Here's how I read the paper that I think we're both discussing, but hasn't been explicitly quoted so far:
The first time pH is mentioned it is because attempting to formulate their first compound at a ph of 7 resulted in deamination - it wasn't stable enough.
"We found that formulation at pH 7 caused chemical instability including deamidation of asparagine residues, as reported for pramlintide, (17,18) and this could not be solved by formulations or minor pH adjustments. In order not to deviate too much from h-amylin, formulation at low pH seemed necessary."
It was much more stable at a pH of 4, but that did not stop fibril formation.
"Formulation at pH 4 was preferred for chemical stability based on published pramlintide data, as described above. (18) However, when 1 was taken into further preclinical development and high doses were given subcutaneously to rats, there were quite severe injection site reactions with signs of necrosis. A precipitate was observed at the injection site that appeared to have a fibrillar structure resembling classical h-amylin fibrils when studied with electron microscopy."
This appeared to be the result of fibril formation at higher a pH when solubility was low
"The hypothesis emerged that compound 1 in low doses binds to albumin at the injection site and is kept soluble by albumin despite low solubility at pH 7.4 in albumin-free aqueous solutions. At high doses, there might not be enough albumin at the injection site, and compound 1 precipitates and forms fibrils as pH rises."
They did not solve this by tinkering with the pH, they tossed this product (compound 1), and started looking for something that had better solubility and lower propensity to form fibrils in the pH range that offered the most stability.
"As fibrils of h-amylin are known to be cytotoxic, (31) the new design strategy therefore focused at identifying compounds with better solubility in the pH range 4.0–7.4 and with low tendency to form fibrils."
The explicit purpose of the centrifuge test was to evaluate tendency to form fibrils.
"Two assays were introduced to accommodate the selection. First, a simple solubility assay (Table 4) in the pH range from formulation pH (4.0) to physiological (7.4) was used... Second a fibrillation assay used in insulin research was introduced to filter out compounds with a high propensity to fibrillate (Table 5). It was based on Thioflavin T fluorescence of fibrils after exposure to mechanical stress over 2 days.
They created over 800 compounds while trying to get this right. Only four resisted fibril formation in the assay (mechanical centrifuge test) for more than 35 hours and showed significant potency.
"Among the more than 800 amylin-based peptides made..."
"Only nine compounds had a lag-time longer than 35 h, and of those, only four had attractive in vivo properties—high potency and long duration of action (8, 22, 23, and 24)."
They ultimately went with #23 because they expected it to perform better when in humans. Compound #23 is Cagrilinitide.
"In the final choice between 22 and 23 (differing at position 17; Figure 2), the species difference in vitro indicated that 23 might have the largest potency difference in favor of the human receptor compared to 22."
