Drug delivery, biodistribution and anti-EGFR activity: theragnostic nanoparticles for simultaneous in vivo delivery of tyrosine kinase inhibitors and kinase activity biosensors

In vivo delivery of small molecule therapeutics to cancer cells, assessment of the selectivity of administration, and measuring the efficacity of the drug in question at the molecule level, are important ongoing challenges in developing new classes of cancer chemotherapeutics. One approach that has the potential to provide targeted delivery, tracking of biodistribution and readout of efficacy, is to use multimodal theragnostic nanoparticles to deliver the small molecule therapeutic. In this paper, we report the development of targeted theragnostic lipid/peptide/DNA lipopolyplexes. These simultaneously deliver an inhibitor of the EGFR tyrosine kinase, and plasmid DNA coding for a Crk-based biosensor, Picchu-X, which when expressed in the target cells can be used to quantify the inhibition of EGFR in vivo in a mouse colorectal cancer xenograft model. Reversible bioconjugation of a known analogue of the tyrosine kinase inhibitor Mo-IPQA to a cationic peptide, and co-formulation with peptides containing both EGFR-binding and cationic sequences, allowed for good levels of inhibitor encapsulation with targeted delivery to LIM1215 colon cancer cells. Furthermore, high levels of expression of the Picchu-X biosensor in the LIM1215 cells in vivo allowed us to demonstrate, using fluorescence lifetime microscopy (FLIM)-based biosensing, that EGFR activity can be successfully suppressed by the tyrosine kinase inhibitor, released from the lipopolyplexes. Finally, we measured the biodistribution of lipopolyplexes containing 125I-labelled inhibitors and were able to demonstrate that the lipopolyplexes gave significantly higher drug delivery to the tumors compared with free drug.


Introduction
Current challenges in next-generation medicine have stimulated the rapid development of theragnostic agents. These are of increasing importance for diseases such as cancer, where no two patients will have exactly the same biomarkers and oncogenic mutations, and where most currently available therapeutic agents have limited target selectivity, poor localization, and undesirable side-effects. Theragnostic agents can be based on small molecules conjugated to a targeting moiety, such as an antibody; on engineered mammalian cells; and on various types of nanoparticles (e.g. iron oxide, gold, polymeric or liposomal). 1 Whilst all of these have limitations, 2 liposomebased nanoparticles have several key advantages. These include the ability to design multimodal nanoparticles with several functionalities contained in, or attached to, a single liposome: rapid cellular uptake; a wide range of cellular compatibilities and low toxicity; and long circulating half-life combined with eventual biodegradability. 3 Many theragnostic agents have now been developed which are capable of diagnosing, through imaging, the presence of cancerous cells and then delivering a precisely targeted therapeutic intervention. 4 Whilst these approaches can be used to measure biodistribution and accumulation of small molecule drugs at the tumor, none of the theragnostic agents reported to date can quantify the effectiveness of the therapeutic at the molecular level in vivo. This is particularly important in targeted cancer therapy. Although diagnostic assays are available to determine protein expression status for some types of cancer, for many tumor types there are either no validated assays, or the expression status of the protein cannot in practice be correlated with the clinical response to the drug. 5 The development of theragnostic nanoparticles which would improve drug delivery, whilst also allowing both the biodistribution and response to the drug to be measured, would be a major step towards the development of more effective cancer therapies, tailored towards the driving molecular phenotype of an individual tumor.
Epidermal growth factor receptor (EGFR) is found to be overexpressed and/or constitutively activated in a variety of tumors including breast, lung and colon 6 and is a validated target for drug development. The EGFR family 7 consists of four members (EGFR, HER2, HER3 and HER4) with more than 10 different ligands 8 able to activate downstream signaling, leading to growth proliferation and inhibition of apoptosis, angiogenesis and metastasis. Several therapeutic strategies targeting EGFR have been devised. In particular, small molecule inhibitors of the EGFR intracellular tyrosine kinase activity (TKI), and antibodies blocking ligand interaction have been developed and tested in preclinical and clinical studies. 9,10 Unfortunately, the factors that predict clinical response to anti-EGFR agents remain unclear, and the success rate for these reagents in the clinic is low. Expression of EGFR does not correlate with efficacy of the EGFR antibody cetuximab 11 and the use of the antibody is limited by toxicity. 12 In recent clinical studies of EGFR-expressing breast 13 and colorectal 14 cancer patients, response rates to EGFR-targeted therapies were low and varied between cancer types, with the majority of patients eventually developing resistance to these reagents. It is clear that in order for these strategies to be successful, much more research is needed to understand how EGFR inhibitors and antibodies mechanistically function in vivo at both the cellular and molecular level. 15 A strategy for overcoming the poor pharmacokinetic profiles and low target selectivity of anticancer drugs is to use liposomes, liposome-based nanoparticles and polymersomes as delivery vehicles. 16 The polyethylene glycol (PEG)-shielded liposomal doxorubicin (DOXIL®) is clinically approved for the treatment of ovarian and breast cancer, with significantly longer circulation times and lower cardiotoxicity compared to the free drug, and several other liposomal drug formulations are in Phase II trials for other cancers. 17 At the preclinical level, radiolabeling 18 of liposome-encapsulated dasatinib with 18 F and liposome-encapsulated Mo-IPQA labeled 19 with 124 I have enabled quantification, demonstrating some of the highest tumor uptakes in vivo of TKI so far observed. Non-targeted liposomal drug formulations were believed to rely on the passive accumulation of these nanoparticles (100-200 nm) in tumors through the Enhanced Permeability and Retention (EPR) effect. However, many clinical studies have failed to show either increased nanoparticle accumulation in tumors or increased efficacy, and it is now clear that this effect is, at best, heterogeneous in humans. 20,21 Ligand-mediated cell-selective targeting offers additional advantages in terms of target cell specificity and cellular uptake. For example, immunoliposomes have been prepared using anti-EGFR antibodies as targeting moieties and were used to deliver drugs such as 5-fluorouracil, 22 cisplatin 23 and doxorubicin 24 to cancer cells in vivo, with moderate to significant improvements in cytotoxicity and cell selectivity compared to non-targeted liposomes. Likewise, polymersomes encapsulating plitidepsin and targeted with an anti-EGFR antibody showed greatly increased cytotoxicity and cellular uptake in colorectal cancer cell lines compared to an untargeted polymersome. 25 Liposomes targeted with EGFRbinding peptides have also been used to deliver doxorubicin 26 and cisplatin 27 effectively to cancer cells. Finally, lipoplexes targeted with anti-EGFR antibodies have been formulated to deliver siRNA to non-small cell lung cancer cell lines 28 and into EGFR-overexpressing hepatocellular carcinoma. 29 Surface modification was necessary for specific uptake of the siRNA into the cells and resulted in enhanced tumor accumulation when compared to untargeted lipoplexes.
We have previously reported 30 the development of EGFRtargeted lipopolyplexes for the delivery of pDNA encoding an EGFR biosensor. 31 This allowed us to monitor EGFR activity in a breast cancer model in vivo, before and after separate treatment with a TKI, Mo-IPQA. 30 As well as successfully demonstrating the inhibition of tyrosine phosphorylation by EGFR in vivo, and showing a significant degree of intratumoral heterogeneity in EGFR activity, our results showed that small molecule TKIs have a very poor uptake in cancer cells, and that the uptake is unselective. In this paper, we report the development of the first targeted, lipopolyplex-based theragnostic nanoparticle that improves drug delivery to cancer cells, allows quantification of the biodistribution, and also provides a readout of TKI mediated enzyme inhibition. As the reasons for the heterogeneity of response of colorectal cancer cells to EGFR-targeted therapies, and the rapid emergence of chemotherapy-resistant strains, are still poorly understood 15 these nanoparticles have been used for preliminary investigation of biodistribution and uptake in a murine xenograft colorectal cancer model. K 16 domain to condense pDNA; a cell-targeting sequence that is displayed at the surface of the nanoparticle; and a linker sequence, RVRR, that is a substrate for the endosomal enzyme furin. Mixtures of cationic and neutral lipids are used and are co-formulated with the neutral lipid dioleoylphosphatidylethanolamine (DOPE). The cationic lipids include both DOTMA 39 and the analogue DODEG4 32 which has a short ethylene glycol oligomer headgroup. These give shielded nanocomplexes that have better stability within the systemic circulation but do not impede cellular uptake.
The targeting peptide sequence at the surface of these lipopolyplexes binds to the cell of interest and mediates internalisation via receptor-mediated endocytosis: cleavage of the peptide at the RVRR sequence allows partial disassembly of the nanoparticle after internalisation, 33 and the neutral lipid DOPE promotes fusion with the endosomal membrane and mediates escape of the pDNA/peptide complex from the endosome. 40 These nanoparticles are able to transfect tumor cells in vivo with transfection efficiencies comparable to viral methods, 34 and lipopolyplexes generally have lower cytotoxicity than commercially available transfection reagents such as Lipofectamine2000. 37,38,41,42 We have previously carried out biophysical characterisation of these lipopolyplexes by a variety of techniques including TEM, 37 freeze-fracture EM 35 and cryo-EM. 36 These EM studies suggest that these lipopolyplexes form spherical particles with a dense internal core and an irregular outer layer.
To report the levels of EGFR activity in cells using these ternary lipopolyplexes, we used a genetically encoded CrkIIbased biosensor (Picchu-X) which undergoes conformational changes upon phosphorylation of tyrosine-221 by EGFR. 31,43 These changes are quantified by fluorescence resonance energy transfer (FRET) as monitored by fluorescence lifetime imaging microscopy (FLIM). We transfected this biosensor into EGFR-positive tumor cells using targeted ternary liposome-base nanoparticles bearing EGFR-binding peptides at the surface, and measured the response of these tumor cells to a TKI, Mo-IPQA, administered separately. 30 In order to achieve our goals of selectively delivering a TKI to colorectal cancer cells, imaging the inhibition of EGFR within the tumor and evaluating the biodistribution of the TKI, we designed enhanced multifunctional lipopolyplexbased nanoparticles (Fig. 1). These lipopolyplex formulations had to include: mixtures of lipids that would enhance the particle stability and provide steric shielding; peptide targeting sequences that would promote uptake from colorectal cancer cells; and a cationic peptide-TKI bioconjugate that would enable incorporation of high levels of both the TKI and the pDNA within the lipopolyplex. Additionally, we needed to develop synthetic strategies to ensure that the cationic peptide-TKI bioconjugate could be radiolabelled with 125 I, thus allowing the biodistribution of both the TKI and the lipopolyplex to be assessed.
With this next generation of theragnostic lipopolyplexes, in addition to DOPE, DOTMA and DODEG-4 we included cholesterol to further increase bilayer stability, 44 and Cy-5-DOTMA 30 to enable the uptake of the nanoparticles to the tumor cells to be easily monitored. In order to target the nanoparticles efficiently to cell lines over-expressing EGFR, we synthesized trifunctional peptides with DNA-binding linear K 16 , RVRR and cell-targeting sequences, as before. We have evaluated three peptide sequences for their ability to target colorectal cancer cells. Peptides P1 and P4 use the LARLLT (D4) sequence, 45 P2 and P5 the AEYLR 46 and P3 and P6 the YHWYGYTPQNVI (GE11) sequence. 47,48 These sequences had been previously reported to bind tightly to EGFR and, when displayed at the surface of nanoparticles, mediate uptake to EGFR+ tumor cells. Indeed, we have previously demonstrated that lipopolyplexes formulated with P1 effectively deliver pDNA encoding for a biosensor to HCC1954 and MDA-MB-231 breast cancer cell lines in vivo, 30 and that lipopolyplexes formulated with P1, P2 or P3 effectively transfect HCC1954 cells in vitro. 49 The major challenge to the design of these self-assembling nanoparticles was the co-encapsulation of the small molecule TKI with pDNA. Whilst several groups have developed multifunctional nanoparticles for combination therapy with small molecules and nucleic acids, 50-54 most of these approaches have focused on the co-delivery of siRNA, and many of the reported approaches are based on dendrimers, which have associated toxicities and result in the formation of heterogeneous mixtures. 55 In some cases where liposomal co-formulation of both nucleic acid and small molecule was attempted, aggregation of the liposome and leakage of the drug was seen. 56 In order to ensure that the TKI was successfully encapsulated in the self-assembling nanoparticle, and was retained within the nanoparticle after formulation, we sought to reversibly bioconjugate the TKI to one of the components of the nanoparticle. We therefore designed a cationic peptide-TKI bioconjugate 1 (Fig. 1) with a disulfide linkage between the two moieties. We based our design on the known EGFR TKI, F-[PEG 6 -IPQA], a modified form of Mo-IPQA 19,51 that had been developed previously as an 18 F PET radiotracer (Fig. 1a). 57 As the PEG moiety of this TKI is predicted to protrude from the active site of the enzyme 58 we reasoned that minimal modification to give a thiol-terminated PEG 6 would not adversely affect the potency or selectivity of this inhibitor, but would allow bioconjugation to K 16 Cys to give K 16 Cys-S-S-[PEG 6 -IPQA] 1. It was also envisaged that the K 16 cationic sequence of this bioconjugate would bind to pDNA with the same efficiency as the trifunctional K 16 -RVRR-[targeting sequence] or [targeting sequence]-RVRR-K 16 peptides P1-P6, and that we could therefore prepare lipopolyplexes using a mixture of 1 and peptides P1-P6. After internalization of the lipopolyplex, we predicted that the disulfide linkage would be cleaved in the reducing environment of the endosome, 59,60 releasing PEG 6 -IPQA into the cytoplasm.
This was subsequently reacted with the sodium salt of triphenylmethanethiol to yield the trityl protected thiol compound 4. Reduction of the nitro group was carried out using tin(II)chloride which gave the desired amino intermediate 5 in 61% yield. Using a previously reported procedure 58 compound 5 was reacted with acryloyl isobutyl carbonate, prepared in situ from acrylic acid and isobutyl chloroformate, to afford compound 6 in 57% yield. At this point, the trityl protecting group was replaced by NPys in a one pot reaction under acidic conditions using 2,2′-dithiobis(5-nitropyridine) 62 to afford 2 in 48% yield. In order to prevent competing Michael reaction between the thiol group and the unsaturated amide, it was necessary to maintain the thiol protection throughout the synthetic sequence, and to carry out the replacement of Trt by NPys at acidic pH. Finally, reaction 63 with the cationic peptide CysLys 16 7 in MeOH followed by HPLC gave 1 in 59% yield.

In vitro optimization of nanoparticle formulation
To test the ability of lipopolyplexes to deliver pDNA and inhibitor we used LIM1215 colon cancer cells. These are known to have a high level of EGFR expression and response to the TKI inhibitor, 64,65 and we had previously demonstrated that a reliable xenograft model could be established. 30 As a control, to check the response of the colorectal cancer cell lines to this class of TKI, we prepared an analogue of 1, with the disulfidelinked peptide replaced with an -OMe group. TKI analogue MeO-[PEG 6 -IPQA] 8 was synthesized from intermediate 3 via a similar route to that of 1 (ESI †) and demonstrated to act as a potent inhibitor of EGFR activation in SW48 cells (Fig. S1 †).
Initially, the sequence of the trifunctional peptides was investigated to see whether placing the targeting sequence at the N or the C terminal had any effect on lipopolyplex uptake and expression in colorectal cancer cells. Preliminary experiments (ESI, Fig. S2 †) suggested that the expression levels were lower for all peptides with the AEYLR sequence (P2, P5), and higher for LARLLT (P1, P4) and GE11 (P3, P6) in DLD1 colorectal cancer cells. In this cell line, the C-terminal LARLLT (P1) and N-terminal GE11 (P3) peptides yielded the highest expression levels. The best performing formulation (CLA1, using the P1 sequence) was tested in DLD1 and LIM1215 cells (Fig. S3 †). Expression levels of Picchu-X biosensor were far higher in LIM1215 cells, which were selected for further optimisation of the lipopolyplex formulations. Subsequently, the impact of plasmid DNA concentration and targeting peptide (P1-P6) on the transfection efficiency in LIM1215 cells were assessed ( Fig. S4 and S5 †). Although formulations with the C-terminal LARLLT peptide (P1) still showed the best transfection efficiency, formulations with N-terminal AEYLR (P5) and N-terminal GE11 (P6) also performed well in this cell line. In order to optimize the receptor binding and transfection properties of these lipopolyplexes, we then investigated formulations using different targeting peptides P1, P5, P6 and differing lipid compositions (Table 1), but initially without K 16 Cys-S-S-[PEG 6 -IPQA] 1. While we observed that all lipopolyplex formulations were equally internalized by the target cells, changes in the lipid composition and in the concentrations of bifunctional peptides P1, P5, P6, had an impact on the transfection of the Picchu biosensor.
In contrast to our initial screen, these studies demonstrated that the CLA1 formulation, formulated with P1 (Table 1) gave the maximum transfection efficiency ( Fig. 2A, S3 and S4 †), and the optimal final concentration of lipopolyplexes in the media was found to be 10 µM (based on the concentration of the lipids) (Fig. S6 †). We then prepared variant lipopolyplex formulations, based on CLA1 and including the EGFR inhibitor K 16 Cys-S-S-[PEG 6 -IPQA] 1 (Table 1). These lipopolyplex formulations (with increasing amount of inhibitor per particle) gave slightly decreased transfection efficiency ( Fig. 2B) compared to particles without the drug ( Fig. 2A), however the overall number of transfected tumor cells was not affected. For efficient perfusion through the leaky endothelial cells of the tumor, lipopolyplexes must be <150 nm in diameter. The lipopolyplex formulations were characterised using dynamic light scattering (DLS) and zeta potential measurements (ESI Table S1 †) and were found to have particle sizes within the ideal diameter for tumor delivery, reasonable polydispersities, and positive zeta potentials which would minimize particle aggregation and afford a reasonable degree of serum stability. We also carried out in vitro studies to check that the disulfide linker between K 16 Cys and PEG-IPQA was labile under pseudointracellular conditions (GSH/GSSG, HEPES, pH 7.2). 66 These indicated that under reducing conditions the linkage would be cleaved within 5 min (ESI Fig. S7 †). Next, we assessed the effectiveness of K 16 Cys-S-S-[PEG 6 -IPQA] 1 as an inhibitor of EGFR activity as a free therapeutic or within lipopolyplexes by utilizing FRET-FLIM and validating our results by western blot (WB) techniques. Activation of EGFR in cells treated with EGF (natural ligand for EGFR) was evident by decreasing lifetime of green fluorescent protein (GFP) in Picchu-X biosensor (Fig. 2c) and by appearance of phosphorylated EGFR band in WB image (Fig. 2d). In both cases pre-treatment with MeO-[PEG 6 -IPQA] 8 for 1 h abolished these changes: lifetime of GFP remained at baseline levels ( Fig. 2c) and the level of EGFR phosphorylation was significantly lower than in the sample without inhibitor (Fig. 2d, line 8 vs. line 2).
Furthermore, incorporation of K 16 Cys-S-S-[PEG 6 -IPQA] 1 within lipopolyplexes not only decreased EGFR activity to equally low levels as the free drug but was effective for longer duration (Fig. 2d, lines 3-7 vs. line 2). We observed EGFR inhibition when 1 was delivered within lipopolyplexes for up to 48 h, whereas free MeO-[PEG 6 -IPQA] 8 was effective only for 24 h (Fig. 2d). These results clearly reveal F1LA1 to be the optimal formulation for lipopolyplexes to achieve maximum transfection efficiency and prolong inhibition of EGFR in cells stimulated with ligand.

Radiolabeling and biodistribution
In order to assess the in vivo biodistribution of these liposome-based theragnostic nanoparticles, we required a [ 125 I]labelled analogue of 1. As we particularly wished to measure the biodistribution of the IPQA-peptide bioconjugate, we elected to attach the radiolabelled PEG-[ 125 I]IPQA moiety to K 16 Cys via a maleimide linkage, which would not be subject to reduction in vivo. The analogue K 16 Cys-SMal-[PEG 3 -[ 125 I]IPQA] 9 was therefore designed with a similar linker length between the K 16 DNA-binding peptide and IPQA analogue 1. PEG 3 -IPQA precursor 10 was prepared using a modification of the literature procedure 58 reported for PEG 6 -IPQA (ESI †). Similar methodology to that developed for the synthesis of 6 was used to access the key stannylated intermediate 16. Thus, the free Table 1 Lipopolyplex formulations. CLA lipopolyplexes are formed from C-terminal LARLLT bifunctional peptide P1 and varying amounts of cholesterol. NAE lipopolyplexes are formed from N-terminal AEYLR bifunctional peptide P5 and varying amounts of cholesterol. NGE lipopolyplexes are formed from N-terminal (GE11) bifunctional peptide P6 and varying amounts of cholesterol (Fig. 2a). The F1LA and F2LA formulations are based on the CLA1 and CLA2 lipopolyplexes, but with the addition of the cationic peptide-EGFR inhibitor bioconjugate K 16 Cys-S-S-[PEG 6 -IPQA] 1 at varying concentrations ( Fig. 2b

In vivo results
LIM1215 xenografts were established in standard immunocompromised CD1 nude mice (Materials and methods section). The biodistribution data (Fig. 3) showed that the free drug, radiolabelled MeO-[PEG 6 -[ 125 I]-IPQA] 8 was rapidly cleared via the hepatobiliary route with high uptake in gallbladder and intestines at 3 and 6 h.
Whereas   After gamma counting, the tumors were snap frozen, allowing simultaneous expression of biosensor and inhibition of EGFR activity by the liposome-encapsulated K 16 Cys-SMal-[PEG 3 -[ 125 I]-IPQA] 9. Fig. 4 shows representative images of tissue from mice injected with CLA1 lipopolyplexes, which only contain the pDNA encoding for the Picchu-X biosensor measurements of GFP revealed that EGFR activity can be successfully suppressed (Fig. 4) by PEG-IPQA released from the F1LA1 K 16 Cys-SMal-[PEG 3 -[ 125 I]-IPQA]-lipopolyplexes. Tissues treated with inhibitor had significantly higher GFP lifetime than control tissues (Fig. 4), which can also be seen on the images (shift from red-yellow to blue colours in pseudocolor map).

Conclusions
Liposome-based theragnostic nanoparticles hold considerable promise for the delivery of small molecule therapeutic drugs to tumours, and for determining their biodistribution and efficacy in vivo. A major barrier to the success of such approaches is the difficulty encountered in loading sufficient quantities of the drug into the liposome-based formulation, particularly where co-delivery of DNA or RNA is required. 67 In this work, we present a novel solution to this problem. The reversible bioconjugation of the small molecule therapeutic to a cationic peptide allows the co-formulation of both a TKI EGFR inhibitor and pDNA coding for a biosensor into a targeted lipopolyplex. The lipopolyplex is designed to disassemble on internalisation into the target cells. Our results indicate that release of the components from the endosome is accompanied by release of the TKI from the bioconjugate and inhibition of EGFR. This approach also allowed us to further investigate the basis for the enhanced circulation and better tumour selectivity afforded by nanoparticle delivery of small molecule therapeutics. The biodistribution data in Fig. 3 suggests different kinetics for either uptake of free drug vs. liposomes, and/or elimination of the free drug vs. drug encapsulated/shielded in liposome in the cells. This phenomenon will need further investigation of potential differences in trafficking of the drug within the cell post-uptake, both in terms of the subcellular compartments involved and kinetics of intracellular trafficking. However, we believe that the in vivo trafficking/clearance differences may be more significant and therefore relevant for the therapeutic effectiveness. The clearance in vivo of radiolabelled Mo-IPQA has recently been studied, 30,57 and has shown that this highly lipophilic inhibitor has limited tumour selectivity and undergoes significant hepatobiliary clearance. More hydrophilic PEGylated analogues, such as [ 18 F]F-PEG 6 -IPQA, 68 which formed the basis for the design of our own bioconjugate, show improved pharmacokinetics and quicker in vivo clearance, but still have little selectivity for tumour cells. 69,70 In contrast, liposomal formulations of the structurally-related, radiolabelled quinazoline TKI inhibitor, SKI-212243 had a significantly longer circulation time, increased overall tumour uptake and selectivity compared to the free drug. 19 Moreover, a different pattern of biodistribution was observed, suggesting that the liposomal formulation is cleared through a different route to the free drug.
Many potent TKI have been developed as anti-EGFR agents, and have shown excellent inhibitory properties in vitro.  However, translation to clinically viable agents, particularly for the treatment of colorectal cancer, has generally been unsuccessful, and the reasons for the poor clinical efficacy of these treatments remains elusive. Knowing how these therapies actually act at a molecular level when administered to patients would be a crucial step towards improving the success rate of such inhibitors. Our results demonstrate that lipopolyplexes are a successful delivery system for both an EGFR biosensor and a TKI inhibitor, which will allow surveillance of EGFR activity while testing the potency of such drugs in vivo. This approach in turn opens a new possibility for manipulating and monitoring the activity of EGFR in situ, which in turn will allow a deeper understanding of the dynamics of EGFR activation/inhibition and sensitivity of target cells to EGFR inhibition. Whilst we have previously demonstrated that it is possible to image EGFR activity in vivo utilizing an endoscopybased fluorescence lifetime imaging methodology, 71 this was based on an endoscope requiring physical contact with the area of interest to be imaged, and is a highly invasive procedure. Currently, we envisage that the most useful application of our methodology will be to investigate human derived xenografts to test the responsiveness of the tumour to different treatment regimes.
Finally, we have demonstrated that the reversible bioconjugation of TKIs to cationic peptides and their formulation into targeted nanoparticles results in greatly improved biodistribution relative to the free drug. This in turn will minimize toxicity due to off-target effects. Ultimately, we believe that this approach can be generalised. It will allow for the targeted codelivery of small molecule/toxic gene, small molecule/gene therapy, and small molecule/siRNA combinations, 72 opening up exciting possibilities for combined therapeutic delivery and sensing.

Liposome and lipopolyplex formulation
Lipopolyplex formulation for in vitro experiments. All lipid components were dissolved in CHCl 3 to a concentration of 1 mM. These stock solutions were mixed to obtain the desired lipid quantities. The solvent was then removed under reduced pressure to form a thin film, which was further dried on a high vacuum line for at least 1 h. The thin film was hydrated with sterilized water to the desired total lipid concentration, sonicated for 10 min, and used immediately. Lipopolyplexes used for in vitro experiments were formulated as described (including Cy5-DOTMA 2mol-%). For every 100 μL of sonicated liposome solution, first peptide (4-8 μg) and/or peptide-drug conjugate (8-32 µM) was added followed by plasmid DNA (2-4 μg) coding for the Picchu-X sensor. The samples were mixed and used immediately. This gave a total concentration of the lipids in the liposomes of 1.3 mM. The solvent in the lipid mixture was slowly removed via evaporation in vacuo to form a lipid thin film and dried for at least 1 h on a high vacuum line. The appropriate amount of distilled water was added to reach a concentration of 1.3 mM total lipid in water. The sample was sonicated for 5 min in a bath sonicator and subsequently on a probe tip sonicator over 10 min using a cycle of 10 s sonication at level 3 and 3 s cooling time. The lipid solution was kept on ice during the sonication process to avoid degradation of the lipid components and overheating of the solution. This procedure gives liposomes of <100 nm and surface charge around +20 meV or above. The liposomes were transferred into a Falcon tube and to this solution was added targeting peptide P1 (K 16 -RVRR-LARLLT (1 mM stock)) to a final concentration of 170 µM followed by the drug peptide bioconjugate (K 16 Cys-S-S-[PEG 6 -IPQA] 1 (0.7 mM stock)) to a final concentration of 100 µM. The liposome peptide solution was then diluted to 3.5 mL. 680 µg Picchu-X plasmid DNA (1.68 µg µL −1 stock, 404 µL) was added to a Falcon tube and diluted to 3.5 mL. The DNA solution was then slowly (dropwise but steadily) added to the liposome-peptide solution while constantly swirling the solution, in order to avoid the concentration of DNA and peptide gets locally too high and precipitates. Once the whole solution has been added the now freshly formed lipopolyplexes have been transferred to a spin-column (MWCO above 5000 Da) and concentrated to the initial volume of 1 mL.

Radiolabeling
General methods for radioiodination work and HPLC traces can be found in the ESI. † Preparation of MeO-[PEG 6 -[ 125 I]-IPQA] (8). A Sep-Pak tC18 Plus Light Cartridge (145 mg, Waters Ltd, Cat. no. WAT036805) was flushed with ethanol (5 mL), HPLC water (10 mL) and was air-dried (10 mL). TKI-tin precursor (S6) (1.8 mg) was dissolved in PBS (0.18 mL, Sigma-Aldrich Ltd, Cat. no. P4417). This stock solution (40 µL) was added to an iodogen coated tube (Thermo-Fisher Ltd, Cat. no. 28601) followed by mixing with [ 125 I]NaI (5 µL, 34.2 MBq). After standing for 30 min at room temperature the reaction was quenched by adding HPLC mobile phase (100 µL of MeOH/H 2 O/TFA 10/90/0.1 v/v/v). The reaction mixture was injected into a semi-preparative HPLC. The product peak (t R = 18.9 min) was collected and diluted with H 2 O (10 mL). A conditioned t C18 SepPak Light cartridge was loaded with that solution, washed with H 2 O (5 mL) and eluted with ethanol in six fractions (0.1 mL). The overall RCY was 38%. The radiochemical purity was >99% as determined by analytical HPLC. The UV channel (254 nm) did not show any stable impurities. The identity of the radioactive product was confirmed by co-injection with a non-radioactive reference solution of 9. The product was formulated by mixing aqueous cold compound (100 µM, 1.1 mL) with an ethanol fraction of [ 125 I]-8 as collected from the SepPak cartridge (6.8 MBq).

In vivo and in vitro testing
Tissue culture, reagents and antibodies. LIM1215, DiFi, SW48 and DLD1 colon cancer cell lines (independently validated by STR DNA fingerprinting at The Institute of Cancer Research (London, UK), were maintained in RPMI-1640 (Life Technologies Ltd). Antibodies: anti-EGFR (clone D38B1) antip-Y1173-EGFR (clone 53A5) were from Cell Signaling Technology Ltd.
In vitro liposomal transfection and imaging. A10 µM solution of the non-radiolabelled F1LA1 K 16 Cys-S-S-[PEG 6 -IPQA] lipopolyplex was applied to cells at 60% confluence in 96-wellplates. Cells expressing Picchu-X biosensor were imaged on customised "open" microscope automated FLIM system. 73 Time-domain fluorescence lifetime images were acquired via time correlated single photon counting (TCSPC) at a resolution of 256 by 256 pixels, with 256 time bins and 100 frames accumulated over 300 seconds, via excitation and emission filters suitable for the detection of GFP fluorescence (Excitation filter: Semrock FF01-470/22 nm; Beam Splitter: Edmund 48NT-392 30R/70 T; Emission filter: Semrock FF01-510/20 nm). Conventional wide field fluorescence images were acquired with filter cubes for FITC (Excitation 480/30 nm, emission 535/ 45 nm) on a CCD camera (Hamamatsu 1394 ORCA-ERA), with an exposure time of typically 100-500 ms. FLIM analysis was performed with the TRI2 software (Version 2.7.8.9, Gray Institute, Oxford) as described previously. 74,75 In vivo experiments. All animal studies were approved by the University College London Biological Services Ethical Review Committee and licensed under the UK Home Office regulations and the Guidance for the Operation of Animals (Scientific Procedures) Act 1986 (Home Office, London, United Kingdom). 12 female 6-8 weeks old CD1 nu/nu mice (Charles River Laboratories, UK) were injected subcutaneously in the right flank with 2 × 10 6 LIM1215 cells in 100 μL PBS. Once palpable tumors were measure in three orthogonal directions using the following equation: length × height × width × π/6. Once tumors had reached approximately 70-100 mm 3 mice were randomly divided into 2 groups; group 1 received an