Ligand binding and partial characterization of sex-specifically expressed lipocalins of submandibular and lacrimal glands of Syrian hamsters

Abstract

Lipocalins are abundantly expressed secretory proteins that perform diverse roles such as ligand-transport, immunomodulation, cell-signaling, chemical communication etc. In Syrian hamsters (Mesocricetus auratus), male-specific submandibular gland proteins (MSP) and female-specific lacrimal gland proteins (FLP) are sex-specifically secreted in saliva and tears respectively. MSP and FLP are lipocalins having 85% protein sequence identity between themselves and they have 58–61% identity with odorant-binding lipocalins (OBP) of rat and mouse. We purified natural MSP and FLP from hamster tissues and recombinant MSP and FLP after cloning and overexpression in E. coli. We found that MSP and FLP have very similar far ultraviolet-circular dichroic (UV-CD) spectra, typical of lipocalins. Natural MSP was found to be expressed as glycosylated and non-glycosylated forms and had no phosphorylation while FLP had no glycosylation or phosphorylation. In vitro ligand binding studies showed that MSP and FLP bind with high affinity to 2-isobutyl-3-methoxypyrazine (IBMP; a potent bell-pepper odorant) and also other test odorants. Therefore, we conclude that the sex-specific MSP and FLP lipocalins are odorant-binding proteins and they may have a role in odor-mediated communication in male and female hamsters.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-12831-9.

Introduction

We had identified in Syrian hamsters (Mesocricetus auratus) two sex-specifically expressed and sex-hormonally regulated secretory proteins, MSP (male-specific submandibular gland protein) and FLP (female-specific lacrimal gland protein)^1–5. MSP and FLP are secreted in saliva of males and in tears of females, respectively^1–5. MSP is abundantly expressed in submandibular salivary gland (SMG) of male hamsters as major 24 and 20.5 kDa protein species and a minor 30 kDa species (together comprising ~ 40% of total soluble proteins of SMG)^1,3,5. MSP proteins are also detectable at very low levels in urine^3,6 and in extracts of body-fur⁷ of adult male hamsters. The three MSP species are immunologically similar and are products of a single MSP gene^2,3,5. On the other hand, FLP is expressed in lacrimal gland (LG) of female hamsters as a single protein species of molecular mass 20 kDa (comprising ~ 20% of soluble proteins of LG)^2,4,5. FLP was found to have immunological similarity with MSP^2,3,5.

The lipocalin protein family consists of abundant, secretory, carrier proteins of molecular mass ~ 20 kDa, which usually have an acidic isoelectric point. Lipocalins usually share low sequence similarity, but whenever investigated, they were found to have a conserved 3D structure with a binding pocket for small hydrophobic molecules such as lipids, retinol, steroids, odorants, siderophores etc^8,9. cDNA cloning and homology search of deduced protein sequence of MSP and FLP (Fig. 1) showed that they belong to the lipocalin family having best match with mammalian OBPs². MSP/FLP share maximum sequence identity (85%) between themselves while both rat OBP and mouse OBP, which are lipocalins expressed and secreted by nasal glands in both sexes, have 61 − 58% identity with MSP/FLP (Fig. 1).

Fig. 1.

Alignment of sequences of mature forms of FLP, MSP and closely related lipocalins, rat OBP1f, mouse OBP1, Syrian hamster aphrodisin and a predicted OBP-like lipocalin of Chinese hamster. Amino acids identical in all sequences are in red. Cysteines are indicated by dots. Arrowheads indicate the 15 amino acids residues known to line the ligand-binding pocket of rat OBP1f¹⁰. Lipocalin signature motif GXW and the CXXXC motif, conserved in all sequences are boxed. Putative N-glycosylation site in MSP is underlined. Numbers of amino acids in each mature protein and percentage identity with FLP are indicated. # The Chinese hamster OBP-like lipocalin sequence included in the alignment is the mature form of a predicted full-length sequence of 173 aa named ‘female-specific lacrimal gland protein [Cricetulus griseus]’ (Uniprot code: A0A8C2LDC7) whose predicted partial mRNA (complete CDS) of 522 bp was annotated from the Chinese hamster genome sequence. Uniprot codes of other protein sequences used in the alignment and number of amino acids in their full-length sequence (including the signal peptide) are as follows: FLP/SyrhamFLP: Q99MG7 (172 aa); MSP/SyrhamMSP: Q9QXU1 (172 aa); aphrodisin/SyrhamAPHR: Q9Z1I7 (167 aa); RatOBP1f: Q9QYU9 (173 aa); MusOBP1: B1AVU4 (173 aa). This figure is an adapted and modified version of an alignment performed using MULTALIN tool, which was published by us earlier².

Notably, the rat and mouse nasal OBP lipocalins have been shown to bind odorants in vitro^11–19. Moreover, aphrodisin, a lipocalin, which is female-specifically expressed in genital tract tissues of Syrian hamsters²⁰ and abundantly secreted in female hamster’s vaginal discharge^21–23 has 39% identity with MSP/FLP (Fig. 1) and it also binds odorants in vitro^24,25. Interestingly, the odoriferous, pungent smelling female hamster vaginal discharge and even aphrodisin protein purified from it were shown to elicit an intense copulatory response (aphrodisiac effect) on male hamsters only after these are contacted orally (i.e., licked) by the males^{21–23,26,27}. As shown in Fig. 1, both FLP and MSP have maximum identities (68%) with a ‘predicted’ OBP-like lipocalin named as ‘female-specific lacrimal gland protein’ of Chinese hamster (UniProt: A0A8C2LDC7), which was annotated from CHO-K1 cell line genome sequence. However, this ‘predicted’ odorant-binding protein has never been isolated and there is no information on whether and where this protein and/or its predicted transcripts are expressed in Chinese hamsters and whether the protein can also bind odorants or other hydrophobic ligands. Both FLP and MSP sequences have the characteristic lipocalin signature motif GXW^8,9 near their N-terminal end and a pair of cysteines corresponding to positions 63 and 154 of FLP (Fig. 1) which are conserved in almost all lipocalins. The two other cysteines (residues 44 and 48 of FLP), which constitute a CXXXC motif in FLP/MSP, are conserved only in a small subset of odorant/pheromone-binding lipocalins², which includes rat/mouse OBPs, aphrodisin and Bos d2 (an allergenic lipocalin found on bovine skin whose odorant/pheromone binding is not yet demonstrated²⁸; not shown in Fig. 1). The CXXXC motif is also conserved in the ‘predicted’ OBP-like lipocalin of Chinese hamster (Fig. 1).

Mammalian OBP lipocalin, first isolated from cow olfactory mucosa, was found to display strong binding affinity for 2-isobutyl-3-methoxypyrazine (IBMP; a potent bell pepper odorant) as well as other odorants²⁹. Subsequently, odorant-binding lipocalins have been identified in the nasal mucosa of several mammals such as rat, mouse, rabbit, porcupine, pig and also human^{11–19,29–33}. It has been shown that OBPs reversibly bind different classes of odorants and they are believed to bind airborne odorants (which are hydrophobic in nature) and carry them across the aqueous nasal mucosa for presentation to the olfactory receptors on sensory cells of the olfactory epithelium^{11–19,29–33}. Unlike many other mammals, no endogenous OBPs were found to be expressed in nasal mucosa of Syrian hamster². Nevertheless, MSP/FLP secreted into saliva and tears of Syrian hamster (from submandibular and lacrimal glands) can access the nasal cavity wherein they can function as OBPs to help in odor perception. However, whether MSP and FLP can indeed bind odorants and other hydrophobic ligands has not been investigated.

Here we report the purification of natural MSP and FLP lipocalins, overexpression and purification of recombinant MSP and FLP, some partial characterization studies and investigations on ligand binding properties of these lipocalins. The characterization studies include far UV-CD spectra of recombinant MSP and FLP, investigation of presence of phosphorylation and the nature of glycosylation in natural MSP and FLP lipocalins. Ligand-binding property of recombinant and natural forms of MSP and FLP lipocalins was investigated using different techniques. Possible functions of MSP and FLP lipocalins are also discussed.

Materials and methods

Animal maintenance

Adult Syrian (golden) hamsters (Mesocricetus auratus) were bred and maintained in 14 : 10 :: light : dark cycle (lights off at 6 pm) in the animal house facility of our Institute (CSIR-CCMB, Hyderabad). Our Institutional Animal Experimentation Ethics Committee had approved all animal experiments, surgical procedure (bilateral ovariectomy) and the sacrifice of Syrian hamsters to collect SMG and LG tissues for purification of specific proteins from them. Bilateral ovariectomy on adult female hamsters was performed under anesthesia (intraperitoneal injections of thiopentone sodium at 5 mg/kg body weight) and hamsters were sacrificed 20 days after surgery. To sacrifice Syrian hamsters, we performed euthanasia by carbon dioxide inhalation until death. The required tissues (SMG or LG) were excised out immediately after sacrifice, wet weight taken and frozen in -70 °C freezer for later use. This study is reported in accordance with ARRIVE guidelines (https://arriveguidelines.org). We confirm that all methods were performed in accordance with the relevant guidelines and regulations of our Institutional Animal Experimentation Ethics Committee.

Purification of natural MSP and FLP proteins

Natural MSP was purified from adult male hamster SMG (Fig. 2A). Pooled SMG (8 g) was minced with a pair of scissors in chilled condition on ice and then homogenized in 25 ml of ice-cold 20 mM Tris-HCl buffer pH 7.4 using a homogenizer (Polytron; Kinematica GmbH). The homogenate was centrifuged at 30,000 g for 30 min at 4 °C to obtain a clear supernatant. The soluble supernatant was fractionated stepwise at 40, 60, 85, and 100% (NH₄)₂SO₄ saturations, and proteins precipitated at each step were dissolved in Milli-Q water and analyzed by SDS-PAGE (sodium dodecyl sulphate-polyacrylamide gel electrophoresis). The precipitate obtained at 85–100% (NH₄)₂SO₄ saturation contained considerable amounts of MSP proteins, comprising of major 24 and 20.5 kDa forms and the minor 30 kDa form, along with trace levels of high Mr contaminating proteins. The dissolved protein precipitate from the 85–100% (NH₄)₂SO₄ fraction was loaded onto a calibrated gel-filtration column (1.6 × 95 cm; Sephadex-G75 superfine; GE) and eluted with 20 mM Tris-HCl pH 7.4 containing 150 mM NaCl. The eluted fractions were monitored by measuring OD at 280 nm and assessed for purity in Coomassie-stained SDS-PAGE. Fractions containing a co-purified mix of all three forms of natural MSP (free from other proteins) were pooled and used for further characterization and ligand-binding studies (Fig. 2A).

Fig. 2.

Purification of natural and recombinant forms of MSP and FLP proteins. Flow charts show, steps involved in purification of natural MSP from male submandibular gland (A) natural FLP from ovariectomized female lacrimal gland (B) and recombinant MSP/FLP after overexpression in E. coli (C); In (D) SDS-PAGE profile of purified natural and recombinant MSP and FLP (top panel) and their Western blot (bottom panel) are shown. In (D) rFLP (lane 2) and rMSP (lane 4) migrate at ~ 22 kDa (rFLP runs slightly faster than rMSP); purified natural FLP (lane 3) migrates at ~ 20 kDa; purified mixture of natural MSP (lane 5) contains major ~ 24 and 20.5 kDa species and a relatively minor ~ 30 kDa species, which were co-purified. Molecular weight marker proteins were run in lane 1. In bottom panel, the blot was probed with antisera raised against purified FLP, which also crossreacts with all forms of MSP. In (B) female symbol with a slash indicates ovariectomized female.

For natural FLP purification, 1 g of pooled LG tissue collected from adult female hamsters (20 days after ovariectomy), was minced with a pair of scissors and homogenized in 25 ml of ice-cold 20 mM Tris-HCl buffer, pH 8.2 (Fig. 2B). The homogenate was centrifuged at 30,000 g for 30 min at 4 °C to obtain a clear supernatant. The supernatant was then loaded on a DE-52 anion exchange column and bound fractions were eluted using a NaCl gradient (0 to 300 mM) in the homogenization buffer. The eluted fractions were monitored for protein content by measuring OD at 280 nm and analyzing individual fractions in Coomassie-stained SDS-PAGE. Fractions containing only the 20 kDa FLP proteins were pooled, concentrated and further purified using a Sephadex G-75 gel-filtration column, as described above for natural MSP purification (Fig. 2B).

Enzymatic deglycosylation of purified natural MSP and FLP

Purified natural MSP (mixture of major 24 and 20.5 kDa and minor 30 kDa forms) and 20 kDa FLP proteins were separately incubated at 37 °C with N-glycosidase F or O-glycosidase (Roche) in 20 mM sodium phosphate buffer pH 7.0. Purified natural MSP was also incubated at 37 °C with sialidase (from Vibrio cholerae, Roche) in 100 mM sodium acetate buffer pH 5.5 containing 4 mM CaCl₂ and 150 mM NaCl and Endoglycosidase H (Endo H; Roche) in 20 mM sodium acetate buffer pH 5.5. At the end of 24 h incubations, the samples were analyzed in Coomassie-stained SDS-PAGE.

Glycan differentiation analysis

Analysis of glycans in natural MSP protein was performed using DIG Glycan Differentiation Kit (Roche) following manufacturer’s recommendations and also by lectin affinity chromatography. The affinity of natural MSP to Sepharose-conjugated lectins Con A and WGA and Agarose conjugated lectin, jacalin was determined by affinity chromatography. The natural MSP (mixture of three forms) was passed through these lectin columns and unbound fractions and fractions obtained after elution with solutions of different sugars (specific for each lectin) were analyzed for presence of MSP forms in SDS-PAGE.

Preparation of overexpression constructs for recombinant MSP (rMSP) and recombinant FLP (rFLP)

Using the full-length cDNA of MSP (GenBank: AF183407.2² as template, forward primer (5’-ATCAATCATATGGCACAGCATCAGAATCTTGAA-3’) and reverse primer (5’-ACTCTCGAGTTGATTACAAGTGTCTGTGGT-3’) having anchored NdeI and XhoI restriction sites respectively (indicated in bold within the primer sequence), the MSP cDNA encoding the complete mature MSP polypeptide (157 residues) was PCR amplified. This PCR product was cloned within the NdeI and XhoI site of pET21a (+) vector plasmid and transformed into ultracompetent DH5α Escherichia coli (E. coli) cells. Similarly, using FLP cDNA (GenBank: AF345648.1² as template, forward primer (5’-ATCAATCATATGCATTATCAGAATCTTGAAGTC-3’) and reverse primer (5’-(ACTCTCGAGTTGATCACAATTGTCTGTGGG-3’) with anchored NdeI and XhoI restriction sites respectively, the FLP cDNA encoding the complete mature polypeptide (156 residues) was PCR amplified and cloned within the NdeI and XhoI sites of pET21a (+) vector plasmid and transformed into ultracompetent DH5α E. coli cells³⁴. The inserts within the two types of purified plasmids were verified by sequencing. DNA sequencing was done using Big Dye Terminator ready reaction kit (Perkin Elmer) and ABI Prism 3700 DNA analyzer. Both strands were sequenced for all clones. The rMSP and rFLP overexpression constructs were used for bacterial overexpression of rMSP and rFLP proteins respectively.

Overexpression and purification of rMSP and rFLP

The overexpression constructs in pET21a (+) were prepared such that only a methionine residue is added at the N-terminal end of the mature polypeptides of MSP and FLP and 8 extraneous amino acids (leucine and glutamic acid encoded by the Xho1 site followed by a 6X histidine-tag) are added at the C-terminal end. The above plasmid constructs for bacterial expression were transformed separately into competent E. coli BL21 (DE3) cells. Transformants from a single colony were inoculated into 10 ml Luria Bertani (LB) medium containing ampicillin (100 µg/ml) and incubated with shaking at 37 °C overnight. This pre-culture was then used to inoculate 1 L of LB medium containing ampicillin (100 µg/ml) and incubated with shaking at 37 °C. After an OD of 0.6 at 600 nm was reached, protein expression was induced by adding isopropyl-β-D-1-thiogalactopyranoside (IPTG) to a final concentration of 1 mM and incubation continued for an additional 4–5 h. The bacterial cells were harvested by centrifugation at 6,000 g for 30 min at 4 °C and the cell pellet was resuspended in ice-cold lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl, pH 8.0) containing 5 mM imidazole, 1 mM PMSF, and protease inhibitor cocktail (Roche). Cells were lysed by sonication on ice and lysate was centrifuged at 30,000 g at 4 °C for 30 min to remove cell debris. The supernatant passed through Ni-NTA column (Qiagen) at 4 °C, which was pre-equilibrated with lysis buffer containing 10 mM imidazole. The column was then washed with lysis buffer containing 20 mM imidazole. Elution was done with 250 mM imidazole in lysis buffer. The eluted fractions containing rFLP or rMSP proteins were pooled, concentrated using centrifugal filter (ultracel-5k; Millipore) and loaded on a calibrated gel-filtration column (1.6 × 95 cm; Sephadex-G75 superfine; GE) and eluted with 20 mM Tris-HCl pH 7.4 containing 150 mM NaCl. Eluted fractions were monitored by measurement of OD 280 nm and checked for purity by SDS-PAGE. Pure rMSP and rFLP eluted in symmetrical peaks at an estimated molecular mass of ~ 20 kDa. Fractions of pure rMSP or rFLP were pooled, desalted and concentrated using centrifugal filter (ultracel-5k; Millipore).

Protein quantification

The protein concentration of purified recombinant MSP and FLP was determined by UV absorbance at 280 nm, using molar extinction coefficients of 14,690 M^− 1 cm^− 1 for rMSP and 17,670 M^− 1 cm^− 1 for rFLP. These coefficients were calculated using the ProtParam tool at www.expasy.org for rMSP (molecular weight: 19,011 Da) and rFLP (molecular weight: 18,933 Da) at the maximum absorption of 280 nm. The nine extraneous amino acids (N-terminal methionine, C-terminal leucine, glutamic acid and 6-histidine tag) were also considered while calculating the molar extinction coefficients and molecular weights of rMSP and rFLP proteins.

SDS-PAGE and Western blot analysis

SDS-PAGE was done essentially as described previously³⁴. Unless otherwise mentioned, gels were of 11% acrylamide (acrylamide : bisacrylamide :: 29 : 1). Gels were either stained with Coomassie R-250 or used for Western blotting³⁵. For Western blots, gels were blotted (0.8 mAcm^− 2 for 1 h) onto Hybond-C nitrocellulose membranes (Amersham) using semi-dry transfer apparatus (Amersham). Membranes were blocked with 2% BSA (bovine serum albumin) in TBST (20 mM Tris-HCl, pH 7.5, 0.9% NaCl, 0.1% Tween-20) followed by incubation for 1 h with FLP antisera (which also crossreacts readily with all forms of MSP^2,3,5 diluted (1:10,000) in TBST containing 2% BSA. After extensive washing with TBST, blots were incubated with horse-radish peroxidase (HRP) conjugated anti-rabbit IgG goat antibodies (Boehringer) (diluted 1:20,000) in TBST containing 2% BSA. Blots were washed with TBST and crossreactions were detected by chemiluminescence reaction of HRP followed by autoradiography. All tissue extracts were freshly prepared for checking their profiles in protein-stained gels and for Western blotting and representative results are shown. For the detection of phosphorylation in natural FLP and MSP, mouse monoclonal antibodies against phosphotyrosine (p-Tyr-20 at 1:2000 dilution; Santa Cruz Biotechnology) and phosphoserine/threonine/tyrosine (1:2000 dilution; Abcam) were used and crossreactions were detected using horseradish peroxidase (HRP) conjugated anti-mouse IgG goat antibodies (diluted 1:20,000; Santa Cruz Biotechnology) and chemiluminescence method.

Circular dichroism (CD) spectroscopy

Far UV-CD spectra of recombinant MSP and FLP proteins (0.1 mg/ml in 20 mM potassium phosphate buffer, pH 7.4) were recorded at room temperature on a Jasco J-715 spectropolarimeter between 195 and 250 nm using 0.1 cm path length quartz cuvette. Four accumulations were recorded per spectra at a response time of 1 s with a bandwidth of 2 nm. Spectra were scanned at a speed of 100 nm/min. The raw CD data were converted to mean residue ellipticity with units (degree.cm².dmol^-1). Buffer subtracted CD-spectra of rFLP and rMSP were analyzed for secondary structure content using web server CAPITO (CD Analysis and Plotting Tool)³⁶.

Fluorescence ligand binding assay

Fluorescence spectra were recorded using a Hitachi-4500 fluorescence spectrophotometer at 25 °C, with a 1 cm path length quartz cuvette and 5 nm slits for both excitation and emission. To detect the binding of the fluorescent probe 1-aminoanthracene (1-AMA) to rMSP, fluorescence was excited at 295 nm, and emission spectra were recorded between 300 and 550 nm. The formation of the rMSP-AMA complex was detected by an increase in fluorescence intensity at 485 nm. For the binding of the fluorescent probe N-phenyl-1-naphthylamine (1-NPN) to rFLP, fluorescence was also excited at 295 nm, and emission spectra were recorded between 300 and 550 nm or 350–550 nm. The formation of the rFLP-NPN complex was detected by an increase in fluorescence intensity at 425 nm. Fluorescent probes 1-AMA and 1-NPN, along with other ligands, were dissolved in 100% methanol as 1 mM solutions. To measure the affinity of the fluorescent ligand 1-AMA to rMSP or 1-NPN to rFLP, a 2 µM protein solution in 20 mM Tris-HCl buffer (pH 7.4) was titrated with aliquots of the 1 mM ligand dissolved in 100% methanol.

To determine the binding constants of the fluorescent probe, the intensity values corresponding to the maximum fluorescence emission (485 nm for the rMSP-AMA complex and 425 nm for the rFLP-NPN complex) were plotted against the concentrations of the free probe. The bound ligand concentration was assessed from the fluorescence intensity values, assuming the protein was 100% active and that the stoichiometry of the protein-probe complex was 1:1 at saturation. The dissociation constant (K_d) of the protein–probe complex was calculated by fitting the binding curve to the experimental data using the computer program Origin 7.0.

The affinities of other ligands were measured using competitive binding assays employing 1-AMA for rMSP and 1-NPN for rFLP, as fluorescent reporter probe. These assays aimed to displace the bound probes with the ligands. The assays were conducted with a 2 µM protein solution and a 2 µM probe concentration. Aliquots of 1 mM ligand solution (competitor) were added to the protein-probe complex at increasing concentrations (0.25–40 µM). The competitor concentrations that caused fluorescence decay to half-maximal intensity (at 485 nm for the rMSP-AMA complex and 425 nm for the rFLP-NPN complex) were recorded as half maximal inhibitory concentration (IC₅₀) values. The apparent K_i values (the concentration of the free inhibitor at 50% inhibition) were calculated using the formula K_i = [IC₅₀] / (1 + [L] / K_d), where [L] is the free fluorophore concentration and K_d is the protein-probe dissociation constant³⁷.

Volatile ligand extraction

To determine whether the purified natural MSP and natural FLP proteins contain any bound volatile ligands, 0.5 ml of each protein solution (1 mg/ml), containing either the mixture of three forms of MSP (24, 20.5 and 30 kDa) or purified 20 kDa nFLP was mixed separately with 0.5 ml dichloromethane (DCM) or 0.5 ml hexane in a glass vial. The mixtures were vigorously vortexed and stored overnight at -20 °C. After thawing, the lower organic phase (in the case of the DCM-protein mix) or the upper organic phase (in the case of the hexane-protein mix) was carefully removed and analyzed for the presence of any volatile ligands by GC-MS (see below).

In experiments directly investigating the binding of the odorant (IBMP) with proteins, elution fractions obtained from gel-filtration of the protein-IBMP mixture were mixed separately with DCM in a glass vial. These mixtures were vortexed and stored overnight at -20 °C. After thawing, the lower organic phase was carefully removed and analyzed by GC-MS for the presence of IBMP.

Gas chromatography and mass spectrometry (GC-MS)

GC–MS analysis was performed using an Agilent 6890N gas chromatograph (Agilent Technologies, USA) coupled with a model 5973i mass selective detector. The system was equipped with an HP-5MS capillary column (30 m length, 0.25 mm inner diameter, 0.25 μm film thickness). The inlet temperature was maintained at 250 °C and helium was used as the carrier gas at a flow rate of 1 ml/min in constant flow mode. A 2 µl volume of organic extracts was injected in splitless mode. The oven temperature program began at 50 °C and was held for 5 min. It was then increased to 100 °C at a rate of 5 °C/min, followed by a ramp to 280 °C at a rate of 10 °C/min, where it was held for 5 min³⁸. The temperature was then further increased to 300 °C at a rate of 5 °C/min and held for 1 min. The ion source and quadrupole temperatures were set to 230 °C and 150 °C respectively, with ionization achieved using a 70 eV electron beam. Masses were acquired in total ion current (TIC) mode with an acceleration voltage turned on after a solvent delay of 240 s. Mass spectra were scanned from 15 to 350 Da. Compound identification was performed by comparing the MS spectra with those in the Wiley 7 N Edition (Agilent Part No. G1035B) library.

Results

Purification of natural MSP and natural FLP from SMG and LG tissues of Syrian hamsters

Natural MSP was purified from SMG of adult male hamsters wherein it comprises approximately 40% of the total soluble proteins of SMG^1,3,5,6. Since ovariectomy results in a marked increase of FLP expression in LG of female hamsters^4,5, therefore natural FLP was purified from LG of ovariectomized female hamsters. The rapid purification procedure for obtaining a co-purified mixture of all three molecular weight forms of natural MSP is summarized in Fig. 2A and the purification process for the 20 kDa natural FLP is summarized in Fig. 2B.

Natural MSP was co-purified as a mixture of the 24 kDa and 20.5 kDa major forms, along with a minor 30 kDa form present in roughly the same relative proportions as they are present in crude soluble supernatants of male SMG (Fig. 2A). The purity of the natural FLP and MSP proteins was confirmed by SDS-PAGE (Fig. 2D).

Bacterial over-expression and purification of recombinant MSP and FLP

Over-expression construct in pET21a (+) vector was prepared in such a way that only a methionine residue was added at the N-terminal end of mature polypeptide of MSP (and also FLP) and eight extraneous amino acids (leucine and glutamic acid encoded by XhoI site followed by a six histidine-tag) was added at the C-terminal ends (see also Methods section). Recombinant MSP and FLP were overexpressed as soluble His-tagged proteins and purified to homogeneity from bacterial cell lysate using Ni-NTA affinity followed by size-exclusion chromatography (Fig. 2C). The purification process yielded approximately 30 mg of recombinant protein per liter of culture. In SDS-PAGE (Fig. 2D), the purified recombinant MSP and FLP proteins exhibited mobility corresponding to an estimated molecular mass of ~ 22 kDa. This increased mass is attributed to the extraneous amino acid residues (His-tag, etc.). Western blotting using antisera raised against purified FLP, confirmed immunological similarity between purified natural FLP (nFLP), all 3 forms of nMSP (present as a mixture in the purified natural MSP) as well as rFLP and rMSP (Fig. 2D).

Investigations on the presence of phosphorylation in natural MSP and FLP proteins

Phosphorylation was earlier detected in several lipocalins including human lacrimal gland/tear lipocalin, human von Ebner’s salivary gland lipocalin, pig OBP of nasal mucosa, pig von Ebner’s gland salivary lipocalin, mouse 24p3 and in odorant/pheromone-binding lipocalins of male mouse urine (MUPs)^39–45. It was proposed that phosphorylation might modulate the binding of ligands by lipocalins or even the binding of the lipocalins to a putative cellular receptor for the lipocalin⁴¹. The sequence analysis of mature MSP and FLP proteins using the NetPhos 3.1 server predicted 14 potential phosphorylation sites in the MSP sequence and 18 in the FLP sequence and similar analysis of other 4 rodent lipocalins aligned in Fig. 1 predicted many (between 11 and 17) potential phosphorylation sites in each. These phosphorylation sites were distributed among serine, threonine, and tyrosine residues (results not shown).

Figure 3 shows the result of investigations on phosphorylation in MSP/FLP lipocalins by Western blots using monoclonal antibodies (anti-phosphoserine/threonine/tyrosine and anti-phosphotyrosine). Figure 3A shows Coomasssie-stained SDS-PAGE of purified nMSP (a mixture of 20.5, 24 and 30 kDa forms) (lane 1), purified nFLP of molecular weight 20 kDa (lane 2), crude extract of male SMG (lane 3) and crude extract of ovariectomized female LG (lane 4). In results of Western blots shown in Fig. 3B and C, no cross-reactions with anti-phosphotyrosine, as well as anti-phosphoserine/threonine/tyrosine antibodies, were detected in all three forms of purified MSP (lane 1) or in crude male SMG extract (lane 3). Figure 3B and C also show that no crossreactions were detected for purified 20 kDa nFLP (lane 2) or 20 kDa nFLP in crude LG extract of ovariectomized female (lane 4). However, crossreactions of the antibodies with some unidentified proteins are clearly detectable in crude extracts of male SMG (lane 3) and in ovariectomized female LG (lane 4), which very likely contain phosphorylated serine, threonine or tyrosine residues. Thus, the Western blot results indicate the absence of phosphorylated serine, threonine or tyrosine residues in all 3 forms of natural MSP as well as in natural FLP.

Fig. 3.

No cross-reactions with natural MSP and FLP proteins were detected in Western blots using anti-phosphoserine/threonine/tyrosine antibodies. Purified natural MSP (lane 1), purified natural FLP (lane 2), SMG extract of male hamster (lane 3) and LG extract of ovariectomized female hamster (lane 4) were run in 12% SDS-PAGE in three similar sets with each containing the above mentioned 4 samples (total 12 lanes). A cut segment of the gel containing one set (with 4 lanes) was Coomassie stained (Panel A) while the remaining gel was blotted. Transfer of proteins after blotting was confirmed by visualizing after Ponceau staining (images of Ponceau-stained blots were not recorded). The blot was then appropriately cut into 2 similar sets containing 4 lanes each, which were kept for blocking and one set was probed with antibody against phosphoserine/threonine/tyrosine (Panel B) while the other set was probed with a separate antibody against phosphotyrosine (Panel C). See “Methods” section for details.

Investigations on glycosylation in MSP and FLP proteins

Analysis of mature MSP and FLP protein sequences using NetNGlyc 1.0 predicted the presence of single potential N-glycosylation site at Asn26 in mature MSP but none in FLP. Similar analysis of mature sequences of rat OBP1f, Mouse OBP1 and the predicted Chinese hamster OBP-like lipocalin showed that none of these had any potential N-glycosylation site. Interestingly, mature sequence of Syrian hamster aphrodisin had 2 potential N-glycosylation sites predicted at Asn41 and Asn69, which were confirmed earlier^23,24,46. NetOGlyc 4.0 predicted no potential O-glycosylation site in both FLP and MSP or in the other rodent lipocalins aligned in Fig. 1.

Figure 4 shows that incubation with O-glycosidase enzyme did not affect the mobility (or levels) of the three forms of natural MSP (lane 2) and natural 20 kDa FLP (lane 8) indicating the absence of O-glycosylation in MSP and FLP proteins. Treatment of the mixture of three nMSP forms with N-glycosidase F enzyme considerably reduced the levels of the 24 and 30 kDa forms (lane 4) and increased the levels of 20.5 kDa form indicating that both 24 and 30 kDa MSP are N-glycosylated while 20.5 kDa MSP is non-glycosylated (or a deglycosylated) form of the 24 and 30 kDa MSP. On the other hand, mobility of 20 kDa FLP was unaffected after N-glycosidase-F treatment (lane 10) confirming the lack of N-glycosylation in the natural FLP protein. However, further treatment of nMSP (lane 5) with Endoglycosidase H (Endo H) enzyme (which cleaves within the chitobiose core of only high mannose and hybrid oligosaccharides of N-linked glycoproteins)⁴⁷ or treatment with sialidase (lane 6) did not affect the mobility of 24 or 30 kDa MSP. The lack of effect of Endo H indicates that the N-linked glycan moieties in both 24 and 30 kDa MSP are of complex type (which are known to be resistant to Endo H treatment⁴⁷ and not high mannose or hybrid type (which are cleaved by Endo H)⁴⁷. The lack of any effect of sialidase treatment on nMSP indicates that N-linked complex type glycans of both 24 and 30 kDa MSP may not contain terminal sialic acids. Thus, only N-linked glycans are present in 24 and 30 kDa forms of MSP and these are of complex type, which may lack terminal sialic acids. On the other hand, the 20.5 kDa MSP and 20 kDa FLP are non-glycosylated.

Fig. 4.

Treatment of natural MSP and FLP proteins with different glycosidase enzymes. Lane 1, marker; Lane 2, nMSP proteins incubated with O-glycosidase; lane 3, nMSP (incubation control); lane 4, nMSP incubated with N-glycosidase F; lane 5, nMSP incubated with Endo H; lane 6, nMSP incubated with sialidase; lane 7, nMSP incubation control for sialidase treatment; lane 8, nFLP incubated with O-glycosidase; lane 9, nFLP incubation control; lane 10, nFLP incubated with N-glycosidase F. After overnight incubation samples were run on 11% SDS-PAGE and stained with Coomassie.

Analysis of N-linked glycans of natural MSP

Since N-glycosidase treatment indicated the presence of a considerable amount of N-glycan in 24 and 30 kDa natural MSP proteins (their deglycosylated form being 20.5 kDa) further analysis of N-glycan moieties was done by investigating lectin binding properties using a glycan differentiation kit (containing digoxigenin labeled GNA, SNA, MAA, DSA and PNA lectins; Table 1) and also separately investigating binding of 24 and 20.5 kDa natural MSP to immobilized lectins (WGA, Con A and Jacalin) conjugated to sepharose or agarose beads. The known binding specificities of the above lectins are listed in Table-1 and the results of lectin binding are summarized in Table 2. Among all the five digoxigenin-labeled lectins, only DSA showed binding with both 24 and 30 kDa MSP indicating that their N-linked complex type glycans contain Galβ1-4GlcNAc (N-acetyllactosamine) disaccharide (and/or its oligomers) in their structure⁴⁷.

Table 1.

Lectins and their known specificities.

Lectin/agglutinin	Specificity
GNA (Galanthus nivalis)	α1–3 and α1–6 linked mannose of high mannose or hybrid glycan structures
SNA (Sambucus nigra)	Neu5Acα2-6Gal, Neu5Acα2-6GalNAc
MAA (Maackia amurensis)	Neu5Acα2-3Galβ1-4GlcNAc
DSA (Datura stramonium)	Galβ1-4GlcNAc (N-acetyllactosamine) and their oligomers
PNA (Peanut agglutinin)	Galβ1-3GalNAc (O-linked glycans)
WGA (Wheat germ agglutinin)	GlcNAcβ1-4GlcNAc, Neu5Ac (Sialic acid)
Con A (Concanavalin A)	α-D- mannose and α-D-glucose residues, branched α-mannosidic structures (high α-mannose types, hybrid and biantennary complex types of N-glycans)
JAC (Jacalin)	(Neu5Ac) Galβ1-3GalNAc, Galβ1-3GalNAc (O-linked glycans)

Table 2.

Binding of different lectins to 24 and 30 kda MSP.

	GNA	SNA	MAA	DSA	PNA	WGA	CON A	JAC
24 kDa	−	−	−	+	−	+/−	+/−	−
30 kDa	−	−	−	+	−	−	−	−

The lectins GNA, SNA, MAA, DSA and PNA were digoxigenin labeled and were components of a glycan differentiation kit (Roche). All other lectins were Sepharose or agarose conjugates. +/− indicates partial binding of 24 kDa MSP to the lectins (indicating heterogeneity). None of the above lectins bound 20.5 kDa MSP or 20 kDa FLP.

When binding was separately tested by incubation and passage through a Con A-Sepharose column, none of the 30 kDa but only a small fraction of 24 kDa MSP bound with Con A (and could be specifically eluted with α-methyl mannoside). Since the N-glycans in 24 kDa MSP are of complex type (being resistant to Endo H treatment), the Con A binding forms of 24 kDa must be of biantennary complex type (Con A does not bind tri- or tetra-antennary complex type N-glycans)⁴⁷ but not hybrid or high mannose type, which although bind Con A, are sensitive to Endo H⁴⁷. The majority of 24 kDa MSP forms and all 30 kDa MSP, which did not bind to Con A, must contain tri- and/or tetra-antennary complex type N-glycans. The complex type N-glycan nature of 24 and 30 kDa MSP explains their lack of binding with Jacalin and PNA (which bind O-glycans) and also GNA, which binds only hybrid and high mannose N-glycans⁴⁷. Moreover, some 24 kDa MSP forms (but not 30 kDa MSP) were found to bind with WGA and could be specifically eluted with N-acetyl glucosamine, which indicates the presence of terminally linked N-acetyl glucosamine or sialic acid⁴⁷. However, SNA and MAA lectins known to bind terminally linked sialic acid⁴⁷, did not bind at all with 24 or 30 kDa MSP while N-acetyl glucosamine could elute the WGA bound 24 kDa forms of MSP (not shown), indicating the absence of terminal sialic acid but presence of terminal N-acetyl glucosamine in the glycans of WGA bound 24 kDa MSP forms.

Conformational characterization of recombinant FLP and MSP

Far UV-CD spectra of recombinant FLP and MSP proteins (Fig. 5) display a minimum ellipticity near 215 nm and a positive peak below 205 nm, clearly indicating the presence of a secondary structure, typical of β-sheet proteins. Secondary structure prediction using the webserver (CAPITO; CD Analysis and Plotting Tool)³⁶ based on CD spectra of rFLP and rMSP, revealed the presence of 8% α-helix, 52% β-strand and 40% random coil in rFLP and 7% α-helix, 47% β-strand and 44% random coil in rMSP. These structural properties of rFLP and rMSP proteins are very similar to those reported for rat and human OBPs^12,14,48,49.

Fig. 5.

Far UV-CD spectra of purified recombinant FLP and MSP proteins. CD spectra were recorded using rFLP and rMSP proteins at a concentration of 0.1 mg/ml. Ellipticity is displayed as mean residue ellipticity [Θ in 10³deg cm² dmol⁻¹].

Investigations on purified natural MSP and FLP lipocalins to detect the presence of any bound ligand

No bound volatile ligands could be detected upon GC-MS analysis of extracts of freshly purified natural MSP or FLP proteins using the organic solvents dichloromethane or hexane (results not shown). Moreover, while purified MUP lipocalins (from male mouse urine)^50,51, purified boar salivary lipocalin, Sal1⁵² and purified aphrodisin from female hamster vaginal discharge^21–25, all retained a characteristic smell (very likely due to the presence of tightly bound odorous pheromonal ligands), purified nMSP and nFLP had no perceptible smell. All these indicated that MSP and FLP purified from SMG and LG did not contain any naturally bound volatile odorous ligand.

Investigations on in vitro ligand binding property of recombinant FLP and MSP using fluorescence ligand binding assays

Competitive displacement of fluorescent probes is widely used to study the in vitro interactions between OBPs and odorants or other small hydrophobic ligands^{12–16,19,37,48,49}. Ligand binding assays were performed on recombinant MSP and FLP lipocalins using the hydrophobic polarity-sensitive fluorescent probes, 1-AMA and 1-NPN respectively. Polarity-sensitive fluorescent hydrophobic probes e.g. 1-AMA and 1-NPN are extensively used to investigate the binding of various test ligands within the hydrophobic pocket of lipocalins (e.g. OBPs)^{12–16,19,37,48,49}. When these probes are in a hydrophobic environment, such as the ligand-binding cavity of a lipocalin, they exhibit a marked increase in fluorescence intensity when excited at a specific wavelength. The addition of a test ligand that enters and binds within the cavity results in the displacement of the probe and a decrease in fluorescence intensity, indicating ligand binding^12,53. In our ligand binding studies using fluorescence spectroscopy, the fluorescent probe 1-AMA was used as reporter for rMSP and the fluorescent probe 1-NPN was used for rFLP in all competitive displacement assays. Competitor test ligands used were different classes of odorants and a variety of other small organic molecules (fatty acids, steroids, siderophores etc.).

Binding analysis using the probe 1-AMA with rMSP was performed (Fig. 6A and B), the binding constant (K_d) for the rMSP-1-AMA complex was calculated to be ~ 1.86 µM. The calculated apparent dissociation constants (K_i) for IBMP and β-citronellol deduced from their half-maximal concentration required to displace 50% of the bound probe (IC₅₀) were found to be 0.22 and 0.46 µM respectively (Fig. 6C and D). K_i values for other competitor ligands (different odorants) for their binding to rMSP obtained in similar displacement experiments are listed in Table 3.

Fig. 6.

Binding of 1-aminoanthracene (1-AMA), a polarity sensitive fluorescent probe to rMSP and its competitive displacement with odorants, 2-isobutyl-3-methoxypyrazine (IBMP) and β-citronellol. IBMP (bell-pepper odorant) is one of the most potent odorants known and is commonly used in odorant binding studies. β-citronellol is found in oils of rose and it is used in perfumery and as an insect repellent. Panel (A) the fluorescence emission spectra of 1-AMA upon binding with rMSP. rMSP was titrated with an increasing concentration of 1-AMA. Panel (B) Fluorescence titration curve of rMSP with 1-AMA. Panels (C) and (D) show competitive binding assays using odorants IBMP and β-citronellol to rMSP. The protein and the fluorescent probe (1-AMA) were used at 2 µM concentrations, while competitors were added as 1 mM methanolic solutions to the final concentrations indicated. K_d value of probe and IC₅₀ and K_i values of competitors are indicated.

Table 3.

Binding properties of different classes of odorants with rMSP and rFLP by competitive displacement of bound fluorescent probe 1-AMA and 1-NPN respectively.

Odorants	rMSP Ki (µM)	rFLP Ki (µM)
7-carbon odorants
Heptanal	3.96	11.43
2-Heptanone	3.46	14.12
8-carbon odorant
2-Octanone	1.48	4.0
10-carbon odorants
R + Limonene	0.38	0.52
Menthol (2-Isopropyl-5-methoxycyclohexanol)	0.54	2.08
Eucalyptol [1,3,3-Trimethyl-2-oxabicyclo (2,2,2) octane]	0.31	0.24
n-Decanal	0.18	1.15
Decanol	0.13	1.83
± Citronellal	0.11	0.19
3,7-Dimethyl octanol	0.17	0.39
β-Citronellol	0.46	1.17
Geraniol	0.63	1.1
Dihydromercenol	2.1	8.0
11-carbon odorant
Undecanol	0.36	0.78
12-carbon odorant
Dodecanol	0.17	0.17
Pyrazine
2-Isobutyl-3-methoxypyrazine (IBMP)	0.22	1.96
Aromatic odorants
Eugenol (4-Allyl-2-methoxyphenol)	0.35	1.69
Vanillin (4-hydroxy-2-methoxybenzaldehyde)	4.45	4.07
Benzyl benzoate	1.34	6.43

Performing similar binding analysis using the probe, 1-NPN with rFLP (Fig. 7A and B), the binding constant (K_d) for the rFLP-probe complex was calculated to be ~ 1.38 µM. In the displacement experiment of bound probe (1-NPN) using the odorants IBMP and β-citronellol as competitors, the apparent K_i values for the odorants were found to be 1.96 and 1.17 µM respectively (Fig. 7C and D). K_i values for binding of other competitor ligands (different odorants) to rFLP (using 1-NPN as probe) obtained in similar competitive displacement experiments are listed in Table 3.

Fig. 7.

Binding of 1-phenyl-n-naphthylamine (1-NPN; a polarity sensitive fluorescent probe) to rFLP and its competitive displacement with the odorants, 2-isobutyl-3-methoxypyrazine (IBMP) and β-citronellol. Panel (A) The fluorescence emission spectra of 1-NPN upon binding with rFLP. rFLP was titrated with increasing concentration of 1-NPN. Panel (B) Fluorescence titration curve of rFLP with 1-NPN. Panel (C) and (D) Competitive binding assays using odorants IBMP and β-citronellol to rFLP. The protein and the fluorescent probe (1-NPN) were used at 2 µM concentrations, while competitors were added as 1 mM methanolic solutions to the final concentrations indicated. K_d value of probe, IC₅₀ and K_i values of competitors are indicated.

The apparent K_i values obtained for all odorants shown in Table 3 are in the micromolar range, which indicates strong binding. However, the values are not in the nanomolar range, which indicates the possibility of ligand release as would be expected (and also required) of an odorant-binding protein. All 10-carbon odorants were found to be better competitors, for the displacement of bound probes, for both recombinant MSP and FLP. Although, the apparent K_i values obtained for most of the odorants listed in Table 3 are higher for rFLP than for rMSP, it needs to be mentioned here that these K_i value for any odorant should not be compared between rMSP and rFLP since they were obtained using different probes for each lipocalin. The affinities of many of the ligands for rMSP and rFLP (shown in Table 3) are however in the similar range when compared to their affinities reported earlier for human and rat OBPs, which were measured in similar displacement assays^{12,13,37,48,49}.

Among other potential ligands for rMSP and rFLP (tested by fluorescent-probe displacement assay), testosterone, progesterone and 17β-estradiol (steroidal molecules) were found to be extremely weak competitors (not shown). Fatty acids (myristic, palmitic, stearic and oleic), siderophores (2,5-dihrodoxy benzoic acid, 2,3-dihydroxybenzoic acid and deferoxamine), hexadecanol (known to bind aphrodisin²⁵ and 1-octen-3-ol (known to bind bovine OBP⁵⁴ had no competitor activity in the fluorescent-probe displacement assay, indicating their lack of any binding to rMSP and rFLP (results not shown).

Detection of IBMP binding to MSP/FLP after in vitro incubation by GC-MS analysis of organic solvent extracts of the incubated protein

The synthetic molecule IBMP (2-isobutyl-3-methoxypyrazine), known for its bell pepper odor, is widely used as a test odorant to evaluate the odorant-binding properties of OBPs. Our studies using fluorescent ligand binding assays detected that this potent odorant binds with high affinity to rMSP and rFLP. The binding of IBMP was confirmed by GC-MS analysis of organic solvent extracts of these lipocalins, obtained after incubating IBMP in vitro with purified rMSP, rFLP, and a purified mixture of the three natural MSP forms. The analysis was conducted after separating the lipocalins from the unbound free ligand upon elution through a gel-filtration (size-exclusion) column.

The 3 different purified lipocalin preparations were separately incubated in vitro with IBMP (which also absorbs strongly at OD 280 nm) and then each incubation mixture was separately resolved in a gel-filtration (Fig. 8, Panel A; see also legend) followed by GC-MS analysis of organic extracts of specific elution fractions of gel-filtration (Fig. 8, Panel C to E). As a negative control, a mixture of IBMP and chymotrypsin (a non-OBP protein having molecular weight close to MSP/FLP) was also similarly incubated and then resolved in a 4th gel-filtration (Fig. 8, Panel A). Elution fractions of each gel-filtration showed two well-resolved OD 280 nm peaks. The specific elution fractions analyzed were: (i) the fraction containing the first OD 280 nm peak, presumably containing the excluded protein (lipocalin or chymotrypsin) eluting approximately at the void volume of the GF column, (ii) the fraction containing the second OD 280 nm peak (major) having maximum absorbance, which contained the free ligand (IBMP), whose elution was greatly retarded within the GF (size- exclusion) column and was well separated from the first OD 280 nm peak, and (iii) an intermediate fraction having nil or negligible OD 280 nm, which eluted between and well resolved from both the first and second OD 280 nm peaks. Aliquots of each of these 3 fractions from each of the 4 separate gel-filtrations were checked in SDS-PAGE (Fig. 8, Panel B) and protein band was detected only in the first peak of all 4 gel filtrations. The second OD 280 nm peak (major), which showed no protein band in SDS-PAGE, was very likely due to the free IBMP (which was found to have strong absorbance at 280 nm). GC-MS analysis of organic solvent extracts of this major OD 280 nm peaks of the 4 gel filtrations showed, as expected, an abundant IBMP peak (Retention time, RT = 16.01 min) (F-52, F-48, F-52 and F-51 in Panels C, D, E and F of Fig. 8). Similar analysis (not shown) could not detect IBMP in the intermediate fractions (fraction 35 for all) located between the first and second OD 280 nm peaks of each of the 4 gel filtrations. Organic solvent extraction and GC-MS analysis did not detect IBMP in the first peak of gel-filtration of chymotrypsin + IBMP incubation mixture (fraction 23, total ion chromatogram in Panel F inset, Fig. 8), indicating lack of binding of IBMP with the control protein, chymotrypsin. However, IBMP (RT = 16.01 min) at low abundance could be detected in the total ion chromatogram of GC-MS analysis of organic extracts of the fractions of first OD 280 nm peak (containing the protein) in gel-filtration of (Fig. 8; nMSP (fraction 23; Panel D inset), rMSP (fraction 25; Panel C inset) and rFLP (fraction 27; Panel E inset), indicating specific binding of IBMP to each of these lipocalins.

Fig. 8.

Detection of binding of the odorant IBMP with nMSP, rMSP and rFLP proteins by GC-MS analysis of peak protein fractions obtained after gel-filtration of protein-odorant mixtures. Solutions of nMSP, rMSP, rFLP and chymotrypsin (-ve control) were separately mixed with IBMP so that final concentration of IBMP in the mix was ~ 7.2 mM and protein concentration was ~ 200 µM. The 4 mixtures were incubated at 25 °C for 1 h on a rotatory shaker. 1 ml of each mix was separately loaded on a Sephadex-G-25 column and eluted. OD of elution fractions (each 1 ml) was monitored at 280 nm (Panel A). Each gel-filtration showed two well-resolved OD-280 peaks. From each of the four gel-filtrations, aliquots of highest OD fractions of 1st peak and 2nd peak and an intermediate fraction between the 2 peaks (having baseline absorbance) were analyzed in SDS-PAGE (Panel B). Aliquots (0.5 ml) of the same fractions of each gel-filtration (which were run in SDS-PAGE) were checked for presence of IBMP after extraction with equal volume of organic solvent dichloromethane (DCM) and analysis of two µl of the DCM extract in GC-MS. Panels C, D, E and F and their insets show GC-MS analysis (total ion chromatogram) of organic extracts of 1st and 2nd peaks of the 4 gel-filtrations and solvent blanks. In each panel, the total ion chromatogram of 1st peak is additionally shown within inset.

Discussion

The procedure reported in this study (Fig. 2A) for rapid co-purification of all three forms of MSP (24, 20.5 and 30 kDa forms) from crude SMG extract of adult male hamsters is different from and also simpler than our earlier procedure³, which had yielded only a specific Con A binding variety of 24 kDa MSP free from other forms of MSP. In the present purification procedure for MSP (Fig. 2A) a mixture of all three co-expressed forms of MSP protein (24 kDa and 20.5 kDa major forms and 30 kDa minor form), co-precipitated in the highest ammonium sulfate fraction with only trace levels of other proteins present as contaminants. After gel-filtration, this mixture of MSP forms was essentially free of other contaminating proteins (Fig. 2D).

The procedure employed here for the purification of 20 kDa FLP from crude extracts of LG tissue of ovariectomized female hamsters (Fig. 2B) has also been modified and simplified compared to our earlier method². Here, ion-exchange was used as the first step and it was followed with a final gel-filtration step. This yielded purified 20 kDa FLP free from other proteins (Fig. 2D).

It has been demonstrated that the ligand binding specificities of lipocalins can be influenced by presence of phosphorylation^39–45. However, unlike many lipocalins that exhibit phosphorylation^39–45, no evidence of phosphorylation was detected by us in purified MSP or FLP lipocalins (Fig. 3). Moreover, no phosphorylation was detected in these proteins even when they are checked within freshly prepared crude extracts of SMG or LG (Fig. 3). This ruled out possibility of any phosphorylation being present in these MSP and FLP proteins in crude extracts, which might have been lost during their purification.

Glycosidase treatment experiments and lectin-binding studies showed FLP lacks both N- and O-glycosylation and O-glycosylation is also absent in all 3 MSP forms (Fig. 4). Moreover, these studies also showed that 24 kDa and 30 kDa forms of MSP must be heterogeneously N-glycosylated, containing bi-, tri- or tetra-antennary complex type N-glycans with no terminal sialic acid residues, while 20.5 kDa form of MSP was non-glycosylated (Tables 1 and 2). Indeed, MSP protein sequence has a single potential N-glycosylation site (Asn21) while FLP sequence lacks any such N-glycosylation site  (Fig. 1). Interestingly, among the four other homologous lipocalin sequences shown in Fig. 1, N-glycosylation site is predicted only in Syrian hamster aphrodisin at its 2 sites, Asn41 and Asn69, both of which had been experimentally confirmed to be linked with only one N-acetyl glucosamine^23,24,46,55.

The heterogeneity in N-glycosylation present in major 24 kDa and minor 30 kDa MSP and absence of N-glycosylation in the major 20.5 kDa MSP have some interesting aspects. Firstly, in addition to saliva, low levels of MSP proteins are detectable in body fur and urine of male hamsters^1,3,6,7. In encounters and consequent interactions between two male hamsters, we and others⁵⁶ observed that they will typically investigate with their snout, each other’s facial region (including mouth, nostrils, cheeks and ocular surface) and also nuzzle each other’s body fur and anogenital area and female hamsters will also similarly investigate a male (see later for investigations by a male on a female). The heterogenous N-glycosylation within 24 and 30 kDa MSP may manifest as subtle differences between male hamsters in the composition of N-glycoforms present in their saliva, body-fur and urine. Such MSP glycoforms, even in the absence of any bound ligand, might function as contact pheromones to signal a male’s dominance status and social rank to other hamsters similar to that proposed earlier for MUP lipocalins in male mouse urine^57–59. MUP lipocalin isoforms displaying microheterogeneity in protein sequence and expressed predominantly in male mouse urine (also secreted in saliva and tears) have been earlier proposed to function as protein pheromone (even in the absence of any bound ligand), with their heterogeneity serving as barcode to signal the male’s uniqueness to other mice^57–59. Secondly, it is known that glycosylation may alter the conformation, dynamics and overall architecture of a lipocalin, which in turn can impact the ability of the lipocalin to interact with its target ligand(s)^60–62. Therefore, it needs to be determined by analysis of 3D crystal structure (or in silico modelling) whether or not the presence or absence of N-glycosylation at Asn21 in MSP (e.g. 24 kDa versus 20.5 kDa MSP) may influence ligand entry into the putative hydrophobic pocket of MSP. Lastly, the consistent ratio between the levels of N-glycosylated 24 kDa and non-glycosylated 20.5 kDa major MSP lipocalins present in soluble extracts of all male hamster SMGs (24 kDa : 20.5 kDa :: ~4 : 1) as observed in Coomassie-stained SDS-PAGE profiles of SMG extracts (e.g. see Fig. 2A and refs.^1,3,5,6, is intriguing. Overnight incubation of freshly prepared soluble extract of male SMG at 37 °C and then resolving equal aliquots of pre- and post-incubation of such SMG extracts in Coomassie-stained SDS-PAGE, revealed no significant changes in the relative proportion of 24 and 20.5 kDa MSP forms or in their apparent levels (data not shown). This indicated absence of any significant deglycosylation of the 24 kDa MSP after SMG extract preparation and also indicated the stability of both 24 and 20.5 kDa MSP proteins, within the SMG extract. So, does the 20.5 kDa MSP originate upon intracellular deglycosylation of its precursor 24 kDa MSP or is its presence likely to be due to a substantial amount of the nascent MSP molecule escaping co-translational N-glycosylation? Whatever the actual reason behind the presence of 20.5 kDa MSP along with the 24 kDa MSP, it is very puzzling how the apparently consistent ratio between the level of two major forms of MSP is maintained within SMG tissue.

MSP and FLP have considerable protein sequence identity (85%) and in pairwise sequence alignment only 24 amino acids are non-identical and out of these 24, 12 are similar (Fig. 1). 3D crystal structure of rat OBP1f identified 15 amino acid residues lining its ligand-binding pocket (Fig. 1)¹⁰ and from this information, the residues lining the hydrophobic ligand binding pocket of other lipocalins, whose sequences are aligned with rat OBP1f, can be predicted (Fig. 1). Out of these 15 amino acid residues of rat OBP1f, 11 amino acids are identical in both FLP and MSP (12 each in FLP and MSP) (Fig. 1). Among the other pocket lining residues of rat OBP1f, Ala37 is identical in FLP but is substituted by Val in MSP and Val68 is identical in MSP but substituted by a similar amino acid, Ile in FLP (Fig. 1). For the remaining two pocket lining residues, Phe39 and Tyr82 (of rat OBP1f), the former is substituted by Ile/Val in FLP/MSP while the latter is substituted by a similar aromatic amino acid, Phe in both FLP and MSP (Fig. 1). Similar comparisons with the pocket lining residues reported for aphrodisin⁵⁵ and the predicted pocket lining residues of Chinese hamster OBP (deduced from Fig. 1) showed that the Chinese hamster OBP had the best match, having 14 residues, which were identical with those of rat OBP1f, while aphrodisin had only 11 residues, which were identical. Thus, the hydrophobic ligand binding cavities of rat OBP1f, MSP, FLP, aphrodisin of Syrian hamster and also the Chinese hamster OBP have close similarities as well as some distinct differences between each other. These differences can influence ligand binding specificity of these lipocalins and therefore, impact their physiological function.

Overexpression of rMSP and rFLP in E coli and purification from the soluble cell lysate yielded abundant quantities of purified rMSP and rFLP (Fig. 2C and D). Far UV-CD spectra (Fig. 5) confirmed that rMSP and rFLP were folded and contained predominantly beta-sheets, as reported for all other lipocalins^{8,12,14,19,48,49}.

Our results of fluorescence probe displacement assays demonstrate for the first time that rMSP and rFLP proteins can indeed bind a variety of odorants (Figs. 6 and 7 hyperlink 7; Table 3). Our in vitro ligand binding assays also indicated that rMSP binds 1-AMA with strong affinity (Fig. 6) while rFLP binds 1-NPN with strong affinity (Fig. 7 hyperlink 7). When we titrated rFLP with 1-AMA we did not get similar emission spectra (not shown) as we obtained with titration of rMSP with 1-AMA. Likewise, rFLP-1-NPN emission spectra were not similar with rMSP-1-NPN emission spectra (not shown). This difference in probe binding properties of MSP and FLP lipocalins again indicates that these two lipocalins can have different ligand specificity.

MSP and FLP lipocalins were until now assumed to be odorant-binding proteins only on basis of homology search results². Despite the homology search results, organic solvent extraction of MSP or FLP lipocalins (freshly purified from hamster SMG and LG tissues) followed by GC-MS analysis of the organic extracts, did not reveal presence of any bound volatile ligands (data not shown). One reason for this could be that, bound natural ligand(s) may have been lost (i.e. released from the MSP and FLP) during the purification process. However, even after purification of the lipocalins MUPs from male mouse urine, SAL1 from SMG of male pig or aphrodisin from vaginal discharge of female Syrian hamsters, the presence of specific natural ligands could be detected bound to these purified odorant/pheromone binding lipocalins^25,50–52 after organic extraction and GC-MS analysis. A more likely reason for our inability to detect bound ligands within MSP and FLP purified from SMG and LG extracts could be that, the MSP and FLP lipocalins do not bind any specific endogenous ligand within the SMG and LG tissue. However, MSP and FLP after secretion into saliva and tears may bind their cognate ligands from the ambient air while they are present, in the oral cavity, in the tear film (on ocular surface) or within the nasal cavity. Finally, purified MSP and FLP did not have any discernable smell whereas the male-specific boar salivary lipocalin SAL1 (which binds odorous pheromonal ligands⁵² and MUP lipocalin of male mouse urine^50,51 and the lipocalin aphrodisin of Syrian hamster vaginal discharge (which binds certain fatty acids and fatty alcohols)^21,22,25 are all reported to have a characteristic odor even after purification. The above-mentioned observations again indicate that the purified MSP and FLP lipocalins, which do not have any perceptible odor, may not have any volatile ligands bound within them. Notably, no odor was also reported for OBPs purified from nasal glands of several species^11,13,29,33.

Finally, using a direct approach (Fig. 8), purified nMSP (mixture of three natural MSP forms), rMSP and rFLP proteins after being incubated with the odorant IBMP (and subsequent separation from free IBMP by size-exclusion chromatography) (Fig. 8A and B) were shown by GC-MS analysis to retain and specifically carry (i.e. bind) detectable levels of the odorant IBMP as a ligand (Fig. 8C, D and E). In similar experiments, a non-lipocalin protein, chymotrypsin when used as negative control (selected because of its beta barrel structure and its molecular mass of 26 kDa, which is close to that of MSP/FLP), did not retain (i.e. bind) any detectable amounts of IBMP (Fig. 8F). Hence our, overall ligand-binding studies provide sufficient evidence in support of the fact that the lipocalins, male-specific submandibular gland protein (MSP) and female-specific lacrimal gland protein (FLP) of Syrian hamster are indeed sex-specific odorant-binding proteins.

Both MSP and FLP are abundantly and sex-specifically expressed in SMG and LG tissues of Syrian hamsters^1–5. Reports of such high levels of sex-specifically expressed proteins in any mammalian non-reproductive tissue are very rare. The stringent sex-specific expression of MSP and FLP in male SMG and female LG strongly suggests a sex-related function for these lipocalins. However, no endogenous ligand could be detected within MSP and FLP. Therefore, it seems possible that unlike the lipocalins of male mouse urine (MUPs)^50,51, boar saliva (SAL1)⁵² and female hamster vaginal discharge (aphrodisin)²⁵, MSP and FLP do not function as carriers (or reservoirs) of pheromones or odors for their dissemination to the surroundings and sensing by a recipient hamster. However, as suggested above by us, the heterogeneously glycosylated MSP may have a male-specific role in advertisement (dissemination) of male status by acting as a protein pheromone and similar role is also possible for MSP in providing a male-like camouflage to the lactating hamster dams, which, as reported by us earlier, displays a massive and temporary male-like expression of MSP in its SMG³. Since, OBP lipocalins similar to rat nasal OBPs (other than FLP/MSP) could not be detected by us in nasal glands or secretions of Syrian hamsters², FLP/MSP of female and male hamsters may instead function as OBPs, in the tear film (which covers the exposed ocular surface), in oral cavity, in nasal cavity and on the snout (all moistened by saliva and tears). Thus, MSP and FLP may bind and capture odors/pheromones from the air for their delivery to cognate receptors on the main olfactory epithelium (MOE) and/or the vomeronasal organ (VNO) of the same hamster (male or female) facilitating the reception of such external olfactory cues. Even if MSP and FLP moonlight as OBPs what sex-specific function do they perform? It is well known that the odoriferous, pungent smelling vaginal discharge of an estrous female hamster and even aphrodisin protein purified from it, can strongly attract male hamsters and elicit an intense copulatory response (aphrodisiac effect) in them but only after, these are contacted orally (i.e., licked) by the males^{21–23,26,27}. Exposure to only the smell of vaginal discharge failed to elicit such copulatory response in males^26,27. It is possible that during the licking process, the male’s salivary MSP may take up and bind a pheromonally active ligand present free in the discharge (or even released from aphrodisin) and relay this as a ‘baton’ to the male’s VNO, which then elicits the intense copulatory response. The possibility of a role served by the male’s salivary MSP in mediating this copulatory response (after they lick the vaginal discharge) needs investigation. Furthermore, we and others also observed^2,4,5,56 that in their initial interaction with female hamsters, males will almost always investigate the ocular region of females possibly to sense the presence of any female-specific signature, like FLP lipocalin itself, or any female-specific compound known to be present in tears of Syrian hamsters^63–65 which may bind to FLP only after the lipocalin is secreted into tears.

Saliva and tears in Syrian hamsters have been long suspected to be involved in olfaction-mediated chemical communication^56,66,67. Since we have now demonstrated that MSP and FLP lipocalins sex-specifically expressed in SMG and LG of hamsters can bind odorants, it seems quite possible that these odorant-binding lipocalins may have some role in olfaction-mediated chemical communication in hamsters, functioning within an individual, between the sexes or even within a sex. Although, no endogenous natural ligand could be detected within purified MSP and FLP lipocalins, it is still possible that MSP and FLP lipocalins when present in saliva, tears, nasal secretions, body fur and urine of hamsters, might bind and carry specific olfactory cues depending on sex and type of secretion in which they are present or even themselves function as protein pheromone. Thus, in different contexts, MSP and FLP may perform different functions related to perception of olfactory cues or even at times, in their dissemination.

Conclusion

Natural MSP of male Syrian hamster SMG exists in 3 forms, major 24 and 20.5 kDa and a minor 30 kDa. The 24 and 30 kDa forms of natural MSP are heterogeneously N-glycosylated and their glycans are of complex type lacking terminal sialic acid whereas 20.5 kDa natural MSP as well as 20 kDa natural FLP (from LG of female Syrian hamster) have no N- or O-linked glycosylation. Phosphorylation was not detectable in all the three forms of natural MSP or natural FLP. MSP and FLP contain predominantly (47–52%) beta-strand structures typical of lipocalins. Although no volatile ligands could be detected bound within purified natural MSP or FLP, our ligand-binding studies strongly indicate that MSP and FLP lipocalins bind a variety of odorants and hence they are established as sex-specific odorant-binding proteins, which may have roles in hamster chemical communication.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

VPD designed and conducted experiments, analyzed results, prepared manuscript, and revised them. PKD conceived the presented idea, encouraged the investigation, supervised the study and helped to prepare and revise the manuscript.

Funding

This work was partly supported by an internal Institutional funding available at CSIR-CCMB, India and a research grant from the Department of Science and Technology, India. A research fellowship from CSIR-India supported VPD.

Data availability

“The datasets analyzed during the current study are already provided in the manuscript.”

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

All the animal procedures used in this study were approved by the Institutional Animal Experimentation Ethics Committee of the Centre for Cellular and Molecular Biology, Hyderabad, India. We confirm that all experiments in this study were performed in accordance with the relevant guidelines and regulations. All the procedure of the study followed the ARRIVE guidelines.