Mol. Hum. Reprod. Advance Access originally published online on December 5, 2005
Molecular Human Reproduction 2005 11(10):715-717; doi:10.1093/molehr/gah223
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Soluble HLA-G isoforms: technical deficiencies lead to misinterpretations
1University of Kansas Medical Center, Kansas City, KS and 2Fred Hutchinson Cancer Research Center, Seattle, WA, USA
3 To whom correspondence should be addressed at: Department of Anatomy and Cell Biology, University of Kansas Medical Center, 3901 Rainbow Blvd, Kansas City, KS 66160, E-mail: jhunt{at}kumc.edu
 |
Introduction |
|---|
 Top Introduction Cell preparationsUse of irrelevant cellsAntibody specificityReaching the maternal...Cell lineage and expression...Copy numberPCR experimentsReferences |
|---|
The recent report authored by Blaschitz et al. (in press)
would lead the reader to believe that human trophoblast cells cannot synthesize or secrete any of the soluble isoforms derived from splice variants of the HLA-G message. They propose that all of the soluble HLA-G presumed to be produced in trophoblast cells is, in fact, cleaved membrane HLA-G1. An interesting idea.
Yet several published works have presented solid, well documented evidence that argues against this position. The original article describing the molecular structure of soluble HLA-G (Fujii et al., 1994) showed that soluble protein encoded by the intron 4-containing mRNA could be clearly distinguished from membrane HLA-G1 by isoelectric focusing (IEF). Further, a class I protein (reactive with W6/32) with an IEF pattern completely overlapping with soluble HLA-G from transfected cells was demonstrated in JEG-3 choriocarcinoma cells. Also in that study, PCR primers specific for soluble HLA-G-encoding mRNA were clearly described and used to confirm the protein work.
In 2002, Solier and colleagues showed that soluble, intron-4 retaining HLA-G1 (now known as HLA-G5) mRNA is present in villous cytotrophoblast cells and that the HLA-G5 protein is secreted by the cells (Solier et al., 2002). In 2003, a team working in the Geraghty laboratory reported that membrane HLA-G1 is essentially absent in placentas, whereas HLA-G5 is abundant in both villous cytotrophoblast cells and syncytiotrophoblast (Ishitani et al., 2003). HLA-G1 was identified only in migrating extravillous cytotrophoblast cells using the highly specific o1G monoclonal antibody. Also in that study, the soluble HLA-G1 specific antibody 16G1 was used to isolate protein directly from term placentas. Bound peptide was isolated from this complex and shown to have a nearly identical profile to peptide isolated from membrane HLA-G1, which was also isolated from placentas. These peptides collectively matched very precisely the peptide binding motif previously described for HLA-G (some peptides were identical), and that work left no doubt as to the source of that peptide by N-terminal sequencing of the heavy chain (Lee et al., 1995).
Taken together, these facts leave no doubt that 16G1 does bind HLA-G produced in vivo, and can only be reconciled with the conclusions made by Blaschitz study if one posits that 16G1 instead binds membrane HLA-G1. That conclusion in turn contradicts several other studies including some of the results presented in the Blaschitz et al. paper (Blaschitz et al., in press). At the next level of study alternative HLA-G protein forms were investigated. Newly generated, fully characterized, isoform-specific monoclonal antibodies developed in the Hunt laboratory demonstrated that villous cytotrophoblast cells contain HLA-G5 and HLA-G6 mRNA but express only HLA-G5 protein (Morales et al., 2003). By contrast, expression of HLA-G2/G6 protein was identified exclusively in extravillous cytotrophoblast cells.
Despite this strong and consistent evidence put forth by three independent laboratories, it is always important to critically examine new evidence contradicting previously held conclusions. Taking the major features of the Blaschitz paper one at a time, let us look at the evidence they present to support their conclusions.
|
Cell preparations |
|---|
|
TopIntroductionCell preparationsUse of irrelevant cellsAntibody specificityReaching the maternal...Cell lineage and expression...Copy numberPCR experimentsReferences |
|---|
There is little disagreement that highly purified cells must be used to study the expression patterns of closely related genes and their protein products as cell type is often related to expression. Yet this attention to detail does not characterize the Blaschitz paper. For example, the technique used in the Blaschitz paper to acquire chorionic cytotrophoblast cells yields not only these cells but also fetal mesenchymal cells lying between the amnion and chorion, and maternal decidual cells and macrophages threaded through the chorion laeve. It is difficult to interpret HLA expression patterns in this mixture of cell lineages. Similarly, the first trimester cytotrophoblast cell harvests appear from staining with MEM-G/01 to be a mixture of villous (negative) and extravillous (positive) cytotrophoblast cells. It is difficult to determine just how data on HLA-G expression that is obtained from a mixture of precursor and differentiated trophoblast cells might be evaluated.
|
Use of irrelevant cells |
|---|
|
TopIntroductionCell preparationsUse of irrelevant cellsAntibody specificityReaching the maternal...Cell lineage and expression...Copy numberPCR experimentsReferences |
|---|
Normal and tumour cells of the same lineage often differ. Thus, although tumour cells are attractive models because of their uniformity they cannot be substituted for normal purified cells. For example, the trophoblast tumour cell lines, Jar and JEG-3, use tumour necrosis factor alpha (TNF
) as a growth factor, whereas this factor kills normal cytotrophoblast cells (Yang et al., 1993; Yui et al., 1994). The authors wish to support their conclusions with information on the AC1-M59 fused cells. Because these cells are artificially constructed and never found in nature and because HLA expression is cell lineage-restricted, the relevance of these experiments is uncertain at best. Even when investigators are not able to obtain pure cell populations, they usually acknowledge the limitations of using a cell line or characterize the cells that they use, but such discussions are lacking in the Blaschitz paper. |
Antibody specificity |
|---|
|
TopIntroductionCell preparationsUse of irrelevant cellsAntibody specificityReaching the maternal...Cell lineage and expression...Copy numberPCR experimentsReferences |
|---|
It is not only essential to use highly purified primary cells but also to choose carefully the tools for dissecting antigen expression. Yet two of the antibodies used in this study require association with ß2m to generate a conformationally correct, recognizable epitope [4H84 (Polakova et al., 2004
), MEM-G/09 (Menier et al., 2003)]. This would eliminate the possibility of detecting free G1 heavy chain as well as HLA-G2 and HLA-G6. Three antibodies are well known to bind either non-HLA-G epitopes or other HLAs (4H84, HCA2, 16G1), leaving Figure 3 uninterpretable. All investigators who have used 16G1 in immunoblots know that many proteins are recognized in placental lysates, although signals from transfected cells are clear and may be highly useful for some studies.
Extensively characterized monoclonal antibodies recognizing HLA-G5 and -G6 have been generated that are entirely isoform specific. They do not bind to epitopes on non-HLA-G proteins, do not recognize intron 4, ß2m or HLA-G1 and do not cross-react with HLA class Ia antigens (Morales et al., 2003). The authors of the present paper might have reached different conclusions if they had used these isoform-specific antibodies.
|
Reaching the maternal circulation |
|---|
|
TopIntroductionCell preparationsUse of irrelevant cellsAntibody specificityReaching the maternal...Cell lineage and expression...Copy numberPCR experimentsReferences |
|---|
It is difficult to understand why the authors would consider it biologically reasonable to propose that HLA-G1 fragments account for the high levels of soluble HLA-G found in pregnancy sera (Hunt et al., 2000a
; Pfeiffer et al., 2000). All investigators, including Blaschitz et al. agree that HLA-G1 is expressed only by extravillous cytotrophoblast cells that are secluded from the maternal blood circulation except when located in the uterine spiral arteries, although production in other types of maternal cells such as macrophages has not been eliminated. Ishitani et al. (2003) have definitively demonstrated the expression of HLA-G1 in extravillous cytotrophoblast cells using the monoclonal antibody, o1G.
Oddly, authors of the Blaschitz paper insist that there is no HLA-G5 in maternal circulation when three independent laboratories (Fujii et al., 1994; Lee et al., 1995; Solier et al., 2002; Morales et al., 2003) have shown that trophoblast cells in placental villi are the source of soluble HLA-G (HLA-G5) and do not contain HLA-G1. Villous syncytio- and cytotrophoblast cells in placental villi are obviously the most likely to contribute soluble HLA-G because of their unique positioning in direct apposition to maternal blood. Blaschitz et al. do not address the question of how they conceive cleavage products of HLA-G1 to arrive in maternal sera.
|
Cell lineage and expression of HLA |
|---|
|
TopIntroductionCell preparationsUse of irrelevant cellsAntibody specificityReaching the maternal...Cell lineage and expression...Copy numberPCR experimentsReferences |
|---|
Blaschitz et al. comment extensively on an original suggestion that the soluble HLA-G in maternal blood might be HLA-G6 (Hunt et al., 2000a
) but fail to mention that an alternative interpretation of the data was offered the same year that focused on the association of trophoblast cell HLA-G5 and ß2m (Hunt et al., 2000b). This paper was not cited but is highly relevant to the studies in the Blaschitz paper because several of the antibodies used in their paper (Blaschitz et al., in press) require association with ß2m, as noted above. Several decades of experimentation have clearly shown that expression of the HLA genes and their proteins is cell type-specific, so the observation that HLA-G5 in recombinant proteins are complexed with ß2m (Morales et al., 2003) does not mean that the same is necessarily true of trophoblast or other types of cells. |
Copy number |
|---|
|
TopIntroductionCell preparationsUse of irrelevant cellsAntibody specificityReaching the maternal...Cell lineage and expression...Copy numberPCR experimentsReferences |
|---|
It would have been extremely helpful if the authors had controlled for copy number when comparing the results from transfected and normal cells. Otherwise, ‘negative’ results could simply be due to concentration differences. The authors conclusion based on the data presented in Figure 4 raises a further question. They concluded that because deglycosylation of basal plate and chorion membrane proteins resulted in ‘...a sharp single band corresponding to the HLA-G-G1 band of the .221-G1 transfectant...’ they must be identical proteins. No microsequencing or, in fact, any other type of data are offered to prove this point, which is essential to their conclusions. Ishitani et al. (2003)
concluded from the results of immunohistochemical experiments using o1G that HLA-G1 is comparatively scarce in the chorion, and results from our laboratory are the same. |
PCR experiments |
|---|
|
TopIntroductionCell preparationsUse of irrelevant cellsAntibody specificityReaching the maternal...Cell lineage and expression...Copy numberPCR experimentsReferences |
|---|
Finally, one must examine conclusions drawn from mRNA data. Other PCR assays that uniquely identify intron 4-containing mRNA encoding soluble HLA-G have been described (starting with Fujii et al., 1994
). For a convincing challenge to this and subsequent studies from other laboratories, the authors should employ precisely those primers and conditions in their experimental system and provide some explanation as to why they were not able to reproduce previous results. If they are able to reproduce previous work, they must then reconcile those findings with the conclusions in this study. Instead, Blaschitz et al. use a system for amplification that has the benefit of ease of performance but introduces the problem of competition for primers in the initial step. Unfortunately, neither the specificity nor the efficiency of the primers in their RT–PCR experiments is documented. Of equal importance, sequencing of the amplicons for authenticity is rarely mentioned. It is very easy to misamplify HLA genes because of the sequence similarity between them. For example, the pseudogene HLA-J is very similar in sequence to HLA-G; it is expressed as mRNA and contains intron 4 sequences. It can easily be mistaken for HLA-G5 or G6 if not sequenced. No one doubts that both HLA-G1 and HLA-G5 are appropriately glycosylated and inserted into the membrane or secreted (Bainbridge et al., 2000; Menier et al., 2000; Morales et al., 2003; Ulbrecht et al., 2004). However, two (Menier et al., 2000; Morales et al., 2003) of the four laboratories that have tested other isoforms agree that smaller isoforms are expressed, and one (Morales et al., 2003) has shown in situ expression of HLA-G2/G6 on migrating cytotrophoblast cells.
This point is deserving of further exploration given that pregnancy proceeds in women who cannot synthesize either HLA-G1 or HLA-G5 (Ober et al., 1998). These data can only be explained if (i) HLA-G is not necessary to pregnancy or (ii) the smaller isoforms are synthesized and compensate for lack of HLA-G1 and -G5. The latter explanation appears much more likely because individuals entirely lacking expression of the HLA-G gene are unknown.
In summary, the Blaschitz paper, although interesting because it challenges established thinking, fails to support its conclusions with solid experimental data. The authors can be credited with arriving at a novel idea, but, unfortunately, the experiments supporting this idea are unconvincing. It is under any circumstances difficult to use negative data (e.g. no PCR product, no signal in histochemical staining) to conclude that something is truly absent. When such a conclusion contradicts previous findings it is incumbent on the authors to address the previous work more directly. That includes in this case using the previously published conditions for histochemical staining, for RT–PCR (including the same primers) and for providing some explanation why their newly designed experiments are, as they conclude, more accurate when compared to results obtained duplicating previous work.
Having failed to perform appropriate studies that would lay a firm foundation for their conclusions, the authors leave open the possibility that the technical sophistication or scrutiny were lacking in the execution of experiments as an explanation for the negative results. Indeed, no doubt the authors were well intended, but technical flaws make it impossible to draw any meaningful conclusions from the work other than this.
|
References |
|---|
|
TopIntroductionCell preparationsUse of irrelevant cellsAntibody specificityReaching the maternal...Cell lineage and expression...Copy numberPCR experimentsReferences |
|---|
Bainbridge DR, Ellis SA and Sargent IL (2000) The short forms of HLA-G are unlikely to play a role in pregnancy because they are not expressed at the cell surface. J Reprod Immunol 47,1–16.[CrossRef][ISI][Medline]
Blaschitz A, Juch H, Volz A, Hutter H, Daxboeck C, Desoye G and Dohr G (in press) The soluble pool of HLA-G produced by human trophoblasts does not include detectable levels of the intron 4-containing HLA-G5 and HLA-G6. Mol Hum Reprod.
Fujii T, Ishitani A and Geraghty DE (1994) A soluble form of the HLA-G antigen is encoded by a messenger ribonucleic acid containing intron 4. J Immunol 153,5516–5524.[Abstract]
Hunt JS, Jedhav L, Chu W, Geraghty DE and Ober C (2000a) Soluble HLA-G circulates in mothers during pregnancy. Am J Obstet Gynecol 183,682–688.[CrossRef][ISI][Medline]
Hunt JS, Petroff MG, Morales P, Sedlmayr P, Geraghty DE and Ober C (2000b) HLA-G in reproduction: studies on the maternal–fetal interface. Hum Immunol 61,1113–1117.[CrossRef][ISI][Medline]
Ishitani A, Sageshima N, Lee N, Dorofeeva N, Hatake K, Marquardt H and Geraghty DE (2003) Protein expression and peptide binding suggest unique and interacting functional roles for HLA-E, F and G in maternal-placental immune recognition. J Immunol 171,1376–1384.
Lee N, Malacko AR, Ishitani A, Chen MC, Bajorath J, Marquardt H and Geraghty DE (1995) The membrane-bound and soluble forms of HLA-G bind identical sets of endogenous peptides but differ with respect to TAP association. Immunity 3,591–600.[CrossRef][ISI][Medline]
Menier C, Riteau B, Dausset J, Carosella ED and Rouas-Freiss N (2000) HLA-G truncated isoforms can substitute in fetal survival. Hum Immunol 61,1118–1125.[CrossRef][ISI][Medline]
Menier C, Saez B, Horejsi V, Martinozzi S, Krawice-Radanne I, Bruel S, Le Danff C, Reboul M, Hilgert I, Rabreau M et al. (2003) Characterization of monoclonal antibodies recognizing HLA-G or HLA-E: new tools to analyze the expression of nonclassical HLA class I molecules. Hum Immunol 64,315–326.[CrossRef][ISI][Medline]
Morales P, Pace JL, Platt JS, Phillips TA, Morgan K, Fazleabas AT and Hunt JS (2003) Placental cell expression of HLA-G2 isoforms is limited to the invasive trophoblast phenotype. J Immunol 171,6215–6224.
Ober C, Aldrich C, Rosinsky B, Robertson A, Walker MA, Willadsen D, Verp MS, Geraghty DE and Hunt JS (1998) HLA-G1 protein expression is not essential for fetal survival. Placenta 19,127–132.[CrossRef][ISI][Medline]
Pfeiffer KA, Rebmann V, Passler M, van der Ven K, van der Ven H, Krebs D and Grosse-Wilde H (2000) Soluble HLA levels in early pregnancy after in vitro fertilization. Hum Immunol 61,559–564.[CrossRef][ISI][Medline]
Polakova K, Kuba D and Russ G (2004) The 4H84 monoclonal antibody detecting beta2m free nonclassical HLA-G molecules also binds to free heavy chains of classical HLA class I antigens present on activated lymphocytes. Hum Immunol 65,157–162.[CrossRef][ISI][Medline]
Solier C, Aguerre-Girr M, Lenfant F, Campan A, Berrebi A, Rebmann V, Grosse-Wilde H and Le Bouteiller P (2002) Secretion of pro-apoptotic intron 4-retaining soluble HLA-G1 by human villous trophoblast. Eur J Immunol 32,3576–3586.[CrossRef][ISI][Medline]
Ulbrecht M, Maier S, Hofmeister V, Falk CS, Brooks AG, McMaster MT and Weiss EH (2004) Truncated HLA-G isoforms are retained in the endoplasmic reticulum and insufficiently provide HLA-E ligands. Hum Immunol 65,200–208.[CrossRef][ISI][Medline]
Yang Y, Yelavarthi KK, Chen H-L, Pace JL, Terranova PF and Hunt JS (1993) Molecular, biochemical and functional characteristics of tumor necrosis factor- produced by human placental cytotrophoblastic cells. J Immunol 150,5614–5624.[Abstract]
Yui J, Garcia-Lloret M, Wegmann TG and Guilbert LJ (1994) Cytotoxicity of tumour necrosis factor-alpha and gamma-interferon against primary human placental trophoblasts. Placenta 15,819–835.[ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. van der Meer, H.G.M. Lukassen, B. van Cranenbroek, E.H. Weiss, D.D.M. Braat, M.J. van Lierop, and I. Joosten Soluble HLA-G promotes Th1-type cytokine production by cytokine-activated uterine and peripheral natural killer cells Mol. Hum. Reprod., February 1, 2007; 13(2): 123 - 133. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||