Human Bone Morphogenetic Protein‐3 Mature Ligand Is Larger Than Recorded Indicating Earlier Publications Are Flawed

ABSTRACT

Bone morphogenetic protein (BMP)‐3 (osteogenin) is an important regulatory cytokine during development and acts as a tumor suppressor in several cancers. BMPs are synthesized as proproteins that are cleaved by proprotein convertases to release biologically active ligands. The BMP‐3 mature ligand is recorded as 16 kDa consisting of carboxy‐terminal Q363‐R472 (110 amino acids). The BMP‐3 ligand sequence was determined based on the expectation that BMP ligands are similar in size and has not been experimentally verified. Here we show the BMP‐3 ligand is S290‐R472 (183 amino acids) consistent with the approximately 25 kDa BMP‐3 ligand secreted from human fibroblasts. This larger ligand size is consistent with proprotein convertase cleavage site prediction and supported by analyses using liquid chromatography tandem mass spectrometry. Data are available via ProteomeXchange with identifier PXD072827. These data indicate that the human BMP‐3 mature ligand is S290‐R472 and that earlier studies using recombinant human BMP‐3 ligand Q363‐R472 are flawed.

BMP‐3 is synthesized as a proprotein that is cleaved by furin/proprotein convertases (PC) to release the biologically active mature ligand. Although never experimentally verified, the BMP‐3 ligand is recorded as C‐terminal Q363‐R472 (110 amino acids) approximately 16 kDa. Experimental analysis using bioinformatics and LC–MS shows that the BMP‐3 ligand is S290‐R472 (183 amino acids) approximately 25 kDa. These data indicate that the BMP‐3 mature ligand has an additional 73 N‐terminal amino acids to that initially predicted and recorded for the BMP‐3 ligand.

1. Introduction

Bone morphogenetic protein (BMP)‐3 (osteogenin) is an important regulatory cytokine during development [1], osteogenesis [2], and disease [3, 4, 5]. BMP‐3 acts as a tumor suppressor in several cancers [6].

BMP‐3 is a member of the transforming growth factor (TGF)‐β superfamily of cytokines, which includes BMPs and activin. BMPs are synthesized as proproteins with an N‐terminal signal peptide directing secretion, a prodomain for proper folding, and a carboxy (C)‐terminal mature growth factor/ligand. BMP proproteins are cleaved by proprotein convertases (PCs), such as furin, to release ligands that form biologically active homo‐ or heterodimers (3–5).

BMP‐3 was first cloned in 1988 and cleavage was predicted to release a ligand of 12‐16 kDa (Q363‐R472, 110 amino acids [7]), similar in size to other BMPs isolated at that time [8]. This BMP‐3 ligand is recorded in Uniprot [7] and NCBI databases [9] and has been accepted in the literature as correct without experimental evidence.

We have recently reported that the endogenous human BMP‐3 ligand secreted by human fibroblasts was approximately 25 kDa [4], consistent with earlier reports of recombinant human BMP‐3 ligand between 25 and 30 kDa in size [2, 10, 11]. These studies suggest that the BMP‐3 ligand of 12–16 kDa is incorrect. Functional studies on BMP‐3 have frequently been contradictory [12, 13], possibly due to the use of a truncated recombinant BMP‐3 ligand. We investigated BMP‐3 ligand in human fibroblasts to determine the mature ligand sequence.

2. Materials and Methods

2.1. Cell Culture and Immunoprecipitation

Normal human BJ fibroblasts (ATCC CRL‐2522, Manassas, VA, USA) and fibroblast stable cell lines expressing recombinant BMP‐3 were cultured as described previously [4]. BMP‐3 was immunoprecipitated from concentrated cell supernatant [4] using V5‐Trap Magnetic Agarose (Proteintech Europe, Manchester, UK) following manufacturer's instructions; incubation with V5‐Trap Magnetic Agarose beads was overnight at 4°C. Western blotting was performed as described [4]. Antibodies: V5, IgG Dylight 680 and Dylight 800 (Thermo Fisher Scientific, Dublin, Ireland), BMP‐3 (Proteintech Europe, Manchester, UK).

2.2. Sample Preparation for Mass Spectrometry

BMP‐3 bands separated by PAGE were excised, destained, cut to approximately 3–5 mm cubes, dehydrated with acetonitrile, and reduced in 10 mM dithiothreitol; 50 mM ammonium bicarbonate for 45 min at 56°C followed by alkylation in 55 mM iodoacetamide; 50 mM ammonium bicarbonate for 30 min at room temperature in the dark, followed by dehydration. Gel pieces were digested in proteomics grade trypsin (20 ng/μL in 25 mM ammonium bicarbonate) overnight at 37°C, with gentle shaking. Tryptic peptides were extracted twice using 70% acetonitrile: 5% formic acid, dried under vacuum, resuspended in 0.5% trichloroacetic acid, and desalted on C18 tips (Ziptip. ZTC18S, Merck Life Science Limited, Wicklow, Ireland).

2.3. Mass Spectrometry

Proteomics samples were loaded onto EvoTips and run on a timsTOF Pro mass spectrometer (Bruker Daltonics, Bremen, Germany) coupled to the EvoSep One system (EvoSep BioSystems, Odense, Denmark). The peptides were separated on a reversed‐phase C18 Endurance column (15 cm × 150 μm ID, C18, 1.9 μm) using the preset 30 SPD method. Mobile phases were 0.1% (v/v) formic acid in water (phase A) and 0.1% (v/v) formic acid in acetonitrile (phase B). The peptides were separated by an increasing gradient of mobile phase B for 44 min using a flow rate of 0.5 μL/min.

The timsTOF Pro mass spectrometer was operated in positive ion polarity with TIMS (Trapped Ion Mobility Spectrometry) and PASEF (Parallel Accumulation Serial Fragmentation) modes enabled. The accumulation and ramp times for the TIMS were both set to 100 ms., with an ion mobility (1/k0) range from 0.6 to 1.6 Vs/cm. Spectra were recorded in the mass range from 100 to 1700 m/z. The precursor (MS) Intensity Threshold was set to 2500 and the precursor Target Intensity set to 20 000. Each PASEF cycle consisted of one MS ramp for precursor detection followed by 10 PASEF MS/MS ramps, with a total cycle time of 1.17 s.

A diaPASEF scheme consisting of 46 precursor isolation windows of 25 m/z width, covering a mass range of 300–1450 m/z and an ion mobility range of 0.6–1.6 Vs/cm, was created using the Bruker timsControl interface (2.0.53).

2.4. Data Analysis

Data acquired on the timsTOF using Data Dependent Acquisition (DDA) were searched against the Human Reference Proteome downloaded from Uniprot.org (reviewed entries, 29/07/2022) using MaxQuant (v 2.0.3.0, https://maxquant.org). The following parameters were used—trypsin was the digesting enzyme with up to two missed cleavages allowed for. Oxidation on methionine and acetylation on the N terminus were selected as variable modifications with a fixed modification of carbamidomethylation on Cysteine. Output from MaxQuant (msms.txt file) was used to build a spectral library in Skyline (Seattle Quant, MacCoss Lab, https://skyline.ms) to allow peptide spectra from the data acquired during Data Independent Acquisition (DIA) to be extracted and visually examined.

3. Results and Discussion

Prediction of furin and PC sites using ProP1.0 [14] indicated that human BMP‐3 has a predicted shared furin/PC site that contains the minimum consensus motif R‐X‐X‐R for PCs and the more rapidly cleaved furin consensus motif R‐X‐K/R‐R motif [15, 16], indicating that it is the most likely site for proteolytic processing (Figure 1A,B). Most BMP proproteins have a single PC cleavage site [16] suggesting that the second predicted PC site is unlikely to release the BMP‐3 mature ligand. This second PC site is predicted with lower probability, lacks the minimum consensus motif, and cleaves 12 amino acids from the amino (N)‐terminal of the secreted BMP‐3 proprotein (Figure 1A).

FIGURE 1.

(A) General PC cleavage site prediction for human BMP‐3 (P12645). ProP 1.0 score > 0.5 predicts propeptide cleavage site, the higher the score the more confident the prediction. (B) Furin specific cleavage site prediction for human BMP‐3. (C) Concentrated cell supernatants from human fibroblasts expressing recombinant BMP‐3 were immunoprecipitated using ChromoTek V5‐Trap and were analyzed by western blot with V‐5 tag antibody (red (700‐nm channel), V5 tag‐carboxy‐terminal rhBMP‐3), BMP‐3 antibody (green (800‐nm channel), both endogenous and rhBMP3), both antibodies (yellow (700 nm and 800‐nm channels), merged). Elution (E), M (molecular weight ladder). The increased size of recombinant protein is due to the presence of the V5 affinity tag (approx. 5 kDa). (D) BMP‐3 amino acid sequence showing peptide coverage following analysis by LC MS/MS. Blue peptides are present in BMP‐3 proprotein, red peptides are present in both BMP‐3 proprotein and ligand. Furin/PPC cleavage recognition site (IERRKKR↓ST) in bold. BMP‐3 ligand predicted by similarity is underlined and in italics. BMP‐3 signal peptide is underlined with a broken line.

Cleavage at the predicted shared furin/PC site releases a 183 amino acid BMP‐3 ligand, with a predicted molecular weight of 21 kDa. Conventional TGF‐β superfamily ligands are between 100 and 140 amino acids [17] indicating that BMP‐3 is unusually large, consistent with earlier reports that BMP3 is highly divergent from other BMPs [18]. The 21 kDa ligand is close in size to the endogenous BMP‐3 mature ligand expressed in human fibroblasts (approximately 25 kDa) [4]. Increased size is probably due to post‐translational modification such as glycosylation as previously reported [8].

Recombinant BMP‐3 proprotein and mature ligand expressed in human fibroblasts were immunoprecipitated using the C‐terminal V5 epitope tag (Figure 1C). Both recombinant (proprotein a) and endogenous BMP‐3 (proprotein b, only detected by BMP‐3 antibody) were observed to be immunoprecipitated, suggesting that these proteins complex following secretion from fibroblasts (Figure 1C).

BMP‐3 proprotein and ligand individually excised from PAGE‐SDS gels were analyzed using liquid chromatography tandem mass spectrometry (LC–MS/MS) and MaxQuant. 11 unique BMP‐3 peptides were identified, 8 of these peptides were identified both in BMP‐3 proprotein and mature ligand fractions (Figure 1D), the remaining three peptides were identified solely in the BMP‐3 proprotein (Figure 1D).

Inspection of the spectra in Skyline for the peptide K.DITQLLR.K [201, 207] showed clear consistent co‐eluting peaks for MS1 and MS2 for BMP‐3 proprotein and the expected ratio of fragments that should be seen for the peptide at MS1 and MS2 levels (Figure 2A). This peptide was not present in the BMP‐3 ligand (Figure 2A). In contrast, the peptide R.STGVLLPLQNNELPGAEYQYK.K [290, 310] immediately following the predicted furin/PC cleavage site at amino acid 289 was identified in both BMP‐3 proprotein and ligand (Figure 2B). These data indicate that BMP‐3 proprotein is cleaved between K. DITQLLR.K [201, 207] and STGVLLPLQNNELPGAEYQYK.K [290, 310] consistent with furin/PC cleavage at position 289, IERRKKR↓ST (Figure 1A,B).

FIGURE 2.

Skyline chromatogram views (i) and (ii) for precursor MS1 and fragment MS2 ions are shown for the peptide indicated. Skyline peak area view (iii) displays the relative abundance of the indicated peptide across duplicate runs for the proprotein and ligand in bar graph form. In each Skyline view the upper panel represents the MS1 ions while the lower panel represents the fragment MS2 ions. (A) Peptide K.DITQLLR.K [201, 207] for BMP‐3 proprotein and ligand. (B) Peptide R.STGVLLPLQNNELPGAEYQYK.K [290310] for BMP‐3 proprotein and ligand. (C) Peptide K.KDEVWEER.K [311, 318] and peptide K.TLQAQAPEK.S [323331] for BMP‐3 ligand.

An additional three peptides, K.KDEVWEER.K [311, 318], K.TLQAQAPEK.S [323, 331] (Figure  3C) and K.SQTLQFDEQTLK.K [346, 357] (Figure 1D) were identified that are located between the furin cleavage site and the start of the Q363‐R472 BMP‐3 ligand (Figure 2C). The Skyline idotp values were 0.96 or greater, indicating high confidence identification. These LC–MS/MS analyses clearly demonstrate that the human BMP‐3 ligand contains amino acids S290 to K357 in addition to amino acids Q363‐R472. These data support the prediction that S290‐R472 is the human BMP‐3 ligand.

Our studies indicate that the recorded BMP‐3 ligand Q363‐R472 is in fact a C‐terminal fragment of the human BMP‐3 ligand. The loss of greater than 70 amino acids from the BMP‐3 ligand is likely to alter biological signaling and activity. Investigations using this BMP‐3 fragment may have led to errors in the functional characterization of human BMP‐3, as structural differences between BMPs regulate specificity of ligand‐receptor binding [19]. Further studies are needed on the structure, receptor affinity, specificity, and activity of the 183 amino acid BMP‐3 mature ligand.

Author Contributions

David E. Phelan: investigation, methodology, visualization, validation, formal analysis, writing – reviewing and editing. Caitriona Scaife: investigation, methodology, visualization, validation, formal analysis, writing – reviewing and editing. Emma R. Dorris: conceptualization, writing – reviewing and editing. John Russell: conceptualization, funding acquisition, writing – reviewing and editing. Madeline Murphy: conceptualization, funding acquisition, supervision, writing – original draft preparation.

Funding

This work was supported by the National Children's Research Centre (NCRC), C/19/4.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Immunoprecipitation of recombinant BMP‐3 secreted by human fibroblasts. (A) Concentrated cell supernatants (secreted) from BMP‐3‐BJ fibroblasts were immunoprecipitated (IP) using ChromoTek V5‐Trap. Excised regions for analyses by mass spectrometry in red boxes (i) Proprotein and (ii) mature growth factor/ligand. Elution (E), M (molecular weight ladder). Protein was eluted in loading buffer (187.5 mM Tris–HCl, pH 7.0, 6% SDS (w/v), 30% glycerol, 0.03% bromophenol blue (w/v), 125 mM DTT) at RT for 10 min (E1) and in fresh loading buffer at 95°C for 10mins (E2). (B) Immunoprecipitated proteins in (A) were analyzed by western blot with V‐5 tag antibody (red (700‐nm channel), V5 tag‐carboxy‐terminal rhBMP‐3), BMP‐3 antibody (green (800‐nm channel), both endogenous and rhBMP3), both antibodies (yellow (700 nm and 800‐nm channels), merged).

Figure S2: Uncropped image of western blot used in Figure 1C and Figure S1.

Data Availability Statement

The mass spectrometry proteomics data that support the findings of this study have been deposited to the ProteomeXchange Consortium via the PRoteomics IDEntifications Database (PRIDE) [20] partner repository with the dataset identifier PXD072827. All other data that support the findings of this study are available in the Materials and Methods, Results, and Supporting Information S1 of this article.