CircRNAs are single-stranded RNAs that form covalently closed loops, with no free ends. They were first discovered as the genomic material of plant viroids and hepatitis delta virus, and were later found to be produced by eukaryotes as well. Advances in next-generation sequencing and bioinformatics then revealed that circRNAs are abundant, ubiquitously expressed, regulated, and functional, upending the long-held view that they were rare and useless byproducts of splicing.

Whether viruses encode and produce circRNAs remained an open question until oncogenic DNA viruses were found to generate circRNAs using host spliceosome machinery. We started this project by focusing instead on viruses with RNA genomes, both because of their diverse life cycles and because the question was still unexplored. To begin, we selected HIV-1 as one of our model viruses, because in some ways, it stacked the odds in favor of finding viral circRNAs. HIV-1 has an RNA genome, but it is reverse transcribed into DNA and integrated into the host genome. From there, the integrated provirus is transcribed and spliced using host machinery. Alternative splicing is also essential for HIV-1, which increased the likelihood that circRNAs could form and that they might matter in viral pathogenesis.

The Return to CircRNA: Discovery and investigation of circRNAs distributed across the HIV-1 genome

We enriched circRNAs from cell lines carrying integrated HIV-1 genomes at one or two copies and performed RNA-seq. Using specialized bioinformatics pipelines to identify putative circRNAs, we found nine viral circRNAs distributed across the HIV-1 genome. We selected the most abundantly expressed viral circRNA common to both cell lines for validation and named it circHIV. RT-qPCR using divergent primers spanning the unique backsplice junction produced a distinct band, and Sanger sequencing of that band perfectly matched the predicted backsplice junction. We received these data in late February 2020, shortly before COVID-19 measures were implemented in the lab and around the world.

During our time away from the lab, we brainstormed experiments, tested models on paper, and read widely. Everybody in the lab also wrote a review. By the time we returned full-time, we were eager to continue the investigation.

The Two Towers: CircHIV is physiologically relevant and plays a virus-promoting role

We subsequently found that circHIV is present during wild-type and pseudotype HIV-1 infection in primary CD4+ T cells and other cell lines, as well as in plasma from people living with HIV-1. Once we saw that circHIV appeared during natural infection and was not limited to a particular experimental system, we felt confident that it was physiologically relevant. From there, we moved on to study its cellular and virion localization, translation potential, and biological function.

We found that circHIV localizes to both the nucleus and the cytoplasm, and is highly enriched in virions. We could not detect translation from circHIV, although we do not exclude the possibility that translation might occur under specific physiological conditions or in particular cellular contexts.

To probe biological function, we depleted circHIV by inducibly expressing shRNAs targeting the backsplice junction. Subsequent viral infection and flow cytometry showed that circHIV knockdown significantly reduced both the percentage and mean fluorescence intensity of GFP-HIV-1-positive cells compared with scrambled shRNA controls. The complementary experiment told the same story in reverse: when we overexpressed circHIV by delivering a synthetic circHIV RNA generated by in vitro transcription and splint ligation, circHIV increased GFP-positive cells relative to a circ-mCherry control. Together, these findings suggested that circHIV may play a virus-promoting role.

We then determined that circHIV acts during HIV-1 transcription by stimulating promoter activity through interaction with the viral protein Tat. We wanted to identify the contact regions for circHIV and Tat, so we generated circHIV mutants in which selected regions were scrambled to disrupt primary sequence and higher-order structure. One mutant markedly reduced binding to Tat compared with wild-type circHIV, revealing that those scrambled nucleotides were critical for the interaction. We also wondered whether circHIV and TAR competed for Tat binding. In vitro pulldown assays showed no loss of binding when the reciprocal RNA competitor was added, suggesting that circHIV and TAR bind Tat independently rather than competing for the same RNA-binding site. These results reveal that Tat may have a non-canonical binding domain for circHIV.

The Fellowship of the CircHIV: A circRNA journey marked by interdisciplinary collaboration

This project came with both expected and unexpected challenges. CircRNA studies are inherently difficult because circRNAs are low in abundance, lack a common positive-enrichment feature like the polyA tail of mRNAs, and share high sequence similarity with their linear RNA counterparts. We kept these issues in mind throughout the study to make sure we were only perturbing the circRNA. To support our conclusions, we included both linear and circular RNA controls throughout the experiments.

The biggest unexpected challenge, of course, was the COVID-19 pandemic. Our momentum paused when we were out of the lab. We are grateful to the interdisciplinary community around us. Colleagues put us in touch with their colleagues, and piece by piece, different aspects of the project came together. Everyone in the lab generously donated blood as negative controls for experiments. Among the co-authors are people with expertise in molecular biology, immunology, virology, biochemistry, and cell biology. There are physicians and basic scientists. At times, the project felt a little like The Lord of the Rings: a difficult journey that only worked because it was carried forward by a fellowship.

Since our work began, many researchers have helped advance our understanding of circRNAs encoded by both hosts and pathogens. It has been exciting to watch this field grow over the past decade, and we are looking forward to the discoveries still ahead.