Our recent post compared stable vs transiently humanized models, which are needed when human-specific therapeutics are assessed. One issue with these models is donor variability, which has led to the development of unique, chimeric, humanized drug target models.
Overcoming the Variability of Humanized Models
It’s well known that humanized models can provide a variable response to treatment. In stably humanized mice, donor to donor variability can be observed, and original donors are not available if a study needs to be repeated. With transiently humanized mice, this can be partially overcome by using PBMC from at least two different donors in each study, but variability can still occur.
Humanized drug target models were designed to overcome this variability, and provide a practical approach to in vivo immunotherapy assessment, before entering clinical trials.
Human Drug Targets Knocked In Using CRISPR/Cas9
The principle of the platform is to provide a range of models for the in vivo evaluation of human specific biological therapies, such as checkpoint inhibitors. In each mouse model a murine protein drug target e.g. PD-1, CTLA-4, OX40 is knocked out, and the human counterpart knocked in.
This is achieved using CRISPR/Cas9 technology, and creates chimeric models with either specific exons from a gene replaced (e.g. PD-1 exon 2), or in some cases the entire receptor gene switched (e.g. OX40). FACS analysis is commonly employed to ensure the human protein is expressed.
Double knock in models are also being developed for testing human-specific combination therapies. Currently, for example, for a single knock in model you could only test a human anti-PD-1 antibody combined with a second, targeted murine antibody.
Humanized Ligand Models also Available
Obviously not all therapies target receptors, and the same technology can also create syngeneic models featuring humanized ligands, e.g. MC38 models with humanized PD-L1.
The humanized PD-L1 ligand xenograft and PD-1 receptor mouse models can then be combined, to recreate perturbing a full human ligand-receptor complex. PD-L1 can also be replaced within mouse models – combining tumor and mouse both expressing hPD-L1 will ensure a full expression complement across both tumors and host dendritic cells.
Efficacy Testing and Immunoprofiling Models
The main use of these model types is efficacy testing – single agent and combination regimen checkpoint inhibitors, to provide preclinical data to inform on clinical trial strategy.
Downstream analyses such as evaluating TIL is also common, to assess levels such as CD3+ T cells, CD8+ cells, NK cells entering the tumor (and other organs) following different treatments.
A Variety of Models Available
Models are already available for many checkpoint proteins – PD-1, CTLA-4, OX40, CD137, TIM3, PD-L1, with a wide range under development: LAG3, GITR, CD40 etc.
Mouse genes other than checkpoint inhibitors have also been humanized for different immuno-oncology applications:
- Human FcγR receptor knock in enabled the study of FcγR function in vivo and assessment of therapeutic antibodies against this target
- Chimeric MHC-II transgenic mice were shown to mount a human DR-restricted immune response
- Mice with human MCH-I and TCR knock in were used to generate optimal-affinity T cell receptors for T cell therapy
For more information, the concept of these models and where they fit with other in vivo I/O models can be found in this review, Li et al Experimental animal modeling for immuno-oncology.
