Improving the predictive power of xenograft and syngeneic anti-tumour studies using mice humanised for pathways of drug metabolism

Abbreviations

PDX: patient-derived xenograft

p.o.: per os, oral gavage

s.c.: sub-cutaneous

Introduction

The pre-clinical stage of drug development provides crucial information for the decision process as to whether a drug candidate will proceed to ‘first in human’ and phased clinical trials.¹ The failure rate through the pre-clinical stages of drug development can be high and many candidate molecules taken forward to clinical trials subsequently fail to recapitulate the safety profile and efficacy found in animal studies.^2–5 There are many reasons for this, but significant species differences in drug metabolism between animals (rats, mice) and humans – with concomitant changes in pharmacokinetics, metabolite profiles, toxicokinetics and pharmacodynamics – are key components underlying the observed failure rates.⁵ We have developed a sophisticated transgenic mouse model in which the major human drug metabolising enzymes – and the transcription factors regulating their expression – replace their mouse counterparts. In a previous report on this humanised mouse model⁶ we show, using model compounds and anti-cancer drugs, that drug metabolism and disposition in the 8HUM mouse more closely reflects that found in humans. Given the growing importance of drug combinations in cancer therapy,⁷ it is clear that a genetically engineered mouse model such as 8HUM could play a pivotal role in the development of such combinations.⁸

While some pre-clinical work is carried out in vitro, using a variety of cell lines including immortalised human cells, much – and arguably the most important – is carried out in animals, mainly rodents but also dogs, and for certain types of candidate molecules, primates. One such in vivo use in the pre-clinical setting is syngeneic or xenograft work, where anti-tumour efficacy of drug candidates is tested alone or in drug combinations. In a syngeneic model, murine cell lines are implanted subcutaneously or orthotopically and tumour response to candidate drugs tested. Whilst such experiments can account for the effect of immune system, the genetic background of the tumour cells - which must match that of the host animal used – can also give rise to disparate results. More recently xenograft models have come to the fore, where immunodeficient mouse lines are able to grow human tumours either from existing immortalised cell lines or via fresh tissue as patient-derived xenografts (PDX).^9,10 Despite lack of a competent immune system and any issues that this may potentially cause in the interpretation of results, the latter are growing in use, a good example being the EurOPDX consortium, who have a database of PDX models to share with the research community.¹¹ Notwithstanding extensive xenograft use in various guises, such models still have issues arising from retention of murine drug metabolism and disposition.⁸

In this brief report we showcase a modification of the humanised 8HUM model in which we have generated a compromised immune system by deleting the Rag2 locus. Using murine and human melanoma cell lines in 8HUM_Rag2^-/- mice, we show tumour growth in a syngeneic and xenograft setting, respectively, and demonstrate in vivo sensitivity to dabrafenib and trametinib, drugs currently used in combination as standard of care in the treatment of metastatic melanoma. Together, these models have the potential to significantly reduce the number of animals used in the pre-clinical stages of drug development, while generating more human-relevant data and thus improving the chances of a candidate drug replicating a positive pre-clinical finding in successful clinical trials.

Methods

Results

8HUM mice were used to determine syngeneic growth of the murine melanoma cell line, 5555, derived from a C57BL/6_BRAF+/LSL-BRAFV600E;Tyr::CreERT2+/o transgenic model¹³ and reported by Hirata et al. as being sensitive to the selective BRAF inhibitor, and vemurafenib precursor, PLX4720¹⁹ in vitro, but refractory to this drug in vivo.¹⁴ More recently, the second-generation mutant BRAF inhibitor dabrafenib (in combination with the MEK inhibitor trametinib) has become standard of care in UK and Europe in the treatment of unresectable or metastatic BRAF^V600 mutant melanoma (https://www.nice.org.uk/guidance/).²⁰

All data for this report are available online at.^21,22 As shown in Figure 1, 5555 cells were injected s.c. into one flank of adult female 8HUM mice, divided into two groups of five mice. After five days tumours had established in each animal and were measurable, at which point daily oral treatment with either vehicle or dabrafenib was started. While tumours in vehicle-treated mice continued to grow over the following two weeks, over the same period tumours in mice treated with dabrafenib became almost undetectable. These data demonstrate that C57BL/6-origin tumours can be grown in a syngeneic manner, and that the BRAF^V600 mutant murine melanoma cell line is exquisitely sensitive to the BRAF inhibitor dabrafenib. We have also been able to reproduce the known induction of P450 CYP3A4 by dabrafenib in the humanized mice, further demonstrating the utility of the model in anticancer drug development (unpublished data).

Figure 1. Growth of a murine melanoma syngeneic graft in 8HUM mice and response to dabrafenib treatment.

Adult female 8HUM_Rag2^-/- mice (18–21 w, n = 5) were injected s.c. in one flank with 3.5 × 10⁶ 5555 murine melanoma cells, in 100 μl ECM diluted 1:1 with DMEM. Tumours were allowed to establish and on day five after implantation daily treatment was commenced with either vehicle (0.5% (w/v) hydroxypropylmethylcellulose, 0.2% (v/v) Tween 80; closed circles) or dabrafenib methanesulfonate (in vehicle, open circles) suspended at 6.3 mg/ml and administered at 5 ml/kg, such that dabrafenib dose administered was 31.5 mg/kg (arrow). Tumour measurements were taken three times weekly, then daily as required, and tumour volume calculated as detailed in Methods section. The study was terminated 15 days after implantation of cells.

Data shown are mean tumour volume ± SEM.

By creating an immunodeficient variant of the 8HUM mouse line, where the Rag2 locus is deleted, we extended work to xenografts with the BRAF mutant human melanoma cell line, A375. Figure 2 shows change in total mean tumour volume following s.c injection of A375 cells injected into both flanks of adult female 8HUM_Rag2^-/- mice. Daily treatment of these mice started 28 d after injection of cells, either with vehicle or dabrafenib (arrow); while the tumours in the former group continued to grow, tumours in the mice treated with the BRAF inhibitor shrank in size until by d35 (at which point the vehicle-treated mice had to be sacrificed due to tumour size) there was a significant difference in the treatment effect between the two groups (Figure 2). The data from vehicle group follows the exponential growth dependency and does not fit with the exponential decay function. Conversely, the dabrafenib group data follows exponential decay dependency and does not fit the exponential growth function. These data clearly demonstrate not only that it is possible to grow a human melanoma cell line as a xenograft in this immunodeficient version of the 8HUM mouse model, but it is also possible to demonstrate sensitivity of A375 tumours to BRAF inhibitors.

Figure 2. Response of A375 human melanoma xenograft to dabrafenib treatment in 8HUM_Rag2^-/- mice.

Adult female 8HUM mice (11–19 w, n = 3) were injected s.c. in both flanks with 4.4 × 10⁶ A375 melanoma cells, in 100 μl DMEM. Tumours were allowed to establish and on day 28 after implantation daily treatment was commenced with either vehicle (0.5% (w/v) hydroxypropylmethylcellulose, 0.2% (v/v) Tween-80; closed circles) or dabrafenib (in vehicle, open circles) suspended at 6.3 mg/ml and administered at 5 ml/kg, such that dabrafenib dose administered was 31.5 mg/kg (arrow). Tumour measurements were taken three times weekly, then daily as required, and total volume of tumours on both flanks was calculated as detailed in Methods section. The study was terminated on d 35 after implantation of cells, although one vehicle-treated animal had to be removed from the study on d 29 as its total tumour size approached the maximum permitted under legislation.

Data shown are mean tumour volume ± SEM.

Dabrafenib is used in a clinical setting in combination with the MEK inhibitor trametinib. We tested the ability of trametinib to stop tumour growth, using A375 cells injected s.c. in the flank of adult female 8HUM_Rag2^-/- mice (Figure 3). Daily treatment with vehicle or drug commenced on d 22 after cell injection (arrow), and in the following period tumours in mice treated with vehicle continued to grow until by d 30 they had reached the maximum size permitted. At this point (Figure 3, STOP) tumours in the 8HUM_Rag2^-/- mice had regressed to the point where they were essentially undetectable, and trametinib treatment was stopped. Interestingly, over the following 10 days, tumours began to regrow in the absence of treatment until they were again palpable and measurable, and continued to grow over the next week or so until the study was terminated on d 48 (Figure 3).

Figure 3. Response of A375 human melanoma xenograft to trametinib treatment in 8HUM_Rag2^-/- mice.

Adult female 8HUM mice (8–18w, n = 3 or 4) were injected s.c. in one flank with 5 × 10⁶ A375 melanoma cells, in 100 μl DMEM. Tumours were allowed to establish and on day 22 after implantation daily treatment was commenced with either vehicle (0.5% (w/v) hydroxypropylmethylcellulose, 0.2% (v/v) Tween 80; closed circles) or trametinib (in vehicle, open circles) suspended at 0.07 mg/ml and administered at 5 ml/kg, such that trametinib dose administered was 0.35 mg/kg (arrow). Tumour measurements were taken three times weekly, then daily as required, and tumour volume calculated as detailed in Methods section. Vehicle-treated mice were sacrificed on d 30 after implantation of cells; trametinib treatment was discontinued (STOP) at that time for the drug-treated group to determine whether tumour re-growth would occur in the absence of drug.

Data shown are mean tumour volume ± SEM.

Discussion and conclusions

In a previous publication⁶ we have demonstrated the utility of the 8HUM model in better predicting human drug metabolism and disposition, and in the current report show how the 8HUM mouse and its immunodeficient variant 8HUM_Rag2^-/- are capable of hosting both syngeneic tumours and xenografts, respectively.

Although both rats and mice are used by industry in the drug development process, the utility of mice for growing xenografts, including patient-derived xenografts, and the development of ‘mouse clinical trials’,²³ along with the more advanced use of mice in genetically engineered models (such as 8HUM) means that increasingly mice are being considered more useful for drug development, especially in oncology.²⁴ While numbers of animals used in syngeneic or xenograft work in pre-clinical drug development are difficult to assess, the total will be significant given the number of drug candidates being tested across Pharma at any given time, and such growth of tumours in vivo is also carried out in other research areas, for example, toxicology. A PubMed search for papers published in 2019 found ~5,000 papers containing ‘xenograft’ in the title or abstract, illustrating the extent to which the 8HUM and 8HUM_Rag2^-/- models – modified as appropriate by gene editing to recapitulate disease models - may be able to address 3Rs issues by both refinement: generation of better, more human-relevant data, and reduction – use of fewer animals without loss of statistical power. Pre-clinical use of humanised models to prevent failure of a drug candidate during clinical testing because of species differences in drug disposition would undoubtedly save significant numbers of mice. They will also allow complex drug combinations to be tested and treatment regimens optimised in a manner which is not feasible by clinical trial and reduce the chances of drug-drug interactions.

The 8HUM has some limitations, and it should be noted that a minor complement of murine P450 enzymes remain, and that the Phase II enzymes are murine. These may potentially contribute to drug disposition, as may other pathways e.g., drug transporters. However, the advent of CRISPR/Cas9, as used here to delete the Rag2 locus in the 8HUM mouse, means that it should be relatively simple to additionally modify the 8HUM model to further enhance versatility.

Author contributions

Conceptualization    Wolf, Henderson, Kapelyukh

Data Curation      McLaren, Henderson, Kapelyukh

Formal analysis     Kapelyukh, Henderson

Funding acquisition    Wolf, Henderson

Investigation      Kapelyukh, McLaren, Henderson

Methodology      Wolf, Kapelyukh, McLaren, Henderson

Project Administration  McLaren, Henderson

Resources       McLaren, Henderson

Supervision      Wolf, Henderson

Data availability