Metastasis significantly increases the lethality of cancer, as it is responsible for 90% of cancer-associated deaths. Recently, soluble molecules have been identified as biomarkers for early metastasis, where a crosstalk between granulocytes, platelets, and cancerous cells happens and the metastatic niche is formed to favor the extravasation of tumor cells. The CXCR2 receptor present in granulocytes is responsible for its migration towards this site. Thanks to protein engineering we can take advantage of this receptor and enhance it to be part of a more efficient immune therapy. Our aim is to enhance the efficacy of CAR-T cells treatment through improved chemotaxis by adding a biosensor to recognize and potentiate migration towards early metastatic niches in order to avoid the progression of cancerous cells. We have three defined objectives: to improve the selectivity of the CXCR2 receptor towards CXCL5 and CXCL7, to build a hybrid receptor that senses CXCR2 ligands but has a chemotactic response of a native T-cell receptor, and finally, to combine the previous two objectives to have a robust biosensor to guide CAR T-cells to early metastatic sites. We explored the idea of completely redesigning external loops of CXCR2 by using loop modeling and FunFoldDes to explore both the conformational and sequence space of new loops that would enhance selectivity for CXCL5. The obtained models were subjected to experimental validation using BRET assays, a robust and sensitive technique to measure activation. We achieved a shift from the wild-type response in the activation pathway of the Gi proteins towards CXCL5. We plan to further experimentally analyze this mutant with flow cytometry assays to measure binding affinity. Thanks to these results we will select the motifs responsible for the selective binding and subject them to another round of design, where we can use the same protocol of Loop modeling-FFD with fixed motifs or switch to another loop grafting method to predict and obtain stable structures. We also designed hybrids receptors that combined binding sites of CXCR2 and the intracellular domain of either CXCR4 or CCR5. With BRET measurements, we were able to observe activation in one CXCR4/CXCR2 hybrid. We will now move to a round of design to reconstruct the allosteric pathways of the hybrid receptors. Using the described biosensor will solve problems of CAR-T cell therapy of off-targeting and will be easier to detect metastatic sites and navigate the aggressive tumor microenvironment. Engineering a membrane receptor to function as a biosensor is a stringent test of our biophysical understanding of protein structure and function, but also has important applications and relevance for cancer therapeutics.