Recombinant human zona pellucida proteins ZP1, ZP2 and ZP3 co-expressed in a human cell line

The zona pellucida (ZP) is the extracellular matrix that surrounds the oocyte. The mouse ZP is composed of three major glycoproteins designated as ZP1, ZP2 and ZP3 based on their average molecular weight of 185-200 kDa, 120-140 kDa and 83 kDa, respectively [1]. Similarly, the nomenclature of three human zona pellucida proteins was initially based on their electrophoretic mobility in sodium dodecylsulphate polyacrylamide gel electrophoresis (SDS-PAGE): ZP1 90-110 kDa, ZP2 65-85 kDa and ZP3 57-73 kDa [2].

The three major classes of human ZP genes are described as ZPA, ZPB and ZPC, on the basis of the size of the complementary DNAs (cDNAs), corresponding to mouse ZP2, ZP1 and ZP3 proteins respectively. The genes and corresponding zona glycoproteins are highly conserved in mammals [3]. Whereas the length and sequence identity of human and mouse glycoproteins from ZPA (ZP2) and ZPC (ZP3) (715 amino acids, 61 % identity and 424 amino acids, 67 % identity, respectively) are quite high, there is some uncertainty about human ZPB and its homology to the mouse ZP1. The mouse ZP1 gene described by Epifano and associates[4] has been initially accepted as a human ZPB homologue, whose genomic organisation was initially identified by Harris and associates [3]. Interestingly, identification of a ZPB ortholog in mice has never been reported.

Additional studies using antibodies developed against Bonnet Monkey recombinant ZP proteins expressed in Escherichia coli (E. coli), revealed that the protein of 90-110 kDa molecular weight initially classified as human ZP1 protein corresponding to mouse ZP1 is likely to be human ZP2 and the 65-85 kDa protein, is likely to be human ZP1. Human ZP1 migrates as two major bands on SDS PAGE, smaller 45-55 kDa and larger 58-63 kDa. The latter overlaps human ZP3 (58-68 kDa) [5].

Localised on chromosome 19, the mouse ZP1 gene is composed of 12 exons with an open reading frame of 1869 bp. The human ZPB gene, however, has an open reading frame of 1623 bp to encode a protein of 540 amino acids and molecular weight of approximately 60 kDa. It is evident that the human ZPB is shorter than mouse ZP1 (623 amino acids) and only 33 % identical. A gene potentially encoding a 638 amino acid human protein with more then 70 % homology to the mouse ZP1 has been described by Hughes and Barratt [6].

A structural model of the zona pellucida, based on the biochemical and ultrastructural analysis of the mouse zona pellucida, involves long filaments of a repeating ZP2 and ZP3 heterodimer. ZP1 provides structural integrity for the zona pellucida by cross-linking with disulphide bonds the ZP2/ZP3 filaments [1]. In view of the differences in protein composition observed between species, it is uncertain if the structure of the zona pellucida described in mice is applicable to those of other mammals.

Each secreted protein has a signal peptide that directs the protein to be secreted. A trans-membrane domain located on the carboxyl terminal side is followed by a basic amino acid cleavage recognition sequence for furin. There is evidence in mice and hamsters that prior to secretion nascent ZP proteins, at least ZP2 and ZP3, are proteolytically cleaved at their furin cleavage sites and subsequently assembled around the oocyte [7]. However, the furin proteolytic recognition sequence has been slightly altered in the human and the rabbit ZPB proteins.

In the mouse and a variety of other mammalian species including humans, ZP3 has been described as the primary sperm receptor, which subsequently induces the AR [8]. Sperm binding to the ZP and the AR on the ZP are important processes in human fertilisation. Tests for these functions provide powerful diagnostic and prognostic information for male fertility. These tests currently involve laborious assessment of sperm interaction with oocytes that have failed to fertilise in clinical IVF treatments [9]. In addition, the availability of native human ZP for diagnostic and research purposes is severely limited. The importance of using native human ZP in the study of sperm physiology, in particular with respect to the AR, has previously been emphasised [10]. Other approaches, such as stimulating the AR with chemicals, can be inaccurate or misleading [11]. Thus, a biologically active synthetic ZP, consisting of rhZP proteins, would be particularly useful as a reagent for investigating sperm-ZP interactions.

Although several research groups have reported production of rhZP proteins, their various biological activities have been considered insufficient for testing sperm function [12, 13]. It is most likely that these recombinant proteins were not glycosylated appropriately to interact fully with the human spermatozoa. In addition, human ZP2 and ZP1 may be a necessary contribution to primary sperm binding and induction of the AR.

In the present study, recombinant human ZP proteins, ZP1, ZP2 and ZP3 were expressed in the human embryonic kidney cell line 293T to allow a human pattern of glycosylation. In addition, rhZP2 and rhZP1 and rhZP3 were co-expressed to determine whether the combination of these proteins is required for biological activity. Recombinant human ZP proteins were compared to native ZP in terms of size and level of glycosylation as assessed by Western blotting and the ability to induce the AR. Immunofluorescence using anti-human rZP1, rZP2 and rZP3 sera raised in rabbits, was used to localise rhZP1, rhZP2 and rhZP3 proteins in transfected 293T cells and human oocytes. The effect of these antibodies on sperm-ZP binding was also tested.

The cDNA encoding full-length human ZP3, cloned into the p-Bluescript vector, was kindly provided by Professor Jurrien Dean (National Institutes of Health; Alcohol, drug Abuse and Mental Health Administration, Maryland, USA). The cDNA encoding full-length human ZPA (ZP2) and ZPB (ZP1) cloned into p-Bluescript vectors, were kindly provided by Dr. Jeffrey Harris (Zonagen, Texas, USA). Full-length coding sequences were excised from their respective pBluescript constructs using the restriction enzymes BamHI /XhoI (Roche Molecular Bioche-micals, Germany) for ZP3 and KpnI/NotI (Promega, WI, USA) for ZP1 and ZP2 and ligated into pcDNA 3.1/Hyg expression vector (Invitrogen, CA, USA).

Sperm binding to 293T cells was also assessed.

The significance of difference between native human ZP- induced and rhZP protein-induced AR was examined by t-test.

Western blotting analysis of the culture medium and cell lysate from each transfectant is shown in Figure 1. Both rhZP2 and rhZP3 were effectively secreted into the culture medium, whereas rhZP1 was detected in cell lysates only.

Probing the cell lysate from ZP1 transfectants with anti-rZP1 serum recognised a band of 70 kDa. A major band of 55 kDa and a weaker band of 70 kDa were detected in disaggregated human ZP probed with anti- human rZP1 serum. Western blotting with anti- rZP2 serum revealed a band of 110 kDa in cell culture medium and cell lysate from ZP2 transfected 293T cells. Occasionally a weaker band of 90 kDa was detected in cell culture medium, probably representing a precursor or less glycosylated rhZP2 protein. Molecular weights of rhZP2 protein, either secreted or detected in cell lysate, were slightly smaller than those detected in disaggregated human ZP probed with anti-rZP2 serum (120 kDa). Similarly, cell culture medium and cell lysate from ZP3 transfectants probed with anti-rZP3 serum revealed a band of 55 kDa, similar to the band observed in disaggregated human ZP. No reactivity was observed in the control culture media from cells expressing enhanced green fluorescent protein probed with any of the anti-rZP sera.

Recombinant human ZP1, ZP2 and ZP3 proteins were co-expressed in 293T cells. Western blotting analysis was performed on each culture medium and representative blots are shown. The molecular weight of the co-expressed proteins was compared with the molecular weight of the same protein individually expressed (Figure 2). Anti-rZP2 serum recognized a band of 110 kDa in culture medium of ZP2 transfectants and bands of 90 kDa and 110 kDa when rhZP2 was co-expressed with rhZP1 and rhZP3 (panel B). A band of 90 kDa was occasionally observed when rhZP2 was expressed as a single protein. Similarly, rhZP3, secreted into the culture medium from either ZP3 individually transfected cells or cells transfected with ZP1, ZP2 and ZP3, revealed a band of 55 kDa (panel C). However, rhZP1 was not detected in culture medium of individual ZP1 transfectants, but was present in cell lysates as a band of 55 kDa. In contrast, rhZP1 was detected in the culture medium when cells were co-transfected with ZP1, ZP2 and ZP3: there were bands of 85 kDa, 70 kDa and 55 kDa against anti-human rZP1 serum, with the latter similar to the major band observed in disaggregated ZP (panel A). This suggests that the combined expression of these three proteins releases rhZP1 from the cells. However, when secreted rhZP1 was immunoprecipitated with anti-rZP1 sera from the ZP1/ZP2/ZP3 transfectants culture media, neither ZP2 nor ZP3 could be detected in the immuno-complexes.

Culture medium from ZP1/ZP2/ZP3, ZP2 and ZP3 transfectants were treated with N-glycosidase to ascertain glycosylation efficiency of these proteins by 293T cells (Figure 3). Culture medium from ZP1/ZP2/ZP3 transfectants probed with rZP1 serum contained a band of 55 kDa before, and a band of 50 kDa after N-glycosidase treatment (panel A). This indicates that N-linked oligosaccharides were present in the secreted form of rhZP1. Removal of N-linked oligosaccharides from secreted rhZP2 reduced its molecular weight slightly, from 110 kDa to 100 kDa (panel B). A much larger reduction of molecular weight was observed when secreted rhZP3 was treated with N-glycosidase, reducing it from 55 kDa to approximately 40 kDa, which is a molecular weight similar to the mature hZP3 protein backbone (panel C).

In this study, the human embryonic kidney cell line 293T was employed to express rhZP1, rhZP2 and rhZP3 in an attempt to obtain bioactive proteins with a species-specific glycosylation pattern. The 293T cells have the ability to co-express several genes and produce highly glycosylated and biologically active proteins [15]. RhZP2 and rhZP3 were secreted into the culture medium, whereas rhZP1 was retained within the cell and only detected in the cell lysate. Interestingly, when the three proteins were co-expressed in the same cell, all three proteins were detected in the culture medium. To our knowledge, this is the first report that shows co-expression of all three rhZP proteins in the same cell. Deglycosylation of the rhZP proteins showed that they were heavily glycosylated by the 293T cells. Immunofluorescence studies of the protein-producing cells showed long string-like threads extending from the cells that stained with the corresponding serum. However, no evidence of any biological activity could be found with any of rhZP protein preparations whether expressed singly or in combination. Furthermore, direct exposure of sperm to 293T cells expressing rhZP proteins produced no evidence of sperm binding or augmentation of the AR.

The most frequently expressed rhZP protein is a homologue of the mouse primary sperm receptor, ZP3. It has been previously expressed in bacteria and also in mammalian cells [12, 13]. In the present study, similar to other studies, it was observed that rhZP3 was secreted into the culture medium. In addition, rhZP3 secreted into the medium and extracted from the cells after 1 % Triton X-100 treatment had a similar molecular weight to the ZP3 protein in the human disaggregated ZP (55 kDa). Glycosylation of the secreted rhZP3 was analysed with N-glycosidase. Western blotting analysis of N-glycosidase-treated ZP3 culture medium revealed a band of approximately 40 kDa. This is very similar to the expected molecular weight of the mature protein backbone comprised of 331 amino acids (43 kDa). This suggests that the majority of the oligosaccharides in the recombinant protein are N- linked to the protein backbone, accounting for as much as 28 % of the molecular weight of rhZP3 whereas O-linked oligosaccharides comprise only a small part of the glycoprotein. By producing the recombinant protein in a human cell line, this study avoided the possibility of an incompatible interaction between rhZP3 and the human sperm due to species-specific glycosylation, an important consideration in the study of the gamete interaction [16].

Although the mobility of rhZP3 on SDS-PAGE and the level of N-glycosylation indicate a high degree of similarity between rhZP3 and native ZP3, biological activity of the recombinant protein was not observed. Bioactivity of rhZP3 was assessed for both blocking sperm-ZP binding and stimulation of the AR. The densitometry measurements of bands representing rhZP3 and native ZP3 indicated that the concentration of rhZP3 in the concentrated culture medium was equivalent to a concentration of native ZP3 protein sufficient to show biological activity.

The ability of the recombinant proteins in the present study to induce the AR was always compared with that of native ZP for the same sperm sample. This comparison gives an indication of sperm ability to acrosome-react in situations with greater physiological similarity. Although several laboratories have claimed production of biologically active rhZP3, verification by different laboratories and parallel studies using native material is not reported [12, 17]. Additionally, in the reported studies the recombinant material was usually incubated under non-physiological conditions, introducing the strong possibility of an artificial result rather than a physiological effect. For instance, rhZP3 generated by CHO cells promotes calcium influx into human spermatozoa, tyrosine phosphorylation of membrane proteins and induction of the AR, but the activity is slow in onset and detectable only after prolonged incubation, a phenomenon anticipated with time [12, 18]. Recombinant human ZP3, translated in vitro or expressed in bacteria, is not glycosylated and its sperm binding and delayed acrosome reaction-inducing activity differs from native human ZP which readily binds sperm and produces rapid induction of the acrosome reaction. It appears that these rhZP3 proteins were not considered by any of these research groups to be sufficiently active to justify continued production. Acceptable recombinant ZP proteins would be those with an ability to induce the AR similar to that of native ZP. Dong and associates have reported production of rhZP3 in human ovarian teratocarcinoma (PA-1) cells which shows the full spectrum of biological activity, including competitive inhibition of sperm-zona pellucida binding and stimulation of the AR [13]. The chosen cell line assures both a species- and tissue-specific glycosylation pattern. If tissue-specific glycosylation of rhZP3 in concert with species-specificity is a crucial factor for successful interaction with human sperm, then rhZP3 produced in the present study may not be appropriately glycosylated and hence not biologically active.

In the present study the full-length coding sequences were used for each of the human ZP protein: ZP1: 1620bp; 540 amino acids, ZP2: 2235bp; 745 amino acids and ZP3: 1272; 424 amino acids. A convenient purification tag, such as a hexahistidine tail, was not added to the coding sequences. This omission prevented a convenient and efficient protein purification. Although there are some suggestions that various amounts of unwanted proteins and protein fragments may have interfered with the biological activity of the recombinant proteins, this is unlikely since no effect on sperm motility was observed. This design was based on the possibility that interference with the primary structure affects protein processing, changes the protein folding and glycosylation pattern and consequently affects protein function. It has also been suggested that the protein activity is highly dependent upon the native conformation, due to either the importance of conformational epitopes or the greater stability of folded proteins [3, 12]. Therefore, the signal, trans-membrane, cytoplasmic and extracellular protein domains were retained in order to process the expressed proteins in a manner as similar as possible to their native structure. The potential signal sequences in human ZP1, ZP2 and ZP3 proteins comprise the first 21, 38 and 22 amino acids, respectively. Prior to secretion, the signal sequence is removed from the precursor ZP protein by specific signal peptidases in the endoplasmic reticulum (ER). A C-terminal peptide of the immature protein is also cleaved at the consensus furin cleavage site (CFCS), which plays an essential role in regulating protein secretion and activation [19]. Protein maturation and removal of the signal and C-terminal peptides after the CFCS result in mature ZP proteins considerably smaller than that predicted from their exon sequences: ZP1, 446AA (~60 kDa); ZP2, 605AA (~82 kDa) and ZP3, 331AA (~43 kDa) [3, 7]. Since both secreted rhZP3 and rhZP3 detected in the cell lysate have similar molecular weights, it is expected that the latter has probably undergone the same changes and it is retained somewhere in the cytoplasm or the cell plasma membrane as part of an ongoing process of secretion. This study is the first to show production of rhZP2 in a human cell line. The rhZP2 is detected in both the culture medium and the cell lysate of ZP2 transfectants as a band of 110 kDa, which is slightly smaller than the band in native human ZP (120 kDa). When the culture medium of ZP2 transfectants was treated with N-glycosidase, the relative molecular weight observed on SDS PAGE was reduced to 100 kDa, which is noticeably higher than the expected molecular weight of the protein backbone after removal of the signal and C-terminal peptides (~82 kDa). One may alternatively speculate that maturation of the protein did not occur as expected, although this is unlikely since the protein is efficiently secreted and accurate elimination of the signal and C-terminal sequences is important for protein secretion. It is also possible that some oligosaccharides are O-linked to the protein backbone and are not removed by N-glycosidase. Theoretically, 25 % of the molecular weight of mature ZP2 is from the sugar components at six potential N-glycosylation sites. The change following treatment with N-glycosidase is only 9 %. A most likely explanation is that glycosylation of the recombinant protein was either incomplete or different from that in native ZP2.

In contrast to rhZP2 and rhZP3, rhZP1 was not secreted into the medium, but retained within the cells. Western blotting analysis of the cell lysate probed with anti-rZP1 serum revealed a band of 70 kDa, similar to a weak band observed in native disaggregated ZP. The 70 kDa form may be an immature rhZP1 protein in which a C-terminal peptide has not been cleaved. Furin prefers the site R-X-K/R-R. These sequences are in rhZP2 and rhZP3. However, rhZP1 contains such an altered cleavage site, S-R-R-R, that may also be recognised by some furin-like enzymes. There is a possibility that furin in 293T cell does not recognise an alternative sequence in rhZP1, which then becomes resistant to the enzyme. Alternatively, the enzyme active sites may be shielded due to the limited flexibility of the protein and therefore inaccessible to the enzyme. All ZP1 family proteins, including human ZP1, contain a cysteine-rich region called the trefoil domain. It has been demonstrated that many trefoil domain-containing proteins demonstrate high resistance to enzymatic degradation which is associated with the domain's extensive crosslinking and short secondary structure. Similarly, uncleaved recombinant mouse ZP3 accumulates in the ER of mutated embryonal carcinoma cells and 293T cells that do not posses CFCS [7]. Although the recently reported study by Zhao and associates suggests that cleavage at the furin site is not required for formation of the mouse ZP, the majority of the research groups support an essential role of furin in regulating protein secretion and formation of the human ZP [20]. It is of interest that co-expression of the three proteins results in secretion of rhZP1 from the cell. The molecular weight of the secreted rhZP1 detected in the medium (55 kDa) is close to the expected size (~60 kDa) of the mature ZP1 protein backbone with signal and C-terminal sequences cleaved from the protein precursor. However, glycosylation experiments show that rhZP1 detected in the culture medium of ZP1/ZP2/ZP3 transfectants contains some N-linked oligosaccharides. Full-length rhZP1 contains seven potential N-linked glycosylation sites (N-X-S/T). The molecular weight of the secreted protein after N-glycosidase treatment was reduced to 50 kDa. This molecular weight is smaller than expected for the protein backbone of the mature ZP1 (~60 kDa). It is possible that the recombinant protein undergoes additional post-translational modifications as part of the cosecretion with rhZP2 and rhZP3, the nature of which are unknown. However, when rhZP1 was immunoprecipitated with anti-rZP1 serum from the culture medium of rhZP1/ZP2/ZP3 transfectants neither rhZP2 nor rhZP3 were detected in the immunoprecipitate. At this stage it can be speculated that co-expression of all three rhZP proteins facilitates secretion of the rhZP1 in a manner that needs to be further examined.

Co-expression of all three proteins does not change mobility on SDS-PAGE of rhZP2 and rhZP3. Both individually and co-expressed rhZP2 and rhZP3 proteins are secreted into the medium and retain the same molecular weights. However, lower concentration of each protein is detected in the medium in the co-transfection experiments. A weak band of 70 kDa which reacted with anti-human rZP1 serum was also detected in the culture medium of ZP1/ZP2/ZP3 transfectants and is, most likely, an immature protein released from a small number of cells with disturbed membranes rather than a secreted form.

Apart from the necessity for a species and/or tissue specific glycosylation pattern and purity of the recombinant proteins, a tertiary structure of ZP components may also be crucial for biological activity. This may contribute to the relatively low level of sperm binding and acrosome reaction inducing activity of rhZP3 reported by the majority of research groups. Human ZP2 and ZP1 may contribute to primary sperm binding and inducing the AR. To test these alternative explanations, the biological activity of the culture media from ZP1/ZP2/ZP3 transfectants was examined. No augmentation of the AR was observed when sperm were incubated with concentrated culture medium from co-expressed rhZP proteins. One can argue that the interaction between recombinant proteins in the medium and free-swimming sperm is too weak to maintain sufficiently strong binding, given that physiologically this interaction is achieved after binding has been initiated in the case of native ZP. Sperm were incubated with transfected 293T cells in order to test the ability of sperm to bind ZP protein(s) when they are expressed on the cell surface. No interaction was observed.

Although no evidence of biological activity was found with any of the rhZP proteins, whether expressed singly or in combination, some interesting findings have been observed. When sperm were incubated with the culture medium in which FuGENE 6/DNA complex was present, significant impairment of sperm motility was observed after two hours of culture. The FuGENE 6/DNA complex did not, however, affect 293T cells in culture for 72 hours. No notable difference in cell appearance or amount of protein produced (tested by Western blotting) was observed when FuGENE 6 was left in the culture medium. The detrimental effect on the sperm from FuGENE 6 resulted in a high proportion of acrosome-reacted sperm identified by PSA staining. This further emphasises the importance of testing recombinant protein under appropriately controlled conditions.

Immunofluorescence studies of protein-producing 293T cells show strong staining around the cell, but also occasionally on long string-like threads extending from the cells. The latter suggests some polymer formation after protein expression from the cell. Although it is expected, from the evidence discussed earlier, that rhZP1 is most likely retained within the ER compared with rhZP2 and rhZP3 which are located near the plasma membrane prior to secretion, there is no noticeable difference in the observed immunofluorescence pattern between these transfectants. Further experiments need to be performed to determine the specific intracellular location. Tests on human oocytes confirmed that all immune sera reacted with human ZP but not other component of the oocyte.

In conclusion, rhZP1, rhZP2 and rhZP3 were successfully expressed in the human embryonic kidney cell line 293T. Efficient secretion of rhZP2 and rhZP3 into the culture medium was observed, whereas rhZP1 was only secreted when co-expressed with rhZP2 and rhZP3. It appears that an interaction amongst these proteins may be required for release of rhZP1 from the cell. This is the first study that shows co-expression of all three known ZP proteins by the same cell. Although this approach is not satisfactory for producing active human ZP proteins, it makes a significant contribution to understanding structural and functional characteristics of the ZP proteins. Further examinations are required to produce a recombinant material which exhibits the activity essential for use as an alternative to the human oocyte in the study of the human acrosome reaction.

We thank Professor Jurrien Dean for providing human ZP3 cDNA and Dr. Jeffrey Harris for providing human ZP cDNAs and rabbit sera. We thank Ms Mingli Liu for technical assistance, the scientists in both Royal Women's Hospital and Melbourne IVF Laboratories for collecting the oocytes and sperm.