The monoclonal antibody TER-119 recognizes a molecule associated with glycophorin A and specifically marks the late stages of murine erythroid lineage


Dr Tatsuo Kina, Department of Immunology, Institute for Frontier Medical Sciences, Kyoto University, Sakyo-ku, Kyoto 606-8507, Japan. E-mail:


The antigen specificity of a rat monoclonal antibody TER-119 was investigated. In adult mice, TER-119 reacted with mature erythrocytes, 20–25% of bone marrow cells and 2–3% of spleen cells but not with thymocytes nor lymph node cells. In fetal haematopoietic tissues, 30–40% of d 10 yolk sac cells, 80–90% of d 14 fetal liver cells and 40–50% of newborn liver cells were reactive with TER-119. TER-119+ cells in adult bone marrow expressed significant levels of CD45 but not myeloid (Mac-1, Gr-1) or B-cell (B220) markers. Morphological examination and haematopoietic colony-forming assays for isolated TER-119+ cells revealed that TER-119 reacts with erythroid cells at differentiation stages from early proerythroblast to mature erythrocyte, but not with cells showing typical erythroid blast-forming unit (BFU-E) and erythroid colony-forming unit (CFU-E) activities. Erythroleukaemia cell lines do not express the TER-119 antigen even after stimulation with dimethylsulphoxide. TER-119 immunoprecipitated protein bands with molecular masses of 110 kDa, 60 kDa, 52 kDa and 32 kDa from erythrocyte membrane, whereas only a 52-kDa band was detected by TER-119 in Western blot analysis. Further molecular and cellular analyses indicated that the TER-119 antigen is a molecule associated with cell-surface glycophorin A but not with glycophorin A itself.

Haematopoietic cells committed to a particular lineage are considered to be marked by a set of cell-surface molecules unique to that lineage. Monoclonal antibodies (mAbs) that identify such lineage-specific markers are of potential value in studies of cellular differentiation and identification of the cell-surface components involved (Ledbeter & Herzenberg, 1979). These mAbs could also be useful for isolating haematopoietic stem cells or progenitors at various developmental stages (Visser et al, 1984; Spangrude et al, 1988; Ikuta et al, 1990; Kawamoto et al, 1998). Although a number of lineage-specific markers have been established for lymphoid and myeloid lineages, mAbs that can specifically mark the erythroid lineage are still very limited in mice.

The cell-surface sialoglycoproteins (glycophorins) have been shown to be selectively expressed in erythroid cells (Marchesi et al, 1976). Thus, mAbs against glycophorins have been considered as specific markers for erythroid lineage in humans (Robinson et al, 1981; Anstee & Edwards, 1982; Bigbee et al, 1983). Although the expression of glycophorin molecules has been well studied in human erythroid cells, little is known about glycophorin expression in the mouse. There are several mAbs that are shown to be reactive with murine erythroid cells, but none of them have been demonstrated as specific markers for erythroid lineage (Britt et al, 1984; Ardman et al, 1987; Bacon & Sytkowski, 1987). During the preparation of mAbs to mouse fetal antigens, we previously established a hybridoma clone TER-119 which produces a rat mAb selectively reactive with both fetal and adult erythroid cells (Ikuta et al, 1990).

In this study, we investigated the reactivity of TER-119 with mouse haematopoietic cells and the biochemical nature of the antigen recognized by this mAb. Our results demonstrated that TER-119 was highly specific for erythroid cells at the stages from early proerythroblast to mature erythrocyte and that TER-119 recognized a cell-surface molecule which is strongly associated with glycophorin A. It was found that the TER-119 antigen was expressed only on normal erythroid cells but not on erythroleukaemia cells, even after induction of these cells with dimethylsulphoxide (DMSO).

Materials and methods

Animals BALB/c mice and Wistar rats were purchased from SLC (Shizuoka, Japan). For fetuses, the day of appearance of the vaginal plug was designated as d 0 of gestation. Adult mice were used at 10–12 weeks of age.

Cell lines and culture T-cell lines (BW5147, EL4), B-cell lines (DW34, 70Z/3), macrophage lines (P388D1, RAW264·7), a mast cell line (P815), a fibroblast line (BALB-3T3) and a thymic epithelial line (TEC4C18) were maintained in our laboratory. Murine erythroleukaemia lines (MEL) (745A, TSA-8, T3-Cl-1, ELM-1) were kindly provided by Dr K. Ito (Kyoto University). To induce haemoglobin synthesis, MEL cells were cultured with 1·8% DMSO in Dulbecco's minimum essential medium (DMEM) containing 10% fetal calf serum (FCS) for 3 d, washed once with fresh medium to remove DMSO and cultured another 3 d without DMSO. Cultured cells were analysed by FACScan for the expression of the TER-119 antigen and glycophorin A. A portion of cells were also assessed for haemoglobin synthesis with benzidine staining.

Antibodies mAbs RA3-6B2 (anti-B220), M1/70 (anti-Mac-1), RB6-8C5 (anti-Gr-1), IM7.8 (anti-CD44) and MAR18.5 (anti-rat κ chain) were obtained from the American Type Culture Collection (Manassas, VA, USA). ALI-4A2 (anti-CD45.2) was a gift from Dr G. J. Spangrude (University of Utah, UT, USA). These mAbs were purified by affinity chromatography on a protein G Sepharose column (Sigma Chemicals, St Louse, MO, USA) and were biotinylated as previously described (Harlow & Lane, 1988). Normal rat IgG (RIgG) was purified from Wistar rat serum by affinity chromatography on a protein G column and was used as an isotype-matched control antibody. Fluorescein isothiocyanate (FITC)- or phycoerythrin (PE)-conjugated goat anti-rat IgG and streptavidin were purchased from Caltag (South San Francisco, CA, USA). Polyclonal rabbit anti-mouse glycophorin A serum (anti-GPA) was kindly donated by Dr Y. Matsui (Tohoku University, Sendai, Japan) (Matsui et al, 1985).

Flow cytometry and cell sorting For staining, cells were suspended in Hanks' balanced salt solution containing 2% calf serum and 10 mm HEPES (HBSS-CS) with an appropriate dilution of antibodies, and then stained with FITC- or PE-labelled anti-rat IgG. As a negative control, cells were stained with control rat IgG (1 µg/ml), followed by FITC- or PE-labelled anti-rat IgG Ab. Stained cells were analysed by FACScan (Becton & Dickinson, San Jose, CA, USA). Dead cells were excluded by gating with forward light scatter and propidium iodide staining. For sorting, bone marrow cells were first treated with a hypotonic shock to remove mature erythrocytes and stained with TER-119 followed by FITC–MAR18.5. Stained cells were analysed by FACSvantage and defined as TER-119 and TER-119+ populations based on the background staining obtained with control Ab. The sorting gate for TER-119+ cells was set at least 20 channels above the gate for TER-119 cells. The purity of sorted populations was more than 95% for the TER-119+ population and nearly 100% for the TER-119 population, as judged by reanalysis with FACScan.

Hybridoma production Wistar rats were subcutaneously injected with day 14 BALB/c fetal liver cells, and the spleen cells were fused with X63.Ag8.653 myeloma cells, as previously described (Galfre et al, 1977). A hybridoma clone TER-119 producing a rat IgG2b Ab reactive with fetal liver cells was established.

Colony-forming assay Haematopoietic colony-forming assays were performed as previously described (Nijhof & Wierenga, 1983). Briefly, varying numbers of cells (104−5 × 105) were plated in 35-mm dishes in the presence of 0·4% methylcellulose in Iscove's modified Dulbecco's medium (IMDM; Gibco, Grand Islands, NY, USA) supplemented with 20% FCS, 5 × 10−5 m 2-mercaptoethanol, 1% bovine serum albumin, 100 µg/ml streptomycin and 100 U/ml penicillin. For erythroid colonies, cultures were supplemented with 0·2 U/ml of recombinant human erythropoietin (EPO) (Kirin Breweries, Tokyo, Japan) in the presence (for erythroid blast-forming units; BFU-E) or absence (for erythroid colony-forming units; CFU-E) of 10% pokeweed mitogen-stimulated spleen cell-conditioned medium (PWM-CM). For granulocyte–macrophage colony-forming unit (CFU-GM) colonies, cultures were supplemented with 10% PWM-CM. Erythroid colonies were scored at 3 d (for CFU-E colonies) or 10 d (for BFU-E colonies) after culture and haemoglobin synthesis was assessed by benzidine staining. CFU-GM colonies were counted at d 7 after culture and stained with May–Grunwald–Giemsa solution for microscopic analysis.

Western blot analysis Erythrocyte ghosts were prepared by suspending peripheral blood erythrocytes in 5 mm sodium phosphate solution (pH 8·0) and pelleted by centrifugation at 22 000 g for 10 min at 2°C. The ghosts were extracted with lysis buffer (10 mm tris-HCl, pH 8·0, containing 1·0% Triton X-100, 0·15 m NaCl, 2 mm phenylmethylsulphonyl fluoride and 10 U/ml of aprotinin) at 4°C, and the lysates (30 µg of protein/lane) were separated by 10% SDS polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions, then electrophoretically transferred to a polyvinylidene difluoride (PVDF) membrane. After blocking with 4% non-fat milk in Tris-buffered saline (TBS, pH 8·0), the membrane was reacted with 1 µg/ml of TER-119 or rat IgG for 1 h, followed by biotinylated MAR-18.5 (0·3 µg/ml). The membrane was then reacted with streptavidin-conjugated alkaline phosphatase (0·1 µg/ml) (Vector Laboratories, Burlingame, CA, USA) and visualized by adding substrate solution containing 5-bromo-4-chloro-3-indolyl phosphate and nitroblue tetrazolium (BCIP/NBT solution) (Promega, Madison, WI, USA).

Immunoprecipitation and peptide mapping Erythrocytes were radiolabelled with Na125I by a lactoperoxidase method, as previously described (Kina et al, 1991), and extracted with lysis buffer. The lysates were immunoprecipitated with protein-G Sepharose precoated with 10 µg of TER-119 or control rat IgG, and the immunoprecipitates were analysed by SDS-PAGE in 10% gel. Molecular weights of protein bands were calculated by comparison with molecular weight markers (BioRad, Richmond, CA, USA). In situ peptide mapping analysis was carried out as described previously (Gooderham, 1984). Briefly, after SDS-PAGE analysis, the protein bands were cut out, swelled and equilibrated in 125 mm Tris buffer (pH 6·8) containing 2 mm EDTA and 0·1% SDS. Gel slices were then loaded onto a two-dimensional polyacrylamide gel (12·5%) together with 0·5 mg of Staphylococcus aureus V8 protease (Sigma) in 50 µl of reducing SDS sample buffer. The samples were left for 20 min to allow enzymatic digestion before SDS-PAGE.

Northern hybridization analysis Total RNA was extracted from 745A cells or bone marrow cells using the RNA extraction kit (Pharmacia Biotech, Sweden), and 20 µg/lane of RNA was separated on a 1% agarose gel containing formaldehyde and transferred to a nylon membrane filter. After prehybridization with salmon sperm DNA (100 µg/ml), the membrane was hybridized with the [32P]-labelled PstI fragment (1·5 kb) of mouse glycophorin A cDNA (pGP315) (Matsui et al, 1989). A cDNA probe for β-actin was used as an internal control. The membrane was hybridized with labelled probes at 42°C for 16 h in 5× SSC buffer containing 50% deionized formamide, Denhardt's solution, 1% SDS and 200 µg/ml salmon sperm DNA. After washing with 2× SSC containing 0·5% SDS at 65°C and 0·1× SSC at room temperature, the membrane was subjected to autoradiography.

Statistical analysis All data are presented as the means ± standard errors of the mean (s.e.m.). Differences between groups were determined by Student's t-test for paired data using the statview 5·0 program (Abacus Concept, Berkeley, CA, USA). P-values less than 0·05 were considered to be significant.


Reactivity of TER-119 with haematopoietic cells

We first investigated the reactivity of TER-119 with haematopoietic cells of adult and fetal mice. The results obtained in four separate experiments are summarized in Table I. In adult mice, TER-119 stained 20–25% of bone marrow cells and 2–3% of spleen cells, but no thymocytes or lymph node cells were positive for TER-119 expression. All mature erythrocytes were reactive with TER-119. In fetal mice, 30–40% of day 10 yolk sac cells and 80–90% of d 14 liver cells were stained with TER-119. The percentage of TER-119+ cells in fetal liver declined after day 14, but 40–50% of newborn liver cells were still TER-119+ (Table I).

To investigate the relationship between the TER-119 antigen and other lineage markers, two-colour analysis was performed for adult bone marrow cells. As shown in Fig 1, TER-119+ cells of adult bone marrow express significant levels of CD45 but not the Mac-1, Gr-1 or B220 markers. Treatment of bone marrow cells with a hypotonic shock strongly reduced the number of TER-119+ cells but not that of other lineage marker positive cells (not shown), suggesting that TER-119 is mainly reactive with erythroid cells. TER-119+ cells in fetal liver and yolk sac do not express other lineage markers (not shown). In contrast to adult bone marrow, TER-119+ cells in the fetal liver and yolk sac were negative for CD45 (Fig 2) and a hypotonic treatment did not reduce the percentage of TER-119+ cells in the yolk sac and fetal liver, suggesting that TER-119+ cells in fetal mice appear to be composed of nucleated erythroid cells.

Figure 1.

Figure 1.

Flow cytometry of bone marrow cells. Adult BALB/c bone marrow cells were two-colour stained with TER-119 vs. various lineage markers (B220, Mac-1, Gr-1 and CD45) and were analysed by FACScan. Quadrant gates were set to indicate negative and positive populations, as determined using isotype-matched control rat IgG.

Figure 2.

Figure 2.

Flow cytometry of fetal haematopoietic cells. Day 10 yolk sac and d 14 fetal liver cells of BALB/c mice were two-colour stained with TER-119 vs. CD45 and were analysed by FACScan.

Erythroleukaemia lines do not express the TER-119 antigen

The expression of the TER-119 antigen was investigated in various cell lines, and the results are shown in Table I. We were unable to detect TER-119 antigen expression in any cultured cell lines, including T-cell lines (BW5147, EL4), B-cell lines (DW34, 70Z/3), macrophage lines (P388D1, RAW264.7), a mast cell line (P815), a fibroblast line (BALB-3T3) or an epithelial line (TEC4C18). Surprisingly, murine erythroleukaemia lines (745A, TSA-8, T3-Cl-1 and ELM-1) were also negative for TER-119 expression, and this phenotype was not altered even after stimulation with DMSO (Table I). These results suggest that the ability to express the TER-119 antigen might be lost during the leukaemogenic process that leads to the evolution of these cell lines.

Morphological examination of TER-119+ cells

To examine the morphological characteristics of TER-119+ cells, BALB/c bone marrow cells were stained with TER-119 and sorted by FACSvantage. The morphological features of sorted TER-119+ cells are shown in Fig 3. The TER-119+ population consisted of many nucleated erythroblasts and some residual erythrocytes. Most of the nucleated erythroblasts were small- to medium-sized cells with deeply stained nuclei (indicated by arrows). Some of the TER-119+ cells were large nucleated cells that contained basophilically stained granules in the cytoplasm (indicated by an arrowhead). Erythrocyte aggregates which survived the hypotonic treatment were also enriched in the TER-119+ population. These results clearly indicate that TER-119 specifically reacts with erythroid cells at the proerythroblast to mature erythrocyte stages.

Figure 3.

Figure 3.

Morphology of TER-119+ cells. Unseparated and FACS-sorted TER-119+ bone marrow cells of BALB/c mice were stained with May–Grunwald–Giemsa solution. (Left) Unseparated cells; (right) TER-119+ cells. Arrows indicate small- to medium-sized mature erythroblasts and an arrowhead indicates a large proerythroblast with basophilic staining.

Colony formation of TER-119+ cells

Adult bone marrow cells were separated by FACSvantage as TER-119+ and TER-119 populations, and their abilities to form BFU-E, CFU-E and CFU-GM colonies were assessed. Colonies containing more than 30 cells/colony were scored as CFU-E colonies. As shown in Table II, similar numbers of BFU-E, CFU-E and CFU-GM colonies were generated in both unseparated and TER-119 populations. On the other hand, the TER-119+ cells generated only a few CFU-E colonies, and no BFU-E and CFU-GM colonies were observed in this population. Therefore, we concluded that almost all CFU-E, BFU-E and CFU-GM colony-forming activities are located in the TER-119 population.

Biochemical analysis of the molecule recognized by TER-119

To investigate the biochemical nature of the TER-119 antigen, Western blot analysis and immunoprecipitation experiments were performed for erythrocyte membranes. As shown in Fig 4A, a single 52-kDa band was detected by TER-119 in Western blot analysis. However, by immunoprecipitation, TER-119 precipitated four protein bands corresponding to 110 kDa, 60 kDa, 52 kDa and 32 kDa, and all these bands were also precipitated with rabbit antiserum to mouse glycophorin A (anti-GPA) (Fig 4B). The 32-kDa band corresponded to a monomeric form of glycophorin A (Matsui et al, 1985) and the 60-kDa band seemed to be a dimer of the 32-kDa protein. This was confirmed by peptide mapping analysis with V8 protease. We isolated the bands labelled a–d in Fig 4B, which were digested with V8 protease and analysed by SDS-PAGE. As shown in Fig 4C, digestion with V8 protease yielded similar peptide patterns between bands a and c and between bands b and d. No peptide bands are shared between these two patterns, suggesting that bands a and c are structurally unrelated to bands b and d. These results strongly suggest that TER-119 recognizes a 52-kDa molecule which is associated with glycophorin A on the erythrocyte membrane. We speculate that the 110-kDa band might have been formed by the dimerization of the 52-kDa molecule under immunoprecipitation conditions.

Figure 4.

Figure 4.

Biochemical analysis of erythrocyte membranes. (A) Western blot analysis. Membrane extract of erythrocytes was separated by SDS-PAGE (10% gel) under reducing conditions and transferred to a PVDF membrane. The membrane was reacted with 1 µg/ml of rat IgG or TER-119 followed by biotinylated MAR18.5 and streptavidin alkaline phosphatase and was visualized by BCIP/NBT solution. An arrowhead indicates a 52-kDa band detected by TER-119. (B) Immunoprecipitation. 125I-Labelled erythrocyte lysates were immunoprecipitated with protein G Sepharose precoated with anti-GPA, TER-119 or rat IgG, and the immunoprecipitates were run on SDS-PAGE (10% gel) under reducing conditions and were subjected to autoradiography. GPA, glycophorin A; RIgG, rat IgG. (C) Peptide mapping. The protein bands a–d in (B) were excised from the gel and were loaded onto a 12·5% polyacrylamide gel together with V8 protease. After 20 min digestion, samples were run by SDS-PAGE and were subjected to autoradiography.

Erythroleukaemia line 745A expresses glycophorin A but not the TER-119 antigen

To investigate the relationship between glycophorin A and the TER-119 antigen, we examined the expression of these molecules on 745A erythroleukaemia cells before and after differentiation induction by DMSO. Glycophorin A expression was examined by both Northern blot and flow cytometric analyses. The expression of TER-119 antigen and CD44 was monitored by flow cytometry. As shown in Fig 5A, unstimulated 745A cells showed significant levels of glycophorin A messages in Northern analysis and the expression level was greatly enhanced by stimulation with DMSO. Similar results were obtained in flow cytometric analysis with anti-GPA (Fig 5B). On the other hand, no TER-119 antigen was detected in either unstimulated or stimulated 745A cells (Fig 5B). It was noted that CD44 expression by 745A cells was decreased after stimulation with DMSO. These results clearly indicate that the TER-119 antigen is not identical to glycophorin A.

Figure 5.

Figure 5.

Glycophorin A expression in bone marrow cells and erythroleukaemia cells. (A) Total RNA was prepared from BALB/c bone marrow cells (BM) or from 745A erythroleukaemia cells unstimulated or stimulated with DMSO (1·8%). RNAs (20 µg/lane) were electrophoresed in 1% agarose gel and blotted onto a nylon membrane. The membrane was hybridized with 32P-labelled cDNA probes for mouse glycophorin A (GPA) and β-actin and were autoradiographed. (B) 745A cells unstimulated or stimulated with DMSO were stained with anti-GPA, TER-119 or anti-CD44 antibodies followed by FITC-conjugated secondary antibody and were then analysed by FACScan. Broken line, staining control; thin line, unstimulated 745A; thick line, DMSO-stimulated 745A.


In mice, several mAbs that react with the erythroid lineage have been reported. They are directed either to a gp70 retroviral antigen (Britt et al, 1984) or to an idiotypic determinant on the viral receptors (Ardman et al, 1987). Expression of these determinants appears to be limited only to a certain stage of erythroid cells or to erythroleukaemia cells. Another mAb ERY-1, which was raised against Rauscher erythroleukaemia cells, has been reported to be reactive with cells at CFU-E to erythroblast stages but not with mature erythroid cells (Bacon & Sytkowski, 1987). This mAb seems to recognize a molecule involved in the EPO-induced differentiation event of erythroid cells. Thus, TER-119 is different from those mAbs regarding the specificity to erythroid cells and the biochemical nature of the antigen.

From the morphological study and haematopoietic colony-forming assay, we concluded that the TER-119+ population in adult bone marrow lacks typical BFU-E and CFU-E colony-forming activities (Fig 3; Table II). However, we noted that a number of small clusters composed of 20–30 haemoglobin-positive cells were generated in CFU-E assay of TER-119+ cells, suggesting that the earliest TER-119+ cells still possessed the proliferative potential in vitro. So far, we have not detected TER-119 expression in any erythroleukaemia lines, even those induced by different leukaemogenic conditions (Friend virus or radiation) (Table I). Because these erythroleukaemia lines are considered to be transformed at the CFU-E stage, these results support the idea that the TER-119 antigen is not expressed by cells with typical CFU-E activity. Stimulation of these erythroleukaemia cells with DMSO was not effective for inducing the TER-119 antigens (Table I and Fig 5), suggesting that they have lost the mechanism to express the TER-119 antigen.

CD45 is a transmembrane protein which is expressed from early stem cells to all mature cells of haematopoietic origin except erythroid lineage (Nishikawa et al, 1998). It has been shown that CD45 expression is only confined to the progenitor cells with BFU-E and CFU-E activities (Scheid & Triglia, 1979). These results are mostly derived from the results obtained in fetal haematopoietic tissues and are consistent with our present findings (Fig 1). However, in adult bone marrow, we observed significant levels of CD45 expression in TER-119+ cells (Fig 1). Because CD45 is a tyrosine-specific phosphatase which is known to regulate the signalling of haematopoietic cells, our results may suggest that CD45 is functionally involved in the regulation of later stages of erythroid differentiation in the adult bone marrow.

Glycophorins are shown to be expressed on erythroid cells at stages from basophilic erythroblasts to mature erythrocytes (Gahmberg et al, 1978), whereas their function has not yet been established. In humans, mature erythrocytes synthesize at least three kinds of glycophorin molecules (designated as glycophorins A, B and C). Their molecular sizes in the monomeric form are reported to be ≈ 31 kDa, 25 kDa and 28 kDa for glycophorins A, B and C respectively (Anstee et al, 1979). From the amino acid and nucleotide sequences of human glycophorins, it has been showed that glycophorins A and B are structurally related (Colin et al, 1986; Siebert & Fukuda, 1986; Tate & Tanner, 1989). In mice, however, glycophorins have not been extensively characterized, and their expression in developing erythroid cells is largely unknown. So far, only glycophorin A has been well studied, and its cDNA clone has been isolated (Matsui et al, 1989; Gu et al, 1991). The mouse glycophorin A resembles its human counterpart in that it is heavily glycosylated with sialyl residues and easily forms dimers at higher concentrations (Dolci & Palade, 1989). The characteristics of the 52-kDa molecule recognized by TER-119 resemble those of glycophorins, in that it is glycosylated with sialyl residues as revealed by staining with the periodic acid Schiff reaction (unpublished observations) and it easily dimerizes under immunoprecipitation conditions (Fig 4). However, there are considerable differences between the TER-119 antigen and typical glycophorins concerning the molecular weight and the distribution in erythroid cells (Ulmer et al, 1989; see also Figs 4 and 5). No glycophorins corresponding to 52 kDa in the monomeric form have been reported either in humans or in mice. Our attempts to isolate the TER-119 antigen with affinity chromatography also suggest that TER-119 is not a protein abundantly expressed on the erythrocyte membrane (unpublished observation). Taken together, these results strongly suggest that the TER-119 antigen is not a typical glycophorin but a molecule associated with it.


We thank Dr Y. Matsui (Tohoku University, Sendai, Japan) for generously providing us with cDNA and rabbit antiserum to mouse glycophorin A, and Dr K. Ito (Kyoto University, Kyoto, Japan) for erythroleukaemia cell lines. We are also grateful to Dr G. J. Spangrude (University of Utah, Salt Lake, UT, USA) for his valuable advice and critical review of our manuscript.