Detection of FFAR4/GPR120 in the Mouse Brain

Free Fatty Acid Receptor 4 (FFAR4/GPR120) is a member of a family of G-protein-coupled receptors (GPCRs) which can bind free fatty acids (1). FFAR4 is preferentially activated by long-chain saturated and unsaturated fatty acids (2), which are involved in many biological functions, including cell membrane maintenance, metabolism, insulin response and the inflammatory cascade (3). The free fatty acid binding capability of this group of receptors has made them targets of interest in diabetes and cardiovascular research (3, 4), while their involvement in anti-inflammatory signaling (5, 6) is of interest to the fields of stroke (7) and neurodegenerative disease (8). FFAR4 is highly expressed in the gastrointestinal tract, lung tissue, and to a lesser extent in the heart (9, 10) and brain (2, 11). More importantly, the exact cell-type specific expression of FFAR4 in many tissues, including the brain, is not fully described.

In the Journal of Immunology, Volume 202, Ren and colleagues reported that the expression of FFAR4 was increased in neurons and microglia after focal ischemic brain injury in mice (12). Their findings regarding FFAR4 protein localization and abundance were based on immunofluorescence staining in brain slices and western blotting using anti-GPR120 (SAB4501490) from Millipore Sigma. However, GPCR antibodies are notoriously unreliable, and based on many previous studies with multiple GPCR antibodies, current recommendations suggest genetic confirmation of antibodies using knock-out mice (13). Therefore, we used a whole-body constitutive FFAR4 knockout mouse (FFAR4-KO) to validate the specificity of this FFAR4 antibody. FFAR4-KO mice were established from the KOMP Repository (C57BL/6N-Ffar4^{tm1(KOMP)Vlcg}/MbpMmucd) and backcrossed to C57BL/6J mice to generate congenic FFAR4-KO mice on the C57BL/6J genetic substrain. Our results revealed a lack of antigen specificity, calling into doubt the validity of observations made using this anti-GPR120 rabbit antibody.

The expected molecular weight of FFAR4 is approximately 42 kDa based on publicly available databases (https://www.genecards.org/cgi-bin/carddisp.pl?gene=FFAR4). The product information available on the antibody manufacturer website displays double ~38 (main) and 42 (minor) kDa western blot bands in colon epithelial LOVO cells (https://www.sigmaaldrich.com/US/en/product/sigma/sab4501490). Using brain protein lysates, as well as lysates from cultured microglial BV2 and neuronal PC12 cell lines, Ren and coworkers identified multiple bands at 52 kDa (main, in brain tissue), below 52 kDa (minor, in PC12 cells) and above 52 kDa (minor, in BV2 cells). It is worth noting that the authors did not observe a band at 42 kDa, where it is expected by the manufacturer. We also recognize that this variation in electrophoretic migration could potentially be caused by additional post-translational modifications, the presence of crosslinked proteins and protein cleavage or degradation.

To verify these results, we collected extracellular, intracellular, and membrane-bound brain lysates from young adult (2–4 months of age) C57BL/6J wild-type (WT) and FFAR4-KO mouse brains. Samples were analyzed by polyacrylamide gel electrophoresis (PAGE) for western blotting analysis and transferred proteins were immunostained for actin and FFAR4 (Fig. 1A) to test the specificity of the FFAR4 antibody as previously described by our group (14–18). Testing with the same rabbit SAB4501490 antibody using protein lysates from WT and FFAR4-KO mouse brains revealed three clear immunoreactive bands of ~27, 40 and 70 kDa but no differences between protein extracts across genotypes, including in lysates enriched for membrane-bound proteins such as GPCRs. The genotypes of the mice were confirmed via RT-qPCR and subsequent agarose gel electrophoresis, which showed the expected lack of FFAR4 mRNA in the knock-out animals (Fig. 1B). Thus, these results indicated that SAB4501490 was not binding to its predicted antigen under denaturing PAGE conditions.

Figure 1. Characterization of SAB4501490 antibody in wild-type and FFAR4-KO mice.

(A) Western blot of fractionated forebrain tissue lysates from WT and FFAR4-KO mice, stained with anti-FFAR4 (SAB4501490) and anti-Actin. Arrowheads indicate the location of immunoreactive bands. EC, extracellular. IC, intracellular. MB, membrane-bound. (B) RT-qPCR was used to quantify FFAR4 expression in FFAR4-KO mouse brains compared to WT mice (left). An agarose gel with the resulting PCR products is also shown (right). (C) Representative confocal images of WT and FFAR4-KO mouse brains immunofluorescently labeled with anti-Synaptophysin (SYN; magenta) and anti-FFAR4 (SAB4501490; green). White dashed boxes indicate hippocampal (HPP) and cortical (CTX) areas of interest. Scale bar 500 μm.

Recognizing that the SAB4501490 antibody might differentially detect its antigen in native conditions, we further tested the specificity of this rabbit antibody by performing immunofluorescence staining in brain sections of WT and FFAR4-KO mice followed by confocal analysis. Briefly, sagittal brain sections were labeled with an antibody for synaptophysin (Synaptic Systems, Cat# 101004) which served as an internal control and with SAB4501490. We observed that the presynaptic marker synaptophysin was detected equally well in both WT and FFAR4 mice, readily labeling synapses. SAB4501490 appeared to mainly label synapses as exemplified by the pattern observed in the hippocampus of WT and FFAR4-KO mice (Fig. 1C). In both genotypes, nearly all pyramidal neurons in the hippocampus were not labeled with SAB4501490 as opposed to deep layer 6 cortical cells within the orbital and motor cortices (Fig. 1C). Given the undistinguishable patterns observed between WT and FFAR4-KO mouse brains, these results indicated that the alleged anti-FFAR4 SAB4501490 rabbit antibody was not specific to FFAR4.

We subsequently tested another commonly cited FFAR4 antibody, SC-390752 (Santa Cruz BioTechnology, mouse monoclonal) to determine its ability to differentially detect FFAR4 in WT and FFAR4-KO brains by western blotting and immunofluorescence. Using the same forebrain lysates, multiple immunoreactive bands were detected and we found that SC-390752 was unable to differentiate WT from FFAR4-KO extracts (Fig. 2A). Following immunolabeling, we observed that SC-390752 was also unable to differentiate WT from FFAR4-KO brain sections (Fig. 2B). By contrast, immunostaining for synaptophysin readily distinguished synaptic fields within the hippocampus and cortical areas.

Figure 2. Characterization of SC-390752 antibody in wild-type and FFAR4-KO mice.

(A) Western blot of fractionated forebrain tissue lysates from WT and FFAR4-KO mice, stained with anti-FFAR4 (SC-390752) and anti-Actin. Arrowheads indicate the location of immunoreactive bands. EC, extracellular. IC, intracellular. MB, membrane-bound. (B) Representative confocal images of WT and FFAR4-KO mouse brains immunofluorescently labeled with anti-Synaptophysin (SYN; magenta) and anti-FFAR4 (SC-390752; green). White dashed boxes indicate hippocampal (HPP) and cortical (CTX) areas of interest. Scale bar 500 μm.

Since FFAR4 could be differentially expressed across brain regions, we performed TRIzol^™-based extractions for RNA and protein from the same hippocampal and cortical brain tissue of WT and KO mice. Using hippocampal protein lysates, both SAB4501490 (Fig. 3A) and SC-390752 (Fig. 3B) produced similar patterns in FFAR4-KO and WT tissue, despite the lack of FFAR4 transcripts in the FFAR4-KO hippocampi (Fig. 3C). Similarly, western blotting using cortical protein extracts revealed multiple bands when stained with SAB4501490 (Fig. 3D) or SC-390752 (Fig. 3E), despite the absence of FFAR4 mRNAs in the FFAR4-KO cortical tissues (Fig. 3F). Actin was used as internal control for biochemical analyses and showed no overt differences across genotypes. Thus, these results indicate that neither SAB4501490 nor SC-390752 are binding to their predicted antigen under denaturing SDS-PAGE conditions or via immunocytochemistry.

Figure 3. Characterization of SAB4501490 and SC-390752 antibodies in wild-type and FFAR4-KO hippocampus and cortex.

(A, B) Western blot of TRIzol^™-treated, microdissected hippocampus from WT and FFAR4-KO mice, stained with anti-FFAR4 SAB4501490 (A) and SC-390752 (B). Actin was used as internal control. Arrowheads indicate the location of immunoreactive bands. (C) RT-qPCR was used to quantify FFAR4 expression in microdissected hippocampus, and an agarose gel with the resulting PCR products is shown. The expected amplicon for FFAR4 is 223 base pairs. (D, E) Western blot of TRIzol^™-treated, microdissected cortex from WT and FFAR4-KO mice, stained with anti-FFAR4 SAB4501490 (A) and SC-390752 (B). Actin was used as internal control. Arrowheads indicate the location of immunoreactive bands. (F) RT-qPCR was used to quantify cortical FFAR4 expression, and an agarose gel with the resulting PCR products is shown. The expected amplicon for FFAR4 is 223 base pairs.

The non-specificity of the SAB4501490 antibody calls into question whether the observed staining in the mouse brain sections and cell line cultures by Ren and coworkers is truly FFAR4, a related GPCR family member, or non-specific background signal. It is worth noting that Ren and coworkers rigorously used three distinct siRNAs (no. 1, no. 2 and no. 3) to lower FFAR4 expression in BV2 cells and siRNA no. 1 in PC12 cells and they reported ~50% reductions in FFAR4 protein abundance by western blotting using the SAB4501490 antibody in these in vitro studies. By contrast, our own observations indicate that the SAB4501490 antibody is not detecting the endogenous FFAR4 in mouse brain tissue expressed in WT mice when compared to FFAR4-KO animals, so our findings appear inconsistent with that from Ren and colleagues (2019). While lot-to-lot variation can occur in antibody manufacturing and could lead to divergent results, the similarity in the immunoreactive bands observed by western blotting between the SAB4501490 and the SC-390752 antibodies indicate the issue could be more general. Antibodies from both manufacturers were raised against human FFAR4 and the western blot bands are similarly non-specific. Our RT-qPCR analysis of the same tissue lysates used in the western blots indicates that though FFAR4 RNA is expressed in WT but not KO brains, confirmation using antibody staining is unreliable. Our findings are in agreement with previous reports of GPCR antibody unreliability (18). Consequently, the claims regarding the co-localization of FFAR4 to specific cell types in the brain should be re-evaluated with in-situ hybridization or spatial transcriptomics in combination with genetic controls.

In conclusion, our data indicate that both the Millipore-Sigma SAB4501490 antibody and the Santa Cruz BioTechnology SC-390752 antibody for FFAR4 fail to detect a specific signal either via western blot or immunofluorescence staining. We suggest caution be used when interpreting results with these antibodies, and recommend that knock-out or knock-down model validation should be conducted whenever possible. Ultimately, more studies are necessary to determine the cell types and extent to which FFAR4 is expressed in the brain.

Methods

Experimental animals.

Animals were 2–4 month-old C57BL/6J wild-type (WT) and FFAR4-KO (C57BL/6N-Ffar4^{tm1(KOMP)Vlcg}/MbpMmucd) purchased from the KOMP Repository, UC-Davis (Davis, CA, USA) and backcrossed to C57BL/6J mice for more than 15 generations to generate congenic FFAR4-KO mice on the C57BL/6J genetic substrain (n=2–3 per group, per experiment). All animal studies were approved by the University of Minnesota Institutional Animals Care and Use Committee and Institutional Review Board.

Animals were anesthetized, then intra-cardially perfused using cold-PBS. One brain hemisphere was formalin-fixed for immunofluorochemistry, and the other was flash frozen for protein extraction.

Protein extractions and SDS-PAGE analysis.

Briefly, microdissected hippocampus and cortex were subjected to TRIzol^™ solubilization (Ambion, cat# 15596018), mixing, and centrifugation to separate protein and RNA fractions from the same sample (19). Forebrain tissue from frozen hemispheres was dissociated in lysis buffers of increasing strength and centrifuged to separate intracellular, extracellular and membrane bound proteins, as described previously (14).

Forebrain lysates were immunodepleted using Protein A and Protein G Mag Sepharose^™ Xtra beads (GE Healthcare, cat # 28-9670-62 and 28-9670-70) and total protein was determined by Protein BCA Assay (Thermo Scientific, cat# 23225). 50 ug of protein was re-suspended in 4X tricine loading buffer, boiled for 5 min, and loaded on 10.5–14% Tris-HCl gels (BioRad, cat# 3459949). Proteins were transferred to 0.2 μm nitrocellulose membrane and blocked using EveryBlot Blocking Buffer (BioRad, cat# 12010020) for 5 min. Anti-FFAR4 (SAB4501490, Millipore Sigma, 1:1000) and anti-Actin (EMD Millipore, cat# MAB1501, 1:10,000) or Anti-FFAR4 (Santa Cruz, cat# SC-390752, 1:100) and anti-Actin (Sigma, cat# A2066, 1:10,000) were added for overnight incubation in block at 4°C. Secondary antibodies were LI-COR Biosciences IR dyes (LI-COR, cat # 926–68071 and 926–32210), and blots were revealed on the LI-COR Biosciences Odyssey imaging system.

Immunofluorescence labeling and confocal imaging.

Formalin fixed hemispheres were washed, immersed in ice-cold PBS and cut into 30 μm sagittal sections using a Vibratome (Leica). Slices then underwent free-floating immunofluorescent staining. Briefly, slices were permeabilized for 2 hours at room temperature using 0.3% Triton-X in PBS, blocked for 1 hour using 10% donor goat serum in PBS with 0.3% Triton-X, Anti-FFAR4 (SAB4501490, Millipore Sigma, 1:500 or Santa Cruz BioTechnology, cat# SC-390752, 1:100) and anti-Synaptophysin (Synaptic Systems, Cat# 101004, 1:500) were incubated at 4°C overnight in blocking solution. Secondary antibodies were Alexa Fluor^™ goat anti-rabbit 488 (Invitrogen, cat# A11034, 1:500) or goat anti-mouse 488 (Invitrogen, cat# A32723, 1:500) & goat anti-guinea pig 405 (Abcam, cat# 175678, 1:500). Slices were treated for autofluorescence using 1% Sudan black, mounted on Superfrost Plus slides (Fisher Scientific, cat# 12-550-15) and cover slipped using ProLong^™ Diamond Antifade Mountant (Invitrogen, cat# P36961). Confocal z-stack tile images were acquired using a STELLARIS 8 microscope (Leica). Images were analyzed using Imaris 9.3 software (Bitplane Scientific Software, USA).

Gene Expression.

RNA was isolated from brains using TRIzol^™ and ethanol, and purified using the RNeasy Lipid Tissue Mini Kit from Qiagen. cDNA was made from RNA using Quantabio’s cDNA synthesis kit. RT-qPCR was performed with Bio-Rad’s iTaq Universal SYBR Green SuperMix or with LightCycler^®480 SYBR Green I Master reaction mix on a Bio-Rad CFX96^™ Real-Time PCR System. Primer sequences from 5’ to 3’ were: CGGCGGGGACCAGGAAAT and GTCTTGTTGGGACACTCGGA.