DNA in our cells is packed into chromosomes and organized into looping structures in three-dimensional space. These loops are known as topologically associating domains (TADs). Perturbations of these chromosome 3D structures are found to be distinctive landmarks of cancer, aging-associated diseases, neurodegenerative diseases, metabolic diseases, and recently, long-COVID.

The current dogma to explain how these loops are formed is called the "loop-extrusion" model. In this model, a ring-like protein, cohesion, initiates loop formation at a loading site on the chromosome within a “TAD-to-be”, and then it rapidly extrudes until encountering protein roadblocks, namely transcriptin factor CTCF, at the loop boundary to constrain the loop. The structure of the loop generated by cohesin extrusion is called a "stem-loop," and it resembles a bolo tie. The Schedl lab findings challenge the loop-extrusion dogma, which has been echoed in hundreds of research publications and reviews.

Back-to-back papers from the Schedl lab, published in elife, show that the loop extrusion model cannot explain TAD formation in the model organism, the fruit fly.  Instead, these papers show that TADs are generated by a completely different mechanism called “boundary-boundary pairing.”  In this model, proteins that bind to special DNA elements, namely TAD boundaries, physically interact with proteins bound to neighboring boundaries. These interactions are specific and determine which boundaries pair with each other.  Boundary pairing is also orientation dependent. Head-to-tail pairing generates a TAD that has a stem-loop structure like that predicted in the loop-extrusion model.  However, head-to-head pairing generates a TAD that has a circle-loop structure much like the loops in a slinky.  Circle-loops cannot be generated by the loop extrusion mechanism.

Based on the patterns of interactions seen between TADs genome wide, most of the genome appears to be assembled into circle-loop TADs in flies and in mammals.

The boundary model has the additional advantage that specificity in TAD formation and in gene regulation is built into the mechanism: boundaries pair with compatible partners and their pairing interactions are orientation dependent. By contrast, the cohesin loop extrusion-road block model provides no mechanism for specifying which boundaries end up linked together, or which gene gets targeted by which regulatory element. 

For further reading, see:
Stem-loop and circle-loop TADs generated by directional pairing of boundary elements have distinct physical and regulatory properties
and
Chromosome structure in Drosophila is determined by boundary pairing not loop extrusion