Herein, we reported novel adiponectin variants identified in Thai patients with T2D and demonstrated the effects of amino acid substitutions on structural alterations and the defects in their biochemical properties. The novel R55H and R131H variants were likely to be implicated in T2D pathogenesis by affecting adiponectin multimerization and secretion, respectively. Since the patients who carried these two variants had taken anti-hyperglycemic drugs that affect plasma adiponectin levels, the plasma adiponectin levels in these patients were not measured. The decreased adiponectin levels in these patients were suspected as implied from the observation that the novel R55H and R131H variants exhibited the same characteristics as the reported G90S and R112C variants, respectively, which were found to be associated with hypoadiponectinemia 
(). It should be noted that the patients who carried these two novel variants exhibited greater waist circumferences than the average value of the patients with T2D (), pointing out the clinical characteristics that associate with adiponectin abnormalities. These variants were found among T2D patients, but not in a sample of non-diabetic subjects. Collapsing these two functional variants together for analysis did not show significant association with T2D (p>0.05, data not shown). Additionally, an association between these variants and T2D cannot be determined with certainty because of the low power associated with the small sample size and the low frequencies of these variants.
Clinical and laboratory characteristics of studies subjects.
The adiponectin collagenous domain is crucial for stabilizing of adiponectin trimers and further formations of hexamers and HMW multimers via interdisulfide bonds 
. The R55H, G84R, and G90S variations did not disturb adiponectin trimeric and hexameric formations but obstructed their multimerization. These variants are not close to interdisulfide bond forming site (Cys36
) and they were still capable of forming hexamers. However, they might cause conformational change and conceal the remaining free thiol from interacting with other hexamers. Another possible mechanism responsible for multimerization defect might be due to ineffectiveness of these variants in interacting with ER chaperones, such as ERp44 
or Ero1-α 
, which assist the formation of adiponectin multimers. It is remarkable that molecular sizes of the R55H and G90S variants were higher than that of the wild-type, implying a gain-of-glycosylation which has been reported as a characteristic of rare mutations implicated in several diseases 
. However, whether R55H and G90S are gain-of-glycosylation variants, and exactly what is the precise mechanism responsible for their multimerization defects, are still to be elucidated.
Adiponectin globular domain is crucial for initiation of trimer formation through hydrophobic interaction. Variations occurred at the globular domain may impair adiponectin trimerization affecting protein secretion. These included the previously reported R112C and I164T 
as well as the novel R131H variants. We found that protein aggregation exhibited by the R112C and R131H variants disappeared in the presence of a reducing agent (data not shown). These findings suggested disulfide-linked aggregation of these two variants. Generally, protein folding is determined by hydrophobic interactions and hydrogen bonding. The breaking of H-bonds between R112C monomers and the alteration of H-bond forming patterns within R131H monomers, as suggested by in silico
modeling and mutagenesis, may obstruct the formation of adiponectin trimers. The resulting free thiol may cause incorrect disulfide cross-linking that participates in protein aggregation 
While the R55H and G90S variants did not interfere with wild-type protein in the forming of HMW multimers, we showed that the R112C and R131H variants could form aggregated complexes with the wild-type protein, resulting in decreasing wild-type protein secretion. However, some forms of hetero-oligomers between the wild-type and mutant proteins could be secreted from the cultured HEK293T cells. These secreted hetero-oligomers might be predominated with the wild-type protein.
It is known that proteasomal and lysosomal systems are the two major pathways for intracellular degradation of the misfolded and unassembled proteins. Abnormal proteins, which are destined to proteasomal degradation, are usually retained within ER before being degraded by the proteasome pathway. However, the R112C and R131H variants were not found to be accumulated in ER; they trafficked to Golgi, TGN, and even endosomes but could not be secreted. Therefore, we hypothesized that the lysosomal system should play a role in the degradation of these two variants. The increases of both intracellular and extracellular levels of these two variants in the presence of lysosomal inhibitors supported our hypothesis. Nonetheless, the increase in intracellular levels of wild-type proteins by lysosomal inhibitors was probably attributable to a decrease in its secretion, rather than prevention from its degradation. Consistent with this study, Blumer et al.
have demonstrated that the acidotropic agents inhibited secretion of endogenous adiponectin from adipocytes by interrupting an acidic interior of vesicular pathway and showed that the lysosomal system was not responsible for degradation of endogenous (or wild-type) adiponectin 
. It is possible that misfolded or unassembled variants, but not wild-type adiponectin, are degraded by the lysosomal system, and inactivation of lysosomal compartments allows these abnormal proteins to escape from the degradation system to be secreted into the culturing media.
We have shown that inhibition of protein degradation by MG132 could promote secretion of the R112C, R131H, and wild-type adiponectins. In the same way, MG-132 treatment of mdx mice rescued the expression and membrane localization of dystrophin, the protein which is absent in the skeletal muscle of Duchenne muscular dystrophy (DMD) patients and mdx mice 
. This finding is also similar to the increased secretion of the mutant Z form of α1-antitrypsin (α1-AT) which is responsible for greater than 95% of individuals with α1-AT deficiency syndrome by inhibition of intracellular degradation using proteasomal inhibitors 
. In addition, proteasome inhibitions also increased the secretion of active BACE, an aspartic protease involved in the production of toxic peptide accumulation in the brain of Alzheimer patients 
. In most of the cases, it is likely that inhibition of proteasome activities decreases the opportunity of misfolded proteins retained in ER to be degraded, thereby prolonging their time in the ER lumen, and increasing the chance to be retrieved by vesicular trafficking. However, the R112C and R131H variants can traffic from ER to Golgi, trans-Golgi network, and early endosome but can not be secreted, and the lysosomal, rather than proteasomal, system is responsible for their degradations. Thus, MG132 may rescue their secretion by other mechanisms. One possibility is that MG132 elevates a number of ER chaperones necessary for protein folding 
. There is evidence indicating that misfolded proteins can be refolded successfully by modifying the molecular chaperone environment using proteasome inhibitors, including MG132 
. For example, enzymatic function of a cystathionine β-synthase (CBS) variant, T262M, could be rescued by MG132 and was not associated with increased levels of CBS T262M variant. It has also been shown that the rescuing effect of MG132 was accompanied by elevation of Hsp70 chaperone content but was not due to inhibition of mutant protein degradation 
. Thus, it is likely that MG132 may promote secretion of the R112C and R131H variants through its effect on molecular chaperones in ER.
In summary, we have identified novel candidate variants for T2D in the ADIPOQ gene in the Thai population. R55H variant impairs protein mutimerization while the novel R131H variant causes secretion defect in a dominant negative fashion. We have also demonstrated that secretion defect is unlikely to result from retention of the variant proteins in ER and the lysosomal compartment plays a role in degradation of these variants. We have also illustrated that MG132 can enhance wild-type protein secretion and rescue adiponectin variants from secretion defect. This suggests the benefit of using protease inhibitors or other compounds that modify the chaperone environment in ER as therapeutic agents for the treatment of patients carrying adiponectin variants with secretion defect.