The method has been used to detect both exogenous overexpressed genes and endogenous transcripts, including c-Fos and Arc, in the brains of mice. “In a way, this is an in vivo equivalent of qPCR, but now you don’t have to mash up the tissue,” said corresponding final author Jerzy Szablowski.
Monitoring gene expression can help better understand organism development and disease progression, identify and measure biomarkers, and monitor therapy efficacy. However, many of the tools available to researchers are only useful on sectioned or pulverized tissue samples, providing just one snapshot after the animal is dead or a biopsied sample is removed. Alternatively, bioluminescent or MRI-active reporters can be used to visualize the expression of genes or proteins in live animals by introducing synthetic constructs containing engineered gene promoters, but only a few reporters can be used at once, and most depend on a form of energy to penetrate tissues, limiting sensitivity in larger animals or other specialized cases. Additionally, many promoters are too large to be packaged in adeno-associated virus (AAV) vectors commonly used to introduce the constructs.
“They were using RNA that could hybridize with the RNA within the cell in a sequence-specific manner; when it does so, there is recruitment of ADAR, which goes in and replaces one of the ribonucleotides in the triplets that encode amino acids,” Szablowski explained. “If you’re clever, you can put a stop codon in the middle of that sensor that will stop any translation downstream,” he added. In that case, when the designed RNA sensor binds to a transcript of interest, ADAR acts on the duplex, converting the stop codon triplet UAG to UIG, which is then read as UGG (tryptophan). “If the hybridization happens, it actually removes that stop codon and puts a tryptophan instead that can be read through and translated,” Szablowski explained.
Szablowski’s group saw that this technique could take their RMA technology to the next level. “We wanted to have the kind of tool that everybody can use for any gene they want,” he said. They applied the same concept of a brain-escaping reporter protein but put that encoded information downstream of a stop codon in a designed complementary transcript. In this way, a sensor could report on any RNA transcript of interest, using the following general design: target gene complementary sequence with a central stop codon having a single mismatch ➜ encoded reporter (e.g., luciferase) ➜ BBB exit signal. They call the completed method in-vivo tracking of active transcription (INTACT).
“You basically specify the sequence you want to study, and then you look for the markers in blood, and they’ll tell you how much transcription is happening,” Szablowski summarized. He elaborated on the underlying rationale: “The markers that you have in your body don’t really report on everything; there are many genes that are sitting inside the cells, and they never have a marker … that’s why I’m making these synthetic markers.”
Gene Transcript Levels Measured in Living Mice
Ultimately, the new approach worked. The team demonstrated this in a range of experiments, detecting exogenous fluorescent protein transcripts, neuronal subtype transcripts (with up to 1295-fold increased blood luciferase levels), and immediate early genes that are activated quickly after neuron firing. Detecting neuronal subtypes exemplified another advantage, as they detected tyrosine hydroxylase (TH) in dopaminergic neurons using this method; with traditional techniques, the 2.5-kb TH promoter would use up ~50% of the AAV packaging capacity.
Szablowski noted that “my lab studies the brain a lot for strategic reasons, but it should work anywhere.” The method is quite sensitive too. In the 2024 report of RMA development, Szablowski says that “you could record the transduction of gene therapy in [as few as] 12 neurons in the mouse, and that took about 10 microliters of blood, or about 5 microliters of the serum.”
Already, this represents a significant improvement, because the INTACT method can be directed to any gene of interest, and multiple measurements can be acquired from a single living organism over time. However, the real next-level development involves multiplexing, querying several genes at once.
To demonstrate the multiplex idea in principle, Szablowski and colleagues performed an experiment using three different luciferase enzymes (Gaussia, Cypridina, and Nano luciferase) as reporters on three different sensors. They then injected each sensor and other INTACT components into one of three different regions of the brain (hippocampus, striatum, or substantia nigra) of a single mouse. From one blood sample, they were able to identify the Arc transcript levels originating from each region of the brain based on the luciferase type.
However, to approach the step change similar to what qPCR or next-gen sequencing enables, INTACT gene probing would need to dramatically expand. “The next step will be multiplexing, to go from ‘in vivo qPCR’ to ‘in vivo transcriptomics,’” he said. “We’re actually working on barcodes that we’re going to introduce probably later in the year, which will allow you to measure large numbers of signals at once, which is kind of the whole point and advantage of this technology.” Instead of using a protein reporter such as luciferase, simpler unique amino acid barcodes could be applied to differentiate the expression of different genes. “If we can have barcodes that can be detected by mass spec in sufficient precision, then you could imagine connecting, let’s say, a thousand barcodes to a thousand different sensors and then injecting that into some tissue that you want to study over time and see how the transcriptomics of that site changes,” Szablowski said. “MRI, PET, fluorescence, luminescence, or ultrasound do not have the multiplexity ‘number of colors’ to do this,” he added.
There are some additional ideas that may be pursued with these reporters. “You can also make a ladder where you have a general promoter that expresses a sensor that expresses a sensor downstream,” Szablowski said. In other words, the technique could be further modified to enable something like the detection of a specific transcript, but only when expressed in a certain cell type, or co-expressed with a second transcript, although Szablowski cautions that efficiency may suffer as the ladder grows.
Introducing INTACT Components
The INTACT sensors must still be delivered to the right cells. In the Nature Communications report, Szablowski’s group used AAVs injected into precise areas of the brains of mice, “although it doesn’t have to be viruses; we are working on other ways of doing that non-viral,” he said. Lipid nanoparticles are one example of a non-viral gene delivery method that could be pursued, and different strategies can be used to target delivery to a particular tissue. For instance, in their report, the sensor RNA molecules were limited to neurons by placing expression under the control of a neuron-specific promoter.
Spatial Precision Using Special Techniques
Szablowski referenced another 2024 Science Advances report in which his group used a newer technique to gain spatial information. In that case, they used intravenously injected brain-targeted AAVs driving reporter proteins that lacked BBB escape capacity. Focused ultrasound was then used to specifically open the BBB in targeted regions with millimeter precision, thereby releasing reporter molecules from only that area. “That allows you to have this kind of synthetic marker measurement with spatial precision, and in the same individual you can look at different brain regions,” Szablowski said. The technique can also be used in the opposite direction. “You actually can open the BBB to allow for gene delivery to specific brain regions,” Szablowski said.
This work also led to a funded clinical trial (not yet active) for studying Parkinson’s disease, which will use BBB opening with focused ultrasound to release proteins from the brain that would otherwise not make it into circulation. “That can actually guide you: what kind of protein changes are occurring in the real person with real Parkinson’s disease (PD)? And then you can actually start generating hypotheses about what is the mechanism of PD, which is something that we don’t really know in a human”, he said, adding that we aren’t going to biopsy the brains of individuals with early-stage PD.
For now, this method is best understood as an extension of and a complement to other methods for use in animal models, but multiplexing hundreds or thousands of gene transcription reporters could already be transformative in the same way that qPCR and NGS have been. It is conceivable that the methods will find their way to diagnostic use in humans. As a first step toward this possibility, another study published in May in Neuron saw Szablowski’s team extending the method to detect transcripts in non-human primates. He says they basically just swapped the domains that enable BBB crossing in a mouse to ones that work in a macaque. “It’s quite plausible that if you … swap the human domain, it’s also going to work,” he said. Even without that, accomplishing the feat in primates is significant. “You cannot just go willy-nilly and sacrifice 100 of these animals, for cost and ethical reasons, but imagine if you could get 100-fold data per every single one … that already would be revolutionary,” Szablowski said.
For human use, a number of kinks would need to be worked out. Luciferase is immunogenic in humans, and the bar for demonstrating necessity is much higher. A potential first application would be with gene therapy. “Let’s say in the retina where you cannot take a biopsy, having ability to monitor the gene expression would be really useful,” Szablowski said. “You cannot do fluorescence because the lasers that use it are too strong; you cannot do biopsy because it’s a retina,” he said. With this method though, you could assess the gene therapy expression “easily with just a blood test that you can do in any medical center,” and because of the high sensitivity, “it really would just be a small fraction of the gene therapy delivered,” Szablowski said. He says he is pursuing these efforts in conjunction with undisclosed gene therapy companies.
He has bigger ideas too. Down the road, “I’d love to have a transcriptomic profile of a disease over time … let’s say somebody’s bone marrow as they begin to develop aplastic anemia or how the Parkinson’s disease neurons die.”