Each year about a third of deaths globally are due to cardiovascular disease (CVD), which includes coronary heart disease, stroke, and congestive heart failure. Elevated cholesterol or low-density lipoprotein (LDL) in the blood is a well-established causal risk factor for CVD. As a result, LDL-cholesterol is often referred to as the "bad cholesterol". The two main therapeutic interventions for lowering "bad cholesterol" are statins and PCSK9 inhibitors, which indirectly increase the availability of LDL receptor (LDLR) in the liver to lower the amount of LDL in the blood.

The detailed structure of apoB100, the main protein in LDL, bound to its receptor was long sought after given its importance in drug development. The consensus of the field was that the structure of apoB100 would never be resolved to less than a nanometer because of its association with lipids.  With the right people, the right resources, and hard work, the complex of LDL with LDLR was resolved with enough resolution to build most of its completely novel molecular structure. Unexpectedly, there were two receptor binding sites on apoB100, which spanned a much larger region than anticipated.   Notably several variants of the apoB100 protein, which are associated with high LDL-cholesterol and premature CVD, are in one of these binding interfaces. We hope this new structural information on LDL helps to enhance genetic counselling, personalized medicine, and the development of future therapeutic interventions.

There were several challenges to sample preparation. Many groups have struggled to make recombinant LDLR. ApoB100 could not be made recombinantly due to its large size (4536 amino acids after cleavage). Even purified from donated human blood, LDL is large, asymmetric, and size- and shape-heterogeneous, and has a limited shelf-life before it becomes altered by oxidation or aggregation.

Meeting these challenges was a relay race and could not have been done by a single laboratory. Alan Remaley (NHLBI) and Joe Marcotrigiano (NIAID) have entire careers of experience purifying lipoproteins (including LDL) and producing difficult recombinant proteins, respectively. Mart Reimund (NHLBI) and Giorgio Graziano (NHLBI) performed the 20-day marathon of purifying the LDL from donated human blood, followed by a 20-hour sprint to produce and purify the complex. They did this more than once. Just when it looked like they were going to drop from exhaustion, Haotian Lei (NIAID) and Altaira Dearborn (NIAID) prepared grids for cryo-electron microscopy completely in parallel as insurance against losing the sample at this stage.

The resulting sample posed additional challenges to structure determination. Because the data was still incredibly heterogeneous, the investigators kept collecting at each opportunity ultimately exceeding 35,000 movies. Because of the size of the dataset and the box size of the volume, they had to use cloud computing to process the tremendous amount of data generated. This enabled Altaira to experiment with the data in parallel, allowing these structures to evolve out of the data. Altaira handed-off the structure building to Joe who, with his extensive experience in X-ray crystallography built 11,000 amino acids of structure (the equivalent of about three polymerases).

Having everything and everyone so close, the investigators were able to go from the patient to the laboratory to the electron microscope and to the cloud. This proximity of expertise enabled each to perform the job at which they had the most expertise, trading ideas efficiently as they went. Overlapping expertise, created a collegial environment for solving problems and generating robust scientific interpretation. As the interpretation of the structure evolved, Mart worked with Francis O'Reilly (NCI) at testing that interpretation biochemically and with cross-linking. Many works take a village to bring to completion. In the end, several villages at the intramural NIH campus came together to make this work possible.