Discussion

In this study, three main findings arose. First, four structural modifications to glabridin – 3’’,4’’ double bond hydrogenation, C-2’ hydroxy group preservation, C-4’ etherification, and enantiomerization – enhanced anti-obesity efficacy and chemical stability. Second, the optimized analog HSG4112 displayed robust dose-dependent effects on adiposity and fatty liver in HFD-induced obese mice. Third, HSG4112’s mode of action appears to be partially reduction in food intake and mainly increase in energy expenditure. These findings contribute to understanding the structure-activity relationship of glabridin and its anti-obesity effects, and of HSG4112’s preclinical efficacy on obesity and its potential mode of action.
HSG4112 as a synthetic analog superseded glabridin and is a distinctive new chemical entity. Evidence exists in literature to explain its superiority. Dehydrogenation at the C-3’’,4’’ double bond could have increased its activity because the consequent structural flexibility allows the resorcinol group to interact with the compound’s target or targets more effectively [15]. Ethoxylation at the 4’ carbon, as Bae et al. (2020) suggests, may have improved drug bioavailability and stability within the body; metabolic clearance likely happens through the glucuronidation of the C-4 hydroxy group in glabridin [34], which is absent in HSG4112. Still, the exact biochemical mechanisms of the four modified components are unknown. Examining the unanswered questions of our SAR study may provide additional drug compounds and aid in understanding the molecular mechanisms of glabridin and HSG4112.
The anti-obesity effect of HSG4112 in HFD-induced obese mice is striking, especially at 100 mg·kg-1 dose, where all examined parameters were fully normalized. Increased energy expenditure appears to be the main mode of action of HSG4112, while appetite control also plays a notable role. The mean daily food intake was significantly reduced starting from HSG4112-10 mg·kg-1 group but was not further significantly reduced in HSG4112-30 mg·kg-1 and 100 mg·kg-1  groups. Dose-dependency was not as clearly observed in food intake as other obesity-related parameters, suggesting that appetite control is not the main effect of the drug, or that appetite control and energy expenditure occur through two different mechanisms where the former reaches maximum efficacy at a low-dose level and the latter increases dose-dependently in efficacy up to the high-dose level. HSG4112-30 mg·kg-1 had almost equivalent reduction of food intake as HSG4112-10 mg·kg-1 but noticeably greater effect on body weight, fat mass and adipocyte size, and serum markers of obesity; therefore, the energy expenditure-enhancing effect seems to be at play starting at this dosage, and indirect calorimetry data at this dosage would further support this hypothesis. While muscle is mainly responsible for expending energy, only gastrocnemius muscle mass was measured in this experiment; a more comprehensive analysis of the body composition of lean and fat mass through methods like dual-energy X-ray absorptiometry (DEXA) scan [35] will be of great benefit. 
As qRT-PCR was performed on 68 select genes, unbiased RNAseq at an early timepoint before weight reduction could further benefit transcriptomic analysis and interpretation of primary or causal signals induced by HSG4112. Furthermore, while phosphorylated hypothalamic AMPK levels were tested, AMPK levels in peripheral tissues – specifically liver, muscle, and fat – and in respective cells should be measured for further confirmation of the activation of AMPK signaling. Additionally, transcriptomic analysis on white adipose tissue would improve interpreting signals or changes related to energy metabolism, leptin signaling, and inflammation. Still, within our current approach, we observed consistent gene regulation patterns in the direction towards enhanced energy expenditure in the liver and muscles: upregulation of fatty acid oxidation, lipid metabolism, and glucose metabolism. This is consistent with signals induced by exercise; the observed increase in PDK4 and PGC-1ɑ levels in muscle tissue similarly occurs after exercising [36,37]. Because PGC-1ɑ regulates mitochondrial biogenesis and AMPK is known to mediate mitochondrial fission [38], HSG4112 may act to improve mitochondrial function or dynamics. On the other hand, evidence of unaffected UCP1 mRNA and protein levels as well as unchanged body temperature in animals suggest that HSG4112 increases energy expenditure in a UCP1-independent manner. It remains unknown whether the above transcriptomic changes are primary or secondary, and such investigation merits attention. 
The molecular targets of HSG4112 have not been unraveled. However, along with the assumption that the C-4 hydroxy group delays metabolic clearance and the C-2 hydroxy group is the active pharmacophore for both HSG4112 and glabridin, putative targets of HSG4112 may coincide with known or putative targets of glabridin. One potential target is the peroxisome proliferator-activated receptor gamma (PPAR-γ) protein, which is considered the master regulator of adipogenesis, is involved in macrophage inflammatory response, and has been a major target for the pharmacological treatment of type 2 diabetes (T2DM) and obesity [39,40]. Glabridin has been reported to show significant PPAR-γ-binding activity [41] and to upregulate the PPAR-γ mRNA level in HFD-induced obese mice after 8-week administration [42]. Another potential direct target is the paraoxonase 1 (PON1) protein, which is an anti-atherogenic enzyme forming part of the circulating HDL; glabridin was shown to directly interact with recombinant PON1 to reduce linoleic acid-induced oxidation [43]. Potential targets for HSG4112’s appetite-control effect remain elusive; glabridin is reported to have in vivo neuroprotective effects of elevating antioxidants – superoxide dismutase and reduced glutathione – and inhibiting effects on staurosporine-induced apoptosis in vitro [44], but the molecular target mediating these neurological effects is unknown. One must note that HSG4112 may have novel or distinct targets compared to glabridin, given its structural modifications and distinctive weight loss effects. The target deconvolution of small molecules discovered through phenotypic screening remains a considerably significant challenge [45]. Novel techniques, such as mass spectrometry-based cellular thermal shift assay (CETSA), where the drug’s binding to proteins in a cell or cell lysate is gauged by the degree of their thermal shifts [46], will be a comprehensive and unbiased method useful in identifying the molecular target or targets of HSG4112.
As HSG4112 showed robust preclinical efficacy on obesity, NASH and T2DM are ideal secondary target indications. Notable results in liver histology and normalization of parameters related to glucose imbalance [23] support HSG4112’s development for NASH and T2DM, respectively. Furthermore, weight loss itself is immensely beneficial to NASH [47] and T2DM [48]. As the body weight of the HSG4112 group was fully normalized after 6-week administration, it is unclear whether all the beneficial effects are primary or secondary to weight loss; therefore, future investigation into the primary target cells for these diseases is needed. Of note, despite improvement in histological steatosis score, the liver color of the HSG4112 group was similar to that of the vehicle group, and serum triglyceride levels did not fully normalize; this suggests that ameliorating steatosis may not be HSG4112’s primary effect on the liver. Additionally, hepatocyte ballooning and fibrosis were absent in a HFD-induced obese mouse model, and liver biopsy was not done pre-to-post to account for the variability of NASH histology [49]; supplementary liver fibrosis animal models [50] or liver-specific fat accumulation model [51] will be of value in determining HSG4112’s prospect for treating NASH.  
Overall, our results show that HSG4112 is a novel compound derived from glabridin, with potent preclinical efficacy and both energy expenditure-enhancing and appetite-controlling effects. A key factor that enabled this discovery is in vivo phenotypic screening, which takes into account the holistic aspect of biological pathways, is more human-translatable, and applicable towards other relevant indications. Due to this method of approach, the exact mechanism of HSG4112 remains unknown and potentially novel; investigation into such mechanism will provide a meaningful understanding of energy metabolism and enable further development of drugs targeting metabolic diseases. Currently, HSG4112 is in phase 1 clinical trial for obesity. Human translation and proof of concept (POC) will demonstrate the potent effect of HSG4112 in treating metabolic diseases.