterials 2018, 156, 217-237.

[69] Muchekehu, R.;Liu, D.;Horn, M.;Campbell, L.;Del Rosario, J.;Bacica, M.;Moskowitz, H.;Osothprarop, T.;Dirksen, A.;Doppalapudi, V.;Kaspar, A.;Pirie-Shepherd, S. R.; Coronella, J. The Effect of Molecular Weight, PK, and Valency on Tumor Biodistribution and Efficacy of Antibody-Based Drugs. Transl. Oncol. 2013, 6, 562-572.

[70] Zimpel, A.;Al Danaf, N.;Steinborn, B.;Kuhn, J.;Hohn, M.;Bauer, T.;Hirschle, P.;Schrimpf, W.;Engelke, H.;Wagner, E.;Barz, M.;Lamb, D. C.;Lachelt, U.; Wuttke, S. Coordinative Binding of Polymers to Metal-Organic Framework Nanoparticles for Control of Interactions at the Biointerface. ACS Nano 2019, 13, 3884-3895.

[71] Sun, H.;Su, J.;Meng, Q.;Yin, Q.;Chen, L.;Gu, W.;Zhang, P.;Zhang, Z.;Yu, H.;Wang, S.; Li, Y.

20

Cancer-Cell-Biomimetic Nanoparticles for Targeted Therapy of Homotypic Tumors. Adv. Mater. 2016, 28, 9581-9588.

[72] Chen, Z.;Zhao, P.;Luo, Z.;Zheng, M.;Tian, H.;Gong, P.;Gao, G.;Pan, H.;Liu, L.;Ma, A.;Cui, H.;Ma, Y.; Cai, L. Cancer Cell Membrane-Biomimetic Nanoparticles for Homologous-Targeting Dual-Modal Imaging and Photothermal Therapy. ACS Nano 2016, 10, 10049-10057.

[73] Lazarovits, J.;Chen, Y. Y.;Song, F.;Ngo, W.;Tavares, A. J.;Zhang, Y. N.;Audet, J.;Tang, B.;Lin, Q.;Tleugabulova, M. C.;Wilhelm, S.;Krieger, J. R.;Mallevaey, T.; Chan, W. C. W. Synthesis of Patient-Specific Nanomaterials. Nano Lett. 2019, 19, 116-123.

21

Electronic Supplementary Material

Is Tumor Cell Specificity Distinct from Tumor Selectivity In Vivo?: A Quantitative NIR Molecular Imaging Analysis of Nanoliposome Targeting

Girgis Obaid1,2†*, Kimberley Samkoe3*, Kenneth Tichauer4, Shazia Bano1, Yeonjae Park3, Zachary Silber1, Sassan Hodge3, Susan Callaghan1, Mina Guirguis,2 Srivalleesha Mallidi1†, Brian Pogue3, and Tayyaba Hasan1,5 () 1 Wellman Center for Photomedicine, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts 02114, U.S. 2 Department of Bioengineering, University of Texas at Dallas, Richardson, Texas 75080, U.S. 3 Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire 037551, U.S. 4 Armour College of Engineering, Illinois Institute of Technology, Chicago, Illinois 60616, U.S. 5 Division of Health Sciences and Technology, Harvard University and Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, U.S. † Girgis Obaid present address: Department of Bioengineering, University of Texas at Dallas, Richardson, Texas 75080, U.S. † Srivalleesha Mallidi present address: Department of Biomedical Engineering, Tufts University, Medford, Massachusetts 02155, U.S. * Both authors contributed equally to the study Received: day month year / Revised: day month year / Accepted: day month year (automatically inserted by the publisher) ©The Author(s) 2010. This article is published with open access at Springerlink.com

———————————— Address correspondence to Tayyaba Hasan, [email protected]

22

23

Figure S-1. a) Fluorescence emission spectra of free IRDye 680RD and free IRDye 800CW in PBS normalized to the molar concentrations of the respective dyes. b) Raw fluorescence emission spectra of free IRDye 680RD and nano-Cet-680 at a 10 nM dye equivalent concentration in PBS (b) and free IRDye 800CW and nano-IgG-800 at a 100 nM dye equivalent concentration in PBS (c). d) Fluorescence emission profiles of the two nanoconstructs normalized to the respective dye concentrations. e) Fluorescence emission profiles of the two nanoconstructs and their respective free IRDyes normalized to the respective maxima of the free dyes in PBS. Samples containing IRDye 680RD were excited at 660 nm and samples containing IRDye 800CW were excited at 760 nm; all emission spectra are averages of 3 measurements. f) Binding patterns of Cet and irrelevant IgG derivatized with both NHS-PEG4-N3 and Alexa Fluor 488-NHS to U87 cells verifies that the azido and fluorescently derivatized Cet exhibits specific binding to human EGFR expressed on U87 cells with a Kd of 0.106 nM, while the irrelevant IgG sham used in this study exhibits no specific binding. Neither the NIRA-tNL nor the sham mimetic exhibit changes in the mean z-average diameters (g) or in the instability factors (h; product of diameter and polydispersity index) for up to 6 weeks in storage at 4°C. The fluorescence intensity of the IRDye 680RD nanoconstructs (i) and IRDye 800CW nanoconstructs (j) are unaltered in response to varying pH levels.

24

Figure S-2. a) H&E staining (top) and immunohistochemistry staining of EGFR (bottom, horseradish peroxidase, brown) of U251, U87 and 9L tumors used in this study. The EGFR staining patterns confirm the high EGFR expression patterns of U251 tumors, the moderate expression of EGFR in U87 tumors and the negligible EGFR expression in 9L tumors, that is consistent with cellular expression levels determined by flow cytometry. Tumor margins in the EGFR immunohistochemistry images (dotted line) were traced manually using the H&E images as a template. b) Quantitation of relative EGFR expression from immunohistochemistry sections from a total of 5 mice per tumor type. c) A linear relationship exists between relative tumor EGFR expression levels ex vivo and the in vitro EGFR expression levels. (Values are mean ± S.E.M., n = 5 biological replicates. Statistical significance was calculated using ANOVA with a Tukey post-test; ** = P< 0.011, **** = P< 0.0001.)

25

Figure S-3. Representative longitudinal fluorescence traces of the nano-Cet-680 NIRA-tNL in U251, U87 and 9L glioblastoma tumors (a) and the nano-IgG-800 sham mimetic in the same U251, U87 and 9L glioblastoma tumors (b). Tumor and Skin nano-IgG-800 sham mimetic signals are normalized to t = 0 h to correct for differences in brightness in vivo between the two nanoconstructs. Representative longitudinal fluorescence traces of the nano-IgG-800 sham mimetic in the tumor and in distant skin regions in mice bearing (c), U87 (d), and 9L (e) tumors.

26

Figure S-4. a) Combined plots of the binding potential values derived from in vivo quantitative NIR molecular imaging in all tumors and vicinal skin. b) Correlations between EGFR receptor numbers per cell (flow cytometry) and calculated binding potentials. (Values are mean ± S.E.M., biological replicates n = 4 (U251), 3 (U87) and 5 (9L). Pearson’s r correlation analysis was performed with a two-tailed test for statistical significance of correlation. Statistical significance was calculated using ANOVA with a Tukey post-test; * = P< 0.013, ** = P<0.0074, *** = P<0.0004, **** = P< 0.0001.) c) Binding Potentials of U87 tumors derived using the non-invasive skin correction method, comparing it to measurements made with skin removed (Mean ± SD, Two-tailed t test, no significance, biological replicates n = 3). d) Pooled binding potentials of the skin from mice bearing U251 tumors, U87 tumors and 9L tumors (Values are mean ± S.D., biological replicates n = 12).

27

Figure S-5. a) Total NIRA-tNL and sham mimetic delivery (24h-72h) in EGFR-null 9L tumors. b) Total delivery of the nano-IgG-800 sham mimetic in U251, U87 and 9L tumors. Whole animal biodistribution of the NIRA-tNL 24h following administration in mice carrying U251 (c), U87 (d), and 9L (e) tumors. f) Optically corrected NIRA-tNL tumor delivery underrepresented NIRA-tNL specificity for U87 tumors at 24h. Values are mean ± S.E.M., statistical significance was calculated using a two-tailed t test for (a) and a One-Way ANOVA with a Tukey post-test for (b and f).

28

Figure S-6. a) Cross-sectional 21- µm resolution fluorescence images of all three tumor types 24h following administration of NIRA-tNLs showing that the construct accumulates at the periphery of the high-EGFR expressing U251 and U87 tumors, and not in EGFR-null 9L tumors. b) Three-dimensional intensity surface plots of tumor cross sections show that the NIRA-tNLs penetrate the tumor cores more efficiently as the tumor EGFR expression levels decreases. (x and y axes correspond to spatial dimensions in pixels, z axes correspond to fluorescence intensity dimensions in arbitrary units.)

29

Figure S-7. a) in situ ultrasound and multiwavelength photoacoustic imaging of the nano-Cet-680 NIRA-tNL and the nano-IgG-800 sham mimetic in a U87 tumor from a live animal. The green arrows highlight the nano-IgG-800 sham mimetic in the tumor. Tumor delivery (b) and cross-sectional tumor microdistribution of the nano-IgG-800 sham mimetic in all three tumor types 24 h following administration.