Cell culture and downregulation using siRNA

HEK293 and HeLa cells were maintained in 1 g/L glucose DMEM (PAN-Biotech) supplemented with 10% fetal bovine serum (FBS, Capricorn Scientific), 2 mM GlutaMAX (Gibco), 1 mM sodium pyruvate (Gibco) and penstrep (PAN-Biotech, penicillin 100 U/mL and 100 μg/mL streptomycin). HepG2 cells were cultured in 1 g/L glucose DMEM (PAN-Biotech) supplemented with 10% FBS (Capricorn Scientific), 2 mM GlutaMAX (Gibco) and penstrep (PAN-Biotech). C2C12 cells were cultured in 4.5 g/L DMEM (Pan-Biotech) supplemented with 10% FBS (Capricorn Scientific) and penstrep (PAN-Biotech), whereas HAP1 cells were maintained in Iscove’s modified DMEM medium (IMDM, Sigma-Aldrich) supplemented with 20% FBS, 2 mM GlutaMAX (Gibco) and penstrep (PAN-Biotech). All the above-mentioned cells were grown at 37°C supplied with 5% CO₂. For downregulation of MIC26, the cells were transfected with Lipofectamine^™ RNAiMAX reagent (Invitrogen) according to the manufacturer’s protocol. Downregulation was done for 48 h with 10 nM of siRNA. Negative control medium GC duplex (Invitrogen, cat no: 462001) was used as control along with following siRNA sequences (Thermo Fischer Scientific):

1. MIC26 # 1 (ID30917 targeting exon 6 and 7): 5’-GGAUAUAUAGUCAUAGAAGTT-3’;
   5’-CUUCUAUGACUAUAUAUCCTC-3’
2. MIC26 # 2 (ID127368 targeting exon 3 and 4): 5’-GGAAACGUACUCCCAAACUTT-3’;
   5’-AGUUUGGGAGUACGUUUCCTG-3’
3. MIC26 # 3 (ID127366 targeting exon 2): 5’-CCUCCCAAAAAUUCCGUGATT-3’;
   5’-UCACGGAAUUUUUGGGAGGTG-3’

Generation of MIC26 and MIC27 KOs using the CRISPR/Cas method

MIC26 and MIC27 KOs in HepG2, HeLa and HEK293 cells were generated using the double nickase method. This CRISPR/Cas KO system consists of two plasmids where each plasmid contained a sgRNA sequence and a reporter (either GFP or puromycin resistance) for positive selection of cells. The plasmid mix was commercially available (Santa Cruz Biotechnology): MIC26: double nickase plasmid sc-413137-NIC and MIC27 double nickase plasmid sc-414464-NIC. Cells were seeded on 35 mm dishes and transfected one day later with 1 μg of the respective plasmid mix using GeneJuice (Novagen) reagent according to the manufacturers protocol. Two days after transfection, single cells were sorted by FACS (Beckman Coulter) using GFP fluorescence into a 96-well plate (Greiner) containing respective conditioned media to obtain homogeneous KO cell populations from single cells. HAP1 MIC26 and MIC27 KO cells were custom-made by Horizon (UK) as described before [31].

Molecular cloning/Plasmids & Constructs

MIC26-GFP was generated by cloning human MIC26 in to the pEGFP-N1 vector, using the Kpn1 and Age1 restriction enzymes which was followed by ligation. MIC26^S34A-GFP, MIC26^S41A-GFP and MIC26^S50A-GFP mutant variants of MIC26-GFP were generated using the Q5 site-directed mutagenesis kit (New England Biolabs). pMSCV-MIC26 plasmid, generated in a previous study [31], was used as a template for making pMSCV-MIC26-myc plasmid using the Q5 site-directed mutagenesis kit (New England Biolabs) according to manufacturer’s protocol with the following primers:

1. 1. myc-tag forward:
   5’-GAACAAAAACTCATCTCAGAAGAGGATCTCTAGCGAATTCTACCGGGTAG-3’
2. 2. myc-tag reverse:
   5’-CCCAGATCCCTTAGTTCCAGGTGAATTCTTCACA-3’

For generating triple mutant MIC26^{S34A/S41A/S50A}-myc plasmid, insertion of myc-tag and triple amino acid mutations from serine to alanine (S34, S41 and S50) were generated using the following primers:

1. 3. MIC26-S34/41/50A forward:
   5’-CTACTCAG-TTCCTGAGGGTCAAGCGAAGTATGTGGAGGAGGCA-3’
2. 4. MIC26-S34/41/50A reverse:
   5’-AGTGC-AAGCTCATCAACCTTCACGGCATTTTTGGGAGGTGAGTC-3’

In order to express MIC26-GFP and MIC26-myc variants, HEK293 cells were seeded onto 35 mm plates and transfected with 1 μg of corresponding plasmid using GeneJuice (Novagen) reagent according to the manufacturer’s protocol and grown for 48 h until harvest.

SDS gel electrophoresis and Western Blotting

Cellular proteins were harvested from a corresponding number of cells after washing them three times with 2 mL DPBS (PAN-Biotech) followed by scraping and resuspending the cells in an appropriate volume of lysis buffer (210 mM mannitol, 70 mM sucrose, 1 mM EDTA, 20 mM HEPES, 1 x protease inhibitor (Sigma-Aldrich)) or RIPA buffer (150 mM NaCl. 0.1% SDS, 0.05% DOC. 1% Triton-X-100. 1 mM EDTA, 1mM Tris, pH 7.4, 1 x protease inhibitor (Sigma-Aldrich)). Following the incubation of cells for 10 min on ice, they were mechanically disrupted using a 20G canula by repetitive strokes. In order to harvest secreted proteins, cells were seeded into 35 mm plates containing 2 mL standard growth medium. After 24 h incubation, cells were carefully washed three times with 2 mL DPBS (PAN-Biotech) and grown for another 24 h in the corresponding medium lacking FBS. The culture medium was collected and separated from detached cells by centrifugation for 5 min at 1000g and 4°C. Next, proteins in cell culture media were precipitated by the addition of 10% trichloroacetic acid (TCA) for 40 min on ice. The precipitated proteins were pelleted by centrifugation for 10 min at 16,000 g, 4°C and washed twice with 2 mL ice-cold acetone. The pellet was dried at room temperature, dissolved in 4 M Urea, 1 mM DTT, 2% SDS and heated for 5 min at 95°C. For subcellular fractionation, nuclear fraction was removed from cell lysates, obtained by mechanical disruption, upon centrifugation for 5 min at 500g and 4°C. An aliquot of the supernatant was taken as total fraction. Next, the mitochondrial fraction was pelleted by centrifugation at 7000g and 4°C for 10 min. Mitochondrial fraction was washed three times and sequentially resuspended in lysis buffer. The supernatant contained the ER/Golgi and cytosolic fraction. For murine tissue lysates, 20 mg of tissue was homogenized, in Precellys Soft Tissue homogenizing CK14 tubes, with 300 μL RIPA buffer in a Precellys 24 homogenizer at 5000 rpm for 10 sec. Further, solubilized proteins were separated from the tissue debris by centrifugation at 16,000 g for 10 min at 4°C. Protein concentration was determined using Lowry method (Bio-Rad). SDS samples were prepared with Laemmli buffer and heated for 5 min at 95°C. 10% or 15% SDS electrophoresis gels were used for running and separating protein samples. The proteins were blotted onto nitrocellulose membrane and detected using Ponceau S (Sigma Aldrich, P7170) following which the membrane was destained. Further, nitrocellulose membrane was blocked with 5% milk in 1x TBS-T and probed with the following primary antibodies overnight at 4°C: MIC26 (#1: Sigma-Aldrich HPA003187, 1:500; #2: Invitrogen PA5-116197, 1:1000; #3: Invitrogen MA5-15493, 1:1000; #4: Pineda home-made, 1:100), MIC27 (Sigma-Aldrich HPA000612, 1:2000), myc-tag (Cell Signaling Technology 2276, 1:1000), ß-Tubulin (Abcam ab6046, 1:2000), Calreticulin (Cell Signaling Technology 2891, 1:1000), ANT2 (Sigma-Aldrich HPA046835, 1:1000) and GFP (Sigma-Aldrich 11814460001, 1:2000). Goat IgG anti-Mouse IgG (Abcam ab97023, 1:10000) and Goat IgG anti-Rabbit IgG (Dianova SBA-4050-05, 1:10000) conjugated to HRP were used as secondary antibodies. The chemiluminescent signals were obtained using Signal Fire ECL reagent (Cell Signaling Technology) and VILBER LOURMAT Fusion SL equipment (Peqlab).

RNA isolation and quantification

Total RNA was extracted from murine liver tissue using RNeasy Mini Kit (Qiagen) according to the manufacturer’s protocol. RNA quality and quantity were assessed using BioSpectrometer (Eppendorf). cDNA synthesis from 1 μg RNA was performed using the GoScript^™ Reverse Transcriptase Kit (Promega). Next, quantitative real-time PCR was performed in Rotor Gene 6000 (Corbett Research) using GoTagR qPCR Master Mix (Promega) according to manufacturer’s instructions with the following primers:

1. HPRT1 (Housekeeping gene) forward:
   5’- CTGGTGAAAAGGACCTCTCGAAG -3’
2. HPRT1 (Housekeeping gene) reverse:
   5’- CCAGTTTCACTAATGACACAAACG -3’
3. MIC26 forward:
   5’- AGCACCCAAAAAGGACTCGCCT -3’
4. MIC26 reverse:
   5’- GGCTCACAATGATGTCGGAGTTG -3’
5. MIC27 forward:
   5’- GTCTATCTGAAGAATCCTCCGCA -3’
6. MIC27 reverse:
   5’- CAAACAGTCGCTCCTAATGTGGC -3’

C_t values were normalized to housekeeping gene HPRT1 followed by normalization of ΔC_t values to average ΔC_t of db/+ control group.

Mass spectrometry

Coomassie stained SDS-PAGE cut-outs of the ≈ 47 to 63 kDa band were processed as described before [56] with minor changes: In brief, slices were destained and washed (alternating cycles of 10 mM Ammonium bicarbonate (ABC) and 100 mM ABC/50% Acetonitrile (ACN); 3 repeats). Proteins were reduced and alkylated in-gel (10 mM DTT followed by 55 mM IAA), washed again as before and dried. Digestion was performed for 16 h, 37°C with 0.1 μg trypsin in 100 mM NH₄HCO₃. Peptides were extracted twice, using a 1:1 mixture of ACN/0.1% TFA (v/v), combined extracts were dried again and resuspended in 0.1% (v/v) TFA.

Half of each sample was used for LC-MS analysis (Ultimate 3000 rapid separation liquid chromatography system, coupled to a QExactive plus, both Thermo Fisher scientific) using the following parameters: Samples were separated on PepMap (Thermo Fisher scientific) C18 loading and separating columns, using a 2h gradient (going from 4% Buffer B to 40% in 64 minutes, followed by 95%B and 4%B equilibration respectively; Buffer A: 0,1% FA, Buffer B: 84% ACN, 0,1% FA). Data was collected using a resolution of 140k, AGC target of 3e6, maximum injection times of 50ms for MS1 and a scan range of 200-2000m/z, data dependent MS2 spectra were recorded using a resolution of 17,5k, AGC target 1e5, maximum It 50ms, NCE 30 in a TopN [20] experiment using dynamic exclusion for 10 seconds after selection.

Data Processing was performed using MaxQuant 1.6.17.0 [57], using standard parameters if not stated otherwise. Searches were performed against a database containing 79038 human proteins (UP000005640, downloaded 01/2022 from uniport.org). The following modifications were considered: fixed: Carbamidomethylation (C); variable: Oxidation (M); Acetyl (Protein N-term). Potential contaminant IDs were removed and annotation of IDs was performed using Perseus [58], due to the nature of the search, no additional filtering was applied, but results were colour coded, based on the number of peptides assigned to the respective protein IDs in each sample: red: 0 or 1 peptide, green >3 peptides.

Mouse tissues

Tissues from 12-week-old male control and db/db.BKS mice (#000642, Jackson Laboratories, USA) were used for this study. Animal procedures were approved by the Department for Environment and Consumer Protection of North Rhine-Westphalia, Germany (LANUV; #81–02.04.2019.A321) and the DDZ Institutional Animal Welfare Committee.

The presumed 55 kDa glycosylated protein is not decreased in cells depleted for MIC26 and MIC27

Previous studies have either focused on the role of MIC26_55kDa glycosylated protein in the context of diabetes and fatty acid metabolism [43,44,47,51,52,54] or MIC26_22kDa protein in mitochondrial function [31,46,48,59]. Mitochondrial function and diabetes are intricately connected [60]. Therefore, in order to understand their relationship, we investigated the functional link between MIC26_22kDa mitochondrial and MIC26_55kDa glycosylated protein. For this, we first focussed on the downregulation of MIC26 in HEK293 cells (Fig 1A). Western Blots (WBs) consistently revealed a marked reduction of MIC26_22kDa protein upon downregulating MIC26 with three different siRNA, targeting different exons, resulting in negligible amounts of protein levels compared to control siRNA demonstrating the expected sensitivity of MIC26_22kDa protein to MIC26 downregulation. On the contrary, we found no differences in the levels of the MIC26_55kDa protein (Fig 1A, uppermost panel) indicating that the MIC26_55kDa protein is not affected upon downregulation of MIC26. Upon revisiting previous literature, we found two reports showing similar unresponsive nature of the MIC26_55kDa glycosylated protein upon downregulation of MIC26 using multiple siRNA in HeLa cells and shRNA in 143B cells [40,59]. However, this was not further explored. Overall, only the mitochondrial MIC26_22kDa protein was responsive while the MIC26_55kDa was resistant to MIC26 downregulation leading us to question the specificity of the MIC26_55kDa glycosylated protein.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The MIC26_55kDa form is neither sensitive to knockdown nor knockout of MIC26.

A) Western blot (WB) analysis of HEK293 cells knocked down for MIC26 reveal the loss of mitochondrial MIC26_22kDa but not MIC26_55kDa. In order to downregulate MIC26, three siRNAs were used. B)–E) WB analysis of cell lysates and secreted proteins, using WT and MIC26 KO HepG2 cells, were done using four anti-MIC26 antibodies. All four WBs show the presence of intracellular mitochondrial MIC26_22kDa form in WT cells but not in MIC26 KO cells as expected. However, unexpected and inconsistent results were obtained w.r.t the 55 kDa protein. (B) Antibody #1 detects an intra- and extracellular 55 kDa band in WT as well as MIC26 KOs. (C) Antibody #2 detects only an intracellular protein band around 70 kDa. (D) Antibody #3 identifies an intracellular and secreted protein (in culture medium) at ≈ 70 kDa in both WT and MIC26 KO HepG2 cells. The considerable increase in ≈ 70 kDa protein correlates to the production and secretion of albumin by human HepG2 cells, which is nonspecifically recognized by this antibody. (E) Antibody #4 is not recognizing a 55 kDa band in WT and MIC26 KO HepG2 cells.

https://doi.org/10.1371/journal.pone.0286756.g001

The previous reports on MIC26_55kDa glycosylation had not employed any MIC26 KO cell line [43,44,47]. One of our studies used MIC26 and MIC27 KOs in HAP1 cells but rather focussed on mitochondrial MIC26_22kDa and the relationship between mitochondrial ultrastructure and functioning [31]. In order to further explore the unresponsive nature of the MIC26_55kDa protein to MIC26 downregulation, we generated MIC26 KO in HepG2 cells, using the CRISPR/Cas system, which have appreciable secretory properties and where the MIC26_55kDa was described to be secreted [43]. Using WBs, we examined the cell lysates as well as culture media containing the secretory proteins of WT and MIC26 KO HepG2 cells (Fig 1B) to check for the presence of both the intracellular and secreted MIC26_55kDa, respectively. Further, in order to corroborate our previous results, we used a set of four anti-MIC26 antibodies available to us. For clarity, we compiled the details of various anti-MIC26 antibodies used in this study and other studies (S1A Fig). The antigens used to elicit antibody response are also shown with different colour-codes from various studies (S1B Fig). All four antibodies consistently detected the human mitochondrial MIC26_22kDa in WT HepG2 cell lysates while MIC26_22kDa was not observed in MIC26 KOs as expected (Fig 1B–1E). We observed the putative MIC26_55kDa secreted protein in culture media of WT as well as MIC26 KO HepG2 cells when antibody #1 was used (Fig 1B). Moreover, antibody #1 recognised the presumed intracellular 55 kDa protein (Fig 1B) while a protein band having a molecular weight larger than 55 kDa was detected when antibodies #2 and #3 were used (Fig 1C and 1D). Antibody #2 did not reveal any secreted proteins around 55 kDa while the nearest intracellular protein detected had a molecular weight well above 55 kDa (Fig 1C). Using antibody #4, neither an intracellular nor secreted protein was found in the 55 kDa region (Fig 1E). Thus, the inconsistent detection of proteins of varied molecular weight around 55 kDa, in accordance with no loss of the 55 kDa protein in HepG2 MIC26 KOs, with different anti-MIC26 antibodies indicates that MIC26_55kDa protein is of nonspecific nature and that MIC26 most likely exists in the mitochondria as a 22 kDa protein across many cell types.

In order to check the mitochondrial localisation of MIC26 using all four antibodies, we isolated mitochondria from HEK293 cell lysates and performed immunoblotting (S2 Fig). As expected, we consistently found that MIC26 was only present in the mitochondrial fraction using all antibodies #1–4 (S2A–S2D Fig). ANT2, a mitochondrial protein was only found in the mitochondrial fraction as expected. We also loaded the fraction which contain the endoplasmic reticulum (ER), Golgi and the rest of cytosolic fraction. While calreticulin, an ER protein, was detected in this fraction as expected, mitochondrial MIC26 was not observed showing that mitochondrial MIC26 is not present in the secretory fraction (S2A–S2D Fig).

A nonspecific 55 kDa band is still recognised in knockouts of MIC26 and MIC27 in various human cell lines

In order to ascertain if the 55 kDa protein belongs to either MIC26 and/or MIC27 and to overcome cell-type specific effects, we employed KOs of MIC26 and MIC27 in various cell lines. During the course of this study, we generated corresponding single KOs of MIC26 and MIC27 in HEK293, HeLa and HepG2 cells using the CRISPR/Cas system. Additionally, we used MIC26 and MIC27 KO HAP1 cells along with WT cells already available from our previous study [31]. Altogether, we used KOs of MIC26 and MIC27 belonging to four different human cell lines (Fig 2A–2D). Since the 55 kDa MIC26 protein was consistently detected in mammalian cell lysates and cell culture medium using only one antibody (#1) out of four anti-MIC26 antibodies available (Fig 1B–1E), we performed subsequent experiments with WT, MIC26 and MIC27 KOs using antibody #1 (Fig 2A–2D). WBs, using lysates from HEK293 (Fig 2A), HepG2 (Fig 2B), HeLa (Fig 2C) and HAP1 (Fig 2D) cells, revealed that the respective MIC26 and MIC27 proteins were absent when the corresponding genes were knocked out as expected when compared to respective control cells. On the contrary, no loss of 55 kDa protein was observed when using an antibody directed against MIC26 (Fig 2A–2D, topmost panels). Similarly, immunoblotting using an anti-MIC27 antibody does not show a protein in the 55 kDa region in WT cells as well as when cells were knocked out for MIC27 (Fig 2A–2D and S3A–S3D Fig). In conclusion, the 55 kDa protein belongs neither to MIC26 nor MIC27 confirming that MIC26 and MIC27 only exist as bona fide mitochondrial proteins.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. MIC26 KOs reveal the nonspecific nature of the 55 kDa band in various human cell lines.

WB analysis of WT, MIC26 and MIC27 KO cell lysates, obtained from HEK293, HepG2, HeLa and HAP1 cells are shown. MIC26 KOs reveal a loss of mitochondrial MIC26_22kDa but not the 55 kDa band. There is no protein band detected in WT and MIC27 KOs at 55 kDa region when anti-MIC27 antibody was used. A) In MIC26 KOs of HEK293 cells, a loss of MIC26_22kDa is observed contrary to an increased amount of 55 kDa protein in WT. B) The KO of MIC26 in HepG2 cells leads to the loss of the 22kDa form but unchanged expression of the 55 kDa band. C) HeLa cells show similar results to HEK293 cells w.r.t loss of the 22 kDa form and an increased expression of the 55 kDa band. D) In HAP1 cells, the MIC26 KO leads to a loss of the mitochondrial MIC26_22kDa protein contrary to unaltered expression of the 55 kDa protein in WT cells.

https://doi.org/10.1371/journal.pone.0286756.g002

Mutation of various predicted glycosylation sites leads to no change in the level of 55 kDa protein

The experiments involving downregulation (Fig 1A) and KOs of MIC26 and MIC27 in multiple human cell lines (Fig 2) using antibody #1 showed that the 55 kDa protein is nonspecific. Further, in order to rule out the possibility that the glycosylation of MIC26 is a possible reason why all four antibodies employed against MIC26 do not recognise the MIC26_55kDa protein, we used WT HEK293 cells exogenously expressing MIC26-GFP and probed with an anti-GFP antibody which must recognise GFP of the MIC26_55kDa-GFP fusion protein. If the MIC26_55kDa protein exists, exogenous expression of MIC26-GFP in control HEK293 cells should lead to the formation of 82 kDa protein in addition to the untagged endogenous MIC26_55kDa protein. When HEK293 cells expressing MIC26-GFP were probed with an anti-MIC26 (antibody #1) and anti-GFP antibody, a 47 kDa protein (Red asterisks) consistent with the size of mitochondrial MIC26-GFP was observed while no additional band at 82 kDa was detected (Fig 3A and 3B). This provides additional evidence that MIC26 exists only as a mitochondrial protein. In addition, we checked if the presumed MIC26_55kDa was sensitive to the disruption of glycosylation pattern. Here, it was predicted in silico that three serine amino acid residues of MIC26 could be glycosylated at positions 34, 41 and 50 [46]. Thus, we separately mutated the three serine amino acid residues to alanine to render them incapable of glycosylation. Then, we expressed the theoretically predicted single mutant non-glycosylated versions of MIC26-GFP namely MIC26^S34A-GFP, MIC26^S41A-GFP and MIC26^S50A-GFP in control HEK293 cells and used an anti-MIC26 and anti-GFP antibody to check the size of MIC26-GFP (Fig 3A and 3B). While a 47 kDa protein comprising mitochondrial MIC26-GFP was detected (Fig 3A and 3B) (Red asterisks), a protein corresponding to 82 kDa (55 kDa protein tagged with GFP) was not found (Fig 3A and 3B). When HEK293 cells expressing either MIC26-GFP or the glycosylation defective mutants were further probed with another anti-MIC26 antibody #3, we neither detected a 55 kDa nor a 82 kDa protein (S4 Fig). However, exogenously expressed MIC26-GFP or variants of MIC26-GFP were detected as a 47 kDa protein. It was interesting to note that antibody #3 did not recognise the exogenously expressed MIC26^S41A-GFP in HEK293 cells showing that serine at position 41 is an important binding site for antibody #3 (S4 Fig). Taken together, we further corroborate that MIC26 only exists as a mitochondrial protein that is not apparently glycosylated.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Exogenous expression of MIC26-GFP or MIC26-myc does not result in the formation of corresponding MIC26_55kDa-GFP or MIC26_55kDa-myc protein.

WB analysis of HEK293 cells expressing MIC26-GFP or MIC26-myc along with various theoretically predicted glycosylation defective mutants. A) Overexpression of mitoGFP (control), MIC26-GFP and the single mutants of MIC26-GFP namely MIC26^S34A-GFP, MIC26^S41A-GFP and MIC26^S50A-GFP in HEK WT cells leads to detection of the endogenous MIC26_22kDa protein, a MIC26_22kDa-GFP protein (≈ 50 kDa protein below the 55 kDa protein marked in red asterisks) and the endogenous 55 kDa band when antibody #1 was used. B) Immunoblotting with anti-GFP antibody shows a ≈ 50 kDa band detecting MIC26_22kDa-GFP. However, no band correlating to a MIC26_55kDa-GFP protein (≈ 80 kDa) could be observed. C) Exogenous expression of MIC26-myc in HEK293 MIC26 KO leads to the formation of MIC26_22kDa-myc protein with an expected slight shift in molecular weight compared to the expression of untagged MIC26. However, MIC26_55kDa-myc protein was neither detected with anti-MIC26 nor anti-myc tag antibody. Expression of the glycosylation defective triple mutant, MIC26^{S34A/S41A/S50A}-myc, where the respective serine amino acids were mutated to alanine (S34A, S41A, S50A), showed unaltered 55 kDa protein levels.

https://doi.org/10.1371/journal.pone.0286756.g003

Since the addition of a large GFP tag could probably disturb the glycosylation pattern of MIC26, we used MIC26 KOs of HEK293 cells and expressed either an untagged or myc-tagged MIC26 on its C-terminus. WBs of MIC26 KOs expressing MIC26-myc revealed a mitochondrial myc-specific form, when probed with an antibody against MIC26, that was running slightly above the untagged MIC26 protein (Fig 3C). Expression of MIC26-myc in MIC26 KOs also revealed a myc-tagged MIC26 when an anti-myc antibody was used. However, we did not observe any increase of the 55 kDa protein while expressing either untagged or myc-tagged MIC26 in MIC26 KOs. This was also in accordance with a previous report where HeLa cells expressing N-terminal myc-tagged MIC26 were used [46]. Although a mitochondrial MIC26_22kDa-myc was detected with an anti-myc antibody, no protein was detected at 55 kDa. Further, we generated a C-terminal myc tag plasmid expressing a glycosylation defective triple mutant, S34A/S41A/S50A, in MIC26 KOs where all three serine amino acids were mutated into alanine. Mutations of all three serine amino acid residues into alanine should lead to an abolished O-linked glycosylation and loss of the 55 kDa protein. However, despite mutating multiple serines into alanine, we neither observed a loss of the 55 kDa protein nor the expression of myc-tag protein in the 55 kDa region (Fig 3C). These observations show that the 55 kDa protein does not originate from MIC26.

The mitochondrial MIC26 is consistently detected in murine tissues, cells and human cells in contrast to the putative 55 kDa glycosylated version

We have performed MIC26 downregulation with three different siRNAs in HEK293 cells, probed with four different anti-MIC26 antibodies using control and HepG2 MIC26 KO cells and further tested MIC26 KOs in three other human cell lines namely HEK293, HeLa and HAP1 cells (Fig 2), and exogenously expressed MIC26-GFP and MIC26-myc in HEK293 cells (Fig 3 and S4 Fig) which consistently validated that the 55 kDa protein of MIC26 does not exist. In addition, in order to corroborate our observations across species, we compared murine liver lysates from WT C57BL6/J mice with human HEK293 and HeLa cell lysates using multiple antibodies and tested for the presence of a 55 kDa protein. WBs consistently revealed the mitochondrial MIC26_22kDa (Fig 4A and 4B, S5A and S5B Fig) reiterating that all four antibodies recognise the correct protein in human cell lines. In the respective WBs, we also loaded lysates of murine liver for comparing the immunoreactivity of anti-MIC26 antibody in human cells and murine tissue. In congruence with human cell line data, antibodies #1 and #2 detected the mitochondrial MIC26_22kDa protein in murine liver (Fig 4A & 4B). Hence, antibody #1 and #2 could be used to detect the specific mitochondrial MIC26_22kDa protein in human and murine samples. However, one must be cautious in interpreting a band with an apparent molecular weight around 55 kDa using antibody #1 (Fig 4A) as this is not specific to MIC26. Further, we used plasma from WT C57BL6/J mice and found that there is no detection of any protein of any size using anti-MIC26 antibodies #1 and #2 (Fig 4C and 4D) confirming that MIC26 is not secreted into murine plasma. In addition, in order to check whether MIC26_22kDa is recognised by both these antibodies in cells having murine origin along with other murine tissues, we used murine lysates of white adipose tissue, muscle, liver and plasma along with C2C12 myoblast cell line and found that antibody #1 and #2 consistently recognise the mitochondrial MIC26_22kDa (Fig 4E and 4F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Unspecific nature of the 55 kDa band is neither cell line- or species-specific.

WB analysis of human WT HEK293 and HeLa cell lysates along with murine liver tissue lysates and plasma comparing two different anti-MIC26 antibodies: A) Antibody #1 detects MIC26_22kDa and an unspecific 55 kDa protein in HEK293 and HeLa as well as in murine liver. B) Antibody #2 exclusively detects the mitochondrial MIC26_22kDa, but no 55 kDa band. C) Antibody #1 does not recognize a mitochondrial MIC26_22kDa and 55 kDa band in plasma samples from two mice. D) Antibody #2, like antibody #1, does not recognize any protein in murine plasma samples. E) Antibody #1 detects the MIC26_22kDa as well as an unspecific 55 kDa band in C2C12 cell line along with different murine tissue lysates but not plasma. F) Antibody #2 detects MIC26_22kDa in mouse cell lines and tissues but not the 55 kDa band. C2C12-A represents C2C12 cells lysed with RIPA buffer and C2C12-B represents C2C12 cells which were mechanically lysed.

https://doi.org/10.1371/journal.pone.0286756.g004

Further, we compared the WBs of HEK293 and HeLa cell lysates with murine liver lysates using two other antibodies (#3 and #4) against MIC26 (S5A and S5B Fig). As shown before (Fig 1D and 1E), antibodies #3 and #4 recognised the human cell lysates (S5A and S5B Fig). However, antibody #3 did not recognise the murine MIC26_22kDa protein (S5A Fig) but recognised nonspecific proteins in murine liver lysates which includes a ≈ 25 kDa (Green asterisks) and a 55 kDa protein (S5A Fig). Since the antibody #3 was raised in mouse, the 25 and the 55 kDa proteins probably correspond to the IgG light and heavy chains which were only observed in murine plasma (S5C Fig), liver, muscle and white adipose tissue but not in C2C12 myoblast cell lysate (S5E Fig). We conclude that antibody #3 is not suitable for detecting MIC26 in murine samples. Next, in order to validate whether the 55 kDa protein is recognised in a nonspecific manner in murine samples, we used our custom-made antibody #4 which was designed to immunoreact only to human MIC26_22kDa and not murine MIC26_22kDa. The epitope used to elicit the anti-MIC26 antibody production in rabbit contained a total of 12 amino acids which perfectly match the human MIC26 but not mouse MIC26 as there was a four amino acid mis-alignment (S1C Fig). Accordingly, on one hand, in murine liver samples, we did not observe a mitochondrial MIC26_22kDa band while observing a presumed nonspecific band at various sizes including a 55 kDa band (S5B Fig). On the other hand, in human HEK293 and HeLa cells, antibody #4 showed multiple bands including the MIC26_22kDa and a protein at the 55 kDa region (S5B Fig). Therefore, antibody #4 specifically recognised human mitochondrial MIC26_22kDa while multiple other nonspecific bands were detected on the WBs in murine liver, HEK293 and HeLa cell lysates including a 55 kDa protein (S5B Fig). Clearly, this inconsistent detection of the 55 kDa protein upon using different antibodies in mammalian HEK293, HeLa cell and murine liver lysates strongly argues further for its nonspecific nature. Thus, while all the four antibodies were recognising the mitochondrial MIC26_22kDa version in human cell lysates, only two antibodies (#1 & #2) used in this study are suitable to recognise MIC26_22kDa mitochondrial version in murine tissues and cells.

Mass spectrometry data of 55 kDa protein does not reveal MIC26-specific peptides

The experiments performed in this study conclude that the presumed 55 kDa protein recognised using anti-MIC26 antibodies is nonspecific. Finally, we tested for the presence of MIC26-specific peptides using mass spectrometry in WT, MIC26 and MIC27 KO HEK293 cells. For this, we ran the corresponding cell lysates on an SDS gel after which the protein bands at and around the 55 kDa region (≈ 47 to 63 kDa proteins) were excised and the samples were given for mass spectrometry (MS) analysis. We could detect a total of 1273 proteins in the excised gel pieces in all the three samples. In total 278 mitochondrial proteins were detected including e.g. Prohibitin-2, NADH dehydrogenase flavoprotein 1, TIM50, glycerol kinase 2. However, as expected, we did not find MIC26-specific peptides in the MS analysis in any of the three samples analysed which included WT, MIC26 and MIC27 KOs. The list of detected peptides is presented in the S1 Data file. Hence, all the experiments presented in this study including the MS results conclusively show that the 55 kDa protein is of nonspecific nature with respect to MIC26.

The levels of the MIC26_22kDa mitochondrial protein are not altered in heart and livers of a db/db diabetic mouse model

MIC26 mRNA was previously found to be increased in hearts of people with diabetes mellitus [43]. MIC26 was also elevated in the transcriptome of hearts of dogs fed with high-fat diet [50]. Therefore, we examined whether MIC26 levels were changed in a murine model of obesity-related diabetes. For this, we used db/db.BKS mice which during the first three months develop massive obesity and uncontrolled diabetes due to a mutation in the leptin receptor [61–63]. The lean db/+ mice are normoglycemic and served as controls. We obtained heart and liver tissues from three male diabetic (db/db) and control (db/+) mice at the age of 12 weeks (hence diabetic) [64] and performed WBs using antibody #2 which specifically recognised only the MIC26_22kDa mitochondrial band in heart and liver of control and diabetic mice (Fig 5A and 5C). Quantification of the mitochondrial MIC26_22kDa protein level (Fig 5B and 5D) revealed no change in heart and liver tissues between control and diabetic mice. Additionally, we checked Mic26 and Mic27 transcript levels in liver of db/db.BKS diabetic and control mice and found no differences (Fig 5E). Thus, we conclude that in this standard mouse model for obesity and diabetes, we do not observe differences of MIC26_22kDa protein levels in the heart and liver tissues compared to control mice.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The diabetic db/db mouse model does not show different levels of the exclusively mitochondrially localized MIC26 and MIC27 proteins in heart and liver.

WB analysis of lysates from heart (A) and liver (C) tissue, along with associated quantification showing protein levels of MIC26 and MIC27 (B and D) from control db/+ and diabetic db/db.BKS mice, reveal no significant differences in amounts of MIC26 and MIC27. E) Mic26 and Mic27 transcript levels are not changed in liver of control db/+ and db/db.BKS diabetic mice.

https://doi.org/10.1371/journal.pone.0286756.g005