Well technically all transcription factors will be biological. Tramscription factors are what binds to DNA and promote binding of RNA polymerase in that region, which causes gene expression. Now it is not like the gene switches - where you modify a transcription factor such that you control when and where it binds and expresses a gene, haven't been made before. There have been ones that can be controlled by tetracycline. There are others controlled by light. The selling point of this paper is there system requires a cheap, non toxic and a non antibiotic molecule aka caffeine to control genes.

So all they did was to first design a a caffine binding protein, technically an antibody.Lets call it aCaff This molecule will dimerize (bond together) with another molecule if aCaff if both of them has caffine bound to them. Now a gene switch system for Tetracycline already exists. It has a transcription factor with DNA binding domain and a tramsactivation domain. The transption factor causes expression of genes it is bound to only when both domains come together. So they fused this transcription factor to caffine binding protein instead of tetracycline binding protein. Now you express these fusion proteins in the cells, along with the gene you want to express and there it is. See this figure
https://www.nature.com/articles/s41467-018-04744-1/figures/1

To answer your next question no they did not use glandular cells. They used human embryonic kidney 283T cells. Which is a cell line once made and now commercially available to any researcher. They use this because it is extremely easy to put any DNA in these cells as compared to your actual glandular primary cells or any other cell line for that matter. So they put DNA for their fusion protein in these cells so it express it. Then they use a promoter(A DNA sequence specific to transcription factors ) specific toTetR and fuse it with gene for glucagon peptide (a protein which is already favourite biological for treating type 2 diabetes). Now you also insert this gene in HEK cells that you inserted your fusion proteins in. Now these cells will express gene for glucagon of and only if you treat them with caffine.

The rest of the story is pretty simple. You just add these cells with all your synthetic genes in a microcapsule and implant it intraperitoneal in the mice. If the implamted mice is model for type 2 diabetes(the leptin knockout mice), then serving them coffee improves their glucose homeostasis.

Let me know if you have more questions on this.

Sunny Kataria, AMRSB