Edinburgh Research Explorer Targeted microbubbles carrying lipid-oil-nanodroplets for ultrasound-triggered delivery of the hydrophobic drug, Combretastatin A4

A number of highly potent and promising drugs fail to reach the clinic due to poor-water solubility. Lipid-stabilized Oil Nanodroplets (LONDs) were produced specifically 19 for the encapsulation of poorly-water soluble drugs such as the vascular disruptive agent Combretastatin A4 (CA4). Initial pre-clinical work with CA4 LONDs showed an 20 accumulation of CA4 in tumor tissue. Attachment of CA4 LONDs to VEGFR2-targeted Microbubbles (MBs) permitted the controlled and ultrasound triggered release of 21 CA4 in tumors, confirmed by a reduction in tumor perfusion post-treatment. The combination of low dose irinotecan with CA4 LONDs-MBs further enhanced the anti-tumor 22 effects of both compounds Abstract 13 The hydrophobicity of a drug can be a major challenge in its development and prevents the clinical translation of highly potent anti-cancer 14 agents. We have used a lipid-based nanoemulsion termed Lipid-Oil-Nanodroplets (LONDs) for the encapsulation and in vivo delivery of the 15 poorly bioavailable Combretastatin A4 (CA4). Drug delivery with CA4 LONDs was assessed in a xenograft model of colorectal cancer. LC – 16 MS/MS analysis revealed that CA4 LONDs, administered at a drug dose four times lower than drug control, achieved equivalent 17 concentrations of CA4 intratumorally. We then attached CA4 LONDs to microbubbles (MBs) and targeted this construct to VEGFR2. A 18 reduction in tumor perfusion was observed in CA4 LONDs-MBs treated tumors. A combination study with irinotecan demonstrated a greater 19 reduction in tumor growth and perfusion ( P = 0.01) compared to irinotecan alone. This study suggests that LONDs, either alone or attached 20 to targeted MBs, have the potential to significantly enhance tumor-specific hydrophobic drug delivery. 21

Targeted microbubbles carrying lipid-oil-nanodroplets for ultrasound-   37 which are often resistant to conventional chemotherapy. 2 38 Numerous preclinical studies have shown that VDAs cause a 39 significant reduction in tumor blood flow in the core of the tumor 40 followed by secondary necrosis. [3][4][5] However, VDAs spare the 41 rapidly proliferating cells of the tumor periphery, leaving behind 42 a "viable rim" of cells which is a major cause of resistance and 43 the reason for combination treatments with conventional 44 chemotherapy. 6 45 CA4 is a natural product originally isolated from the African 46 tree Combretum caffum. It is a potent inhibitor of microtubule 47 polymerization 7 , 8 binding near the colchicine binding site and 48 preventing the "curved" to "straight" tubulin transition. 9 In vitro, 49 CA4 causes complete disorganization of cytoskeletal microtubules 50 in endothelial cells, which in vivo manifests as the rapid shutdown 51 of tumor vasculature. 10, 11 Due to its poor aqueous solubility, the 52 prodrug CA4 phosphate (CA4P) has been developed. 12      Tumor perfusion was assessed by uptake of Hoechst 33342, a 147 DNA binding dye that when left in the circulation for 1 min, stains 148 the endothelial cells of blood vessels perfused at the time of 149 injection. 34 Tumors were also stained with a primary antibody for 150 CD31, visualized using an appropriate secondary antibody and 151 imaged using a Zeiss Axioimager Z1 fluorescence microscope 152 with AxioVision software (Carl Zeiss Microscopy, USA). 153 Perfusion was scored using a semi-quantitative scoring system to 154 define none (score 0), weak (score 1), moderate (score 2) and high 155 (score 3) intensity of Hoechst 33342 fluorescent staining by two 156 independent observers blinded to the treatment groups. Represen-157 tative images for the perfusion scoring are shown in Supplemen-158 tary Figure S2 for each data set. Data set 1 was used to score images 159 represented in Figure 4 and data set 2 was used to score images 160 represented in Figure 5 of the results section.   The CA4 LONDs-MB construct consists of a lipid/PEG coated MB (light green representing the gas C 4 F 10 core with red shell) covalently conjugated (through biotin/neutravidin interactions) to a layer of LONDs (yellow with purple shell) encapsulating CA4 (dissolved in oil). The construct also carries anti-VEGFR2 antibodies to promote tumor specific delivery by allowing accumulation in tumor vasculature where VEGFR2 is more prevalent than in normal vasculature.
To assess if drug delivery with CA4 LONDs altered the 182 biodistribution of CA4, mice bearing SW480 human CRC 183 xenografts were given a single treatment (preparation 1, Table 1) 184 which equated to a dose of 12.8 mg/kg. Free CA4 dissolved in a 185 combination of DMSO/peanut oil at 50 mg/kg (the standard dose 186 in preclinical trials 10 ) delivered intraperitoneally (i.p.) was used 187 as a control ( Figure 2, A and B). The concentration of CA4 in 188 tumor, liver and plasma 1 h post treatment was determined using 189 LC-MS/MS (Figure 2, C). CA4 was only detected in tumors (2/ 190 3) from the CA4 LONDs group, being below the LOD in liver 191 and plasma while in the free CA4 group, CA4 was detected in 192 tumors (2/3), in 1/3 liver and 1/3 plasma samples.

193
The main metabolite of CA4, CA4 glucuronide (CA4G) was 194 also detected but not quantifiably (due to the lack of appropriate 195 standards) in the plasma, liver and tumor from the free CA4 196 group and liver only from the CA4 LONDs group (Supplemen-197 tary Table S2). The CA4G data suggest that delivery of free CA4 198 leads to its rapid metabolism to CA4G. While Q18 the detection of 199 CA4G in the tumor samples suggested that CA4 was directly 200 converted to CA4G intratumorally. However, it should be noted 201 that due to the intraperitoneal delivery route of free CA4, CA4 202 may have undergone hepatic metabolism before entering the 203 bloodstream. As CA4G was not detected in tumor samples from 204 the CA4 LOND group this suggests that either CA4 is still 205 encapsulated in the LONDs and unavailable for metabolism or 206 encapsulated CA4 is potentially being released slowly into the 207 tumor from the LONDs and the levels of CA4G following 208 metabolism are below the LOD.

209
Nevertheless, these results confirmed that drug delivery to 210 tumors of CA4 in both free and LOND encapsulated forms was 211 possible and also showed that tumors from both groups were   Day 0 is the pre-treatment day where the mean (± SD) starting tumor volumes and mouse weights for each group were 53 (± 63) mm 3 and 33 (± 4) g for CA4 LONDs and 64 (± 22) mm 3 and 33 (± 4) g for free CA4. (C) CA4 concentrations in tumor, liver and plasma tissue 1 h post treatment with CA4 LONDs (12.8 mg/kg) and free CA4 (50 mg/kg). Limit of detection (LOD) for CA4 was 10 ng/mL.  Tumor response was further assessed by determining the % 238 hemorrhage and necrosis in tumor tissue (Figure 3, F and G). 239 CA4 LONDs and free CA4 did not appear to cause any treatment 240 associated hemorrhage as the highest median % hemorrhage was 241 observed in the vehicle group, pointing towards inherently leaky 242 tumor vasculature (Figure 3, F). Despite extensive necrosis in 1/ 243 4 tumors treated with CA4 LONDs, the % necrosis was not 244 significantly different from free CA4 and vehicle treated tumors 245 (Figure 3, G). Treatment with CA4 LONDs and free CA4 did not 246 significantly alter the number of CD31+ blood vessels compared 247 to the vehicle (Figure 3, H).
have shown a reduction in perfusion at concentrations of 50 mg/ 271 kg or above. 3 , 10 , 36 A low frequency, high amplitude US pulse 272 was applied at the tumor site using a specifically designed 273 custom-built single element US system (UARP) 4 min post-MB 274 injection. All groups were exposed to a 5 s US trigger (+T). This 275 was a 10 μs tone burst US pulse, with a peak negative pressure of 276 260 kPa and 1 kHz pulse repetition frequency (PRF), using a 277 2.2 MHz transducer designed to destroy the MBs in situ.

278
Tumor morphology at 1 h post treatment with CA4 LONDs-279 MBs or free CA4P was similar to vehicle treated tumors (data not 280 shown). Most of the tumor was viable with some areas of 281 hemorrhage present to the same extent as the vehicle group. 282 Assessment of perfusion using Hoechst 33342 was adopted as a 283 surrogate biomarker of tumor response, since quantitation of 284 tumor necrosis, hemorrhage, blood vessel number and tumor 285 volume was not sufficiently sensitive to detect any very early 286 tumor responses to CA4 LOND therapy.

287
Perfusion in the tumor core was reduced in treated groups 288 compared to vehicle (Figure 4, C). Perfusion in the tumor 289 periphery was not reduced in the treated groups compared to F 308 by cellular carboxylesterases to its active metabolite SN38, a 309 topoisomerase I inhibitor used in the treatment of CRC and liver 310 metastases. 40 A low and frequent dosing schedule (metronomic) 311 was chosen since reports have shown that frequent lower doses 312 of CA4P are more effective at sustaining its anti-tumor effects. 41 313 Irinotecan administered using a metronomic schedule has also 314 shown enhanced efficacy, potentially by exerting additional anti-315 angiogenic effects. 40 , 42 316 There is conflicting evidence regarding the timing and 317 sequence of administration of CA4P when used in combination 318 therapy. Theoretically the greatest anti-tumor activity should be 319 observed when CA4P is administered after or concurrently with 320 the combination agent, as potential changes to the tumor 321 microenvironment induced by CA4P may hinder partner drug 322 uptake. [43][44][45][46] Irinotecan (10 mg/kg) was administered 1 h prior to 323 CA4 LONDs-MBs (0.001 mg/kg) or CA4P (0.001 mg/kg) to 324 treat SW480 human CRC xenografts (Figure 5, A). CA4 325 LONDs-MBs and CA4P were administered at the same 326 concentration to investigate if the targeted, triggered delivery 327 of CA4 using the LOND-MB construct had a greater effect on 328 tumor growth compared to delivering free drug systemically. 329 Administrating irinotecan 1 h prior to any other treatment allows 330 for the active metabolite of irinotecan, SN38 to reach its peak 331 conversion in the blood. 47

337
Tumor growth was inhibited in both combination groups with 338 irinotecan and CA4 LONDs-MBs or CA4P. Irinotecan + CA4 339 LONDs-MBs significantly inhibited the growth of tumors by day 340 11 compared to vehicle control (PBS) (P = 0.05) (Figure 5, C). 341 In contrast the combination therapy with irinotecan and CA4P 342 significantly inhibited the growth of tumors from day 7 onwards 343 when compared to vehicle (P = 0.01, P = 0.02 and P = 0.04) 344 (Figure 5, C). By day 14 after the fourth treatment, tumor growth 345 was also inhibited in the irinotecan + CA4P group compared to 346 irinotecan alone (P = 0.04) (Figure 5, C). The %TGI compared 347 to vehicle was 22.9% for irinotecan only, 88.5% for irinotecan 348 and CA4P and 47.8% for irinotecan and CA4 LONDs-MBs. 349 Compared to irinotecan the %TGI was 85.1% for irinotecan and 350 CA4P and 32.2% for irinotecan and CA4 LONDs-MBs.

351
Tumors in all treatment groups were smaller in mass than 352 vehicle controls. However, only tumors in the irinotecan + CA4P 353 group reached statistical significance when compared to vehicle 354 (P = 0.02) (Figure 5, D). Tumor doubling times were not 355 significantly increased in the irinotecan and CA4P group 356 compared to vehicle (Supplementary Figure S4). Tumor 357 regression was also observed in all treatment groups (Supple-358 mentary Figure S4). Body weight was monitored throughout the 359 study with only very mild body weight loss observed in the 360 irinotecan only group after the initial treatment (before 361 recovering by treatment three) (Supplementary Figure S5).  The data presented in this study showed that CA4 LONDs 419 administered in vivo at a dose four times lower than free CA4, F 420 delivered a similar concentration of CA4 intratumorally 421 (~1000 ng/g) with none detected (or below the LOD) in the 422 liver or plasma 1 h post treatment. In contrast, CA4 was present 423 in the liver and plasma samples from the free CA4 group. It was 424 therefore concluded that CA4 LONDs were more effective at (F) Hoechst 33342 intensity scoring used to score perfusion. A statistically significant difference (P = 0.02) was observed between vehicle and irinotecan + CA4 LONDs-MBs. The data represent the median score from two independent blinded assessors and the error bars denote the interquartile range. Significance was calculated using a Mann-Whitney U test, two-tailed.
Numerous pre-clinical studies have shown that the effects of a 431 single treatment with CA4 or CA4P on the growth of tumors are 432 very limited with modest growth delays only observed at doses 433 higher than 100 mg/kg. This is mainly attributed to the viable 434 rim remaining post treatment. 38 Therefore in order to obtain a 435 better therapeutic response with CA4, multiple dosing regimens 436 are used. 41 The timings between treatments in our study were 437 over 24 h, as a longer retention and prolonged exposure in the 438 tumor were anticipated with CA4 LONDs, due to the 439 encapsulation as opposed to free drug. However, CA4 LONDs 440 caused a modest but statistically insignificant tumor growth 441 inhibition (Figure 3). This was potentially due to the timings 442 between treatments as tumors generally recover from the effects 443 of CA4 24 and 48 h post treatment. 41 , 45 , 54 , 55 In contrast, 444 administrating free CA4 at 3 mg/kg every other day for a total of 445 five treatments has been shown to significantly inhibit growth of 446 hepatocellular carcinoma. 53

447
The combination of nanoparticles with targeted MBs and US 448 has been shown to enhance the efficacy of drugs by increasing 449 the circulation times and improving drug accumulation in tumor  colon. Two-way ANOVA followed by Holm-Sidak's multiple comparisons test showed that a significantly higher concentration of irinotecan was observed in the irinotecan only group compared to irinotecan + CA4P (P = 0.005) and irinotecan + CA4 LONDs-MBs (P = 0.002). Horizontal line ( ) represents the median value. LOD for irinotecan, SN38 and SN38G was 10 pg/L.
with the chemotherapeutic agent irinotecan. This resulted in a 471 significant tumor growth inhibition when irinotecan was given 472 with either CA4 LONDs-MBs or CA4P demonstrating that 473 longitudinally CA4 LONDs-MBs were as effective at delivering 474 a hydrophobic drug such as CA4 when compared to CA4P. The 475 combination of irinotecan and CA4P also significantly inhibited 476 tumor growth compared to irinotecan only, indicating some 477 potentiation of the anti-tumor effects of irinotecan by CA4P. A 478 number of mechanisms may explain this enhanced activity when 479 the two agents are combined. CA4 could induce changes in the 480 tumor microenvironment that enhance the cell-killing activity of 481 irinotecan. Our results showed that CA4 LONDs-MBs or CA4P 482 caused a reduction in perfusion, potentially leading to the 483 entrapment of irinotecan or its active metabolite SN38 in tumor 484 tissue, thus prolonging the exposure times of tumor cells to the 485 active agent. Trapping of cytotoxic agents by CA4P and in 486 particular using irinotecan in combination with CA4P have been 487 previously demonstrated. 59 A trapping effect was not observed 488 in the present study as the biodistribution data at 1 h post-489 injection showed the highest concentrations of irinotecan, SN38 490 and SN38G in the irinotecan only group. However, these tissues 491 were collected following five treatments; therefore, the tumor 492 vasculature could potentially have already been significantly 493 altered and thus hindered the entry of irinotecan as measured by 494 this final time-point. Single treatments at multiple timings of 495 irinotecan and CA4-LONDs-MBs administered would be 496 required to investigate the trapping effect.

497
The potentiation effect shown with CA4 and irinotecan may 498 also result from their different cell targets; thus, this regimen 499 generates a multi-targeting strategy. Irinotecan acts on the most 500 rapidly proliferating tumor cells, while in contrast, CA4 affects 501 the established vasculature. This has been previously demon-502 strated with CA4P and 5-fluorouracil (5-FU), where 5-FU 503 inhibited tumor cells in the viable rim 44 ; in this case, no trapping 504 was observed. Low dose irinotecan (10 mg/kg × 2 per week) has 505 also been shown to suppress the mobilization of circulating 506 endothelial progenitor cells (CEPs) in a murine colon cancer 507 model. 60 CEPs are a known cause of resistance to CA4 61 508 treatment and could contribute to the enhanced anti-tumor 509 activity seen here.

510
Taken together the data presented in this study suggest that 511 LONDs are a promising hydrophobic drug delivery nanovehicle. 512 Their use alone or attached to MBs suggests that this technology 513 has the potential to enhance drug delivery of poorly water-514 soluble drugs while significantly reducing the need for high 515 circulating concentrations. Furthermore, the combination of low 516 dose irinotecan and CA4 holds great promise for effective cancer 517 treatment.