GPR109A mediates the effects of hippuric acid on regulating osteoclastogenesis and bone resorption in mice

The G protein-coupled receptor 109 A (GPR109A) is robustly expressed in osteoclastic precursor macrophages. Previous studies suggested that GPR109A mediates effects of diet-derived phenolic acids such as hippuric acid (HA) and 3-(3-hydroxyphenyl) propionic acid (3-3-PPA) on promoting bone formation. However, the role of GPR109A in metabolic bone homeostasis and osteoclast differentiation has not been investigated. Using densitometric, bone histologic and molecular signaling analytic methods, we uncovered that bone mass and strength were significantly higher in tibia and spine of standard rodent diet weaned 4-week-old and 6-month-old GPR109A gene deletion (GPR109A−/−) mice, compared to their wild type controls. Osteoclast numbers in bone and in ex vivo bone marrow cell cultures were significantly decreased in GPR109A−/− mice compared to wild type controls. In accordance with these data, CTX-1 in bone marrow plasma and gene expression of bone resorption markers (TNFα, TRAP, Cathepsin K) were significantly decreased in GPR109A−/− mice, while on the other hand, P1NP was increased in serum from both male and female GPR109A−/− mice compared to their respective controls. GPR109A deletion led to suppressed Wnt/β-catenin signaling in osteoclast precursors to inhibit osteoclast differentiation and activity. Indeed, HA and 3-3-PPA substantially inhibited RANKL-induced GPR109A expression and Wnt/β-catenin signaling in osteoclast precursors and osteoclast differentiation. Resultantly, HA significantly inhibited bone resorption and increased bone mass in wild type mice, but had no additional effects on bone in GPR109A−/− mice compared with their respective untreated control mice. These results suggest an important role for GPR109A during osteoclast differentiation and bone resorption mediating effects of HA and 3-3-PPA on inhibiting bone resorption during skeletal development.

The effect of hippuric acid and 33-PPA while interesting, the authors don't quite provide mechanistic information in terms of how GPR109A is involved in their actions. Clearly, they regulate GPR109A expression but in the absence of the receptor, 5% BB diet appears to still have effects in the KO and there are some partial effects with HA. As the authors state in the conclusion, it appears that HA and 33PPA somehow involve GPR109A but how exactly is not clear and is needed for making this manuscript mechanistic.

Specific major concerns:
Mice: It is not clear if the mice are littermate (WT and KO). If not, then the differences in microbiota could be due to different cages, rooms etc. Authors should specific these types of information in the methods. Also please make sure to mention if the KO mice have been backcrossed sufficient times to C57BL6 background.
Since these are growing mice, authors should also provide data on bone length.
While it is understandable that the authors focused on osteoclasts, they should provide markers of osteoblasts to provide a complete picture of the knockout mice.
In Fig 8B, as a control, WB of GPR109A should be shown for the KO mice (to show the absence).
Reviewer #3: Remarks to the Author: In this article Chen et al show that Gpr109a-/-mice have higher bone density than WT mice. They link this finding to decreased number of osteoclast and levels of TNF-a, TRAP, cathepsin K ad extracellular cAMP in the former. Additionally they show that diet derived phenolic acids, HA and 3-3-PPA inhibited the expression of Gpr109a and bone resorption in a Gpr109a-dependent manner. Overall this is an interesting study with several concerns. 1) Fig 1A. GPR109a is expressed at highest levels in adipocytes and innate immune cells (pls see various earlier publications from Dr. Offermanns and others) . In all the other tissue, it is so low that in Gpr109a-promoter driven RFP transgenic mice, Dr. Offermann's gropu could not directly visualize RFP. Therefore how come all the other tissue in Fig 1A have similar levels of Gpr109a expression as adipose tissue and RAW cells. Gpr109a is single exon gene. Almost all the RNApreps are contaminated with DNA. In this scenario, RT-PCR primers will not discriminate between DNA and RNA, and therefore will not accurately depict the expression of Gpr109a. Although authors write that RNA was depleted of DNA. To ascertain whether DNA was completely depleted they should do compare the RT-PCR with equivalent amount of RNA and cDNA. In addition, show the expression by quantitative realtime-PCR. 2) Fig 3D. Signaling through Gpr109a inhibits intracellular cAMP not extracellular cAMP as shown here. In my knowledge there is neither a report nor a rationale in the scientific literature to explain how Gi-coupled receptors influence extracellular cAMP. Fig 4B-D to show lower bone resorption by Gpr109a-/-osteocalsts. Since, cells from Gpr109a-/-mice had lower number of osteoclasts, it completely unclear whether lower bone resorption by Gpr109a-/-cells is due to lower number of osteoclasts or their impaired activity. Therefors, authors cannot interpret lower bone resorption activity by Gpr109a-/-osteoclasts in absence of any data. 4) There is no experiment to show how gut microbiota is altered or altered gut microbiota affects osteoclast or higher bone density in Gpr109a-/-mice. 5) Authors have Gpr109a-/-mice. They must authenticate the Gpr109a antibody used for various western blot experimenst such as Fig 6 and 8.  6) Mechanistically, what do the authors want to convey; Whether HA and 33-PPA binds to Gpr109a to mediate all the effects shown in manuscript or they suppress the Gpr109a expression to mediate their effects. A clear mechanistic data to discriminate between these possibilities are required.

3) Fig 4A shows development of lower numbers of osteoclasts in total non-adherent cells from BM of Gpr109a-/-mice. This non-adherent pool of cells was used in
Reviewer: Fig 1A.

GPR109a is expressed at highest levels in adipocytes and innate immune cells (pls see various earlier publications from Dr. Offermanns and others).
In all the other tissue, it is so low that in Gpr109a-promoter driven RFP transgenic mice, Dr. Offermann's gropu could not directly visualize RFP. Therefore how come all the other tissue in Fig 1A have similar levels of Gpr109a expression as adipose tissue and RAW cells. Gpr109a is single exon gene. Almost all the RNA -preps are contaminated with DNA. In this scenario, RT-PCR primers will not discriminate between DNA and RNA, and therefore will not accurately depict the expression of Gpr109a. Although authors write that RNA was depleted of DNA. To ascertain whether DNA was completely depleted they should do compare the RT-PCR with equivalent amount of RNA and cDNA. In addition, show the expression by quantitative realtime-PCR.
Author's response: GPR109A -/mice were originally from Dr. Offermanns' group, we not only checked their every publications, but also contacted them about using this mouse model for studying skeletal tissue. As the reviewer suggested, we have additionally performed real-time PCR and quantified GPR109A expression in different tissues. It is exactly as expected, GPR109A is expressed at the highest levels in adipocytes and innate immune cells relative to other tissues and cell types (normalized by GAPDH control). We have attached data in Fig 1A in the revision.
Reviewer: Fig 3D. Signaling through Gpr109a inhibits intracellular cAMP not extracellular cAMP as shown here. In my knowledge there is neither a report nor a rationale in the scientific literature to explain how Gi-coupled receptors influence extracellular cAMP.