Avian influenza A virus adaptation to the equine host and identification of host-specific markers.

Avian influenza A viruses (IAVs) are able to overcome the interspecies barrier and adapt to the new non-avian host. The process of adaptation requires the adaptive changes of IAV genome resulting in amino acid substitutions. The aim of this work was the description of amino acid substitutions in avian influenza A viruses (IAVs) occurring during their adaptation to equine host. Today, viruses of the equine influenza H3N8 subtype, first isolated in 1963, represent a single genetic lineage of IAV causing a respiratory disease in horses. We compared the amino acid sequences of the conserved proteins PB2, PB1, PA, NP, M1, M2, NS1 and NEP of equine influenza H3N8 subtype IAV with sequences of avian viruses, both available in the NCBI's Influenza Virus Resource Database. The amino acid substitutions persisting in equine IAV isolates and occurring in avian IAV at f both hosts.

Prediction of additional lymph node involvement in breast cancer patients with positive sentinel lymph nodes.

Axillary lymph node dissection (ALND) has traditionally been the principal method for evaluating axillary lymph node status in breast cancer patients. In the past decades sentinel lymph nodes biopsy after lymphatic mapping has been used to stage the disease. The majority of sentinel lymph nodes (SLN) positive patients do not have additional metastases in non-sentinel nodes (non-SLN) after additional ALND. These patients are exposed to the morbidity of ALND without any benefit from additional axillary clearence. In the present study we would like to asses the criteria for selecting those patients, who have high risk for non-SLN metastases in the axilla in cases of positive SLN. In this retrospective analysis, clinical and pathologic data from 163 patients who underwent SLN biopsy followed by ALND were collected. Following clinical and pathological characteristics were analyzed to predict the likehood of non-SLN metastases: age, staging, histologic type and grading of the tumors, hormonal receptor status, HER-2 receptor status and Ki-67 protein, angioinvasion, metastases in SLN and non-SLN. Relative frequencies of individual characteristics between sample groups were statistically tested by Chi-square test at significance level p=0.5, when sample sizes in groups were small (≤5) by Fisher´s exact test. Metastasis in SLN were present in 67 (41%) of patients, 48 patients (29,4%) had metastasis also in non-SLN. The ratio between non-SLN positive / non-SLN negative lymph nodes in patients with positive SLN increases with the stage of the disease, the difference between values for the pT1c and pT2 stadium was statistically significant (p = 0.0296). The same applies to grading, but the differences were not significant (p>0.05). We could not find significant differences for angioinvasion of the tumor, probably for small number of patients with angioinvasion (p>0.05).Only the stage of the tumor was shown to be significant in predicting the metastasis in non-SLN in our group of breast cancer patients with positive SLN Nearly 80% of the patients of 70 years and older displayed no benefit from axillary staging, because of negative SLN as well as non-SLN, although thanks to the small sample size this was not a statistically significant result. Furthermore, current recommendations for axillary staging in breast cancer patients are discussed.

Immune response of mice to non-adapted avian influenza A virus.

Human infections with avian influenza A viruses (IAVs) without or with clinical symptoms of disease were recently reported from several continents, mainly in high risk groups of people, who came into the contact with infected domestic birds or poultry. It was shown that avian IAVs are able to infect humans directly without previous adaptation, however, their ability to replicate and to cause a disease in this new host can differ. No spread of these avian IAVs among humans has been documented until now, except for one case described in Netherlands in the February of 2003 in people directly involved in handling IAV (H7N7)-infected poultry. The aim of our work was to examine whether a low pathogenic avian IAV can induce a virus-specific immune response of biological relevancy, in spite of its restricted replication in mammals. As a model we used a low pathogenic virus A/Duck/Czechoslovakia/1956 (H4N6) (A/Duck), which replicated well in MDCK cells and produced plaques on cell monolayers, but was unable to replicate productively in mouse lungs. We examined how the immune system of mice responds to the intranasal application of this non-adapted avian virus. Though we did not prove the infectious virus in lungs of mice following A/Duck application even after its multiple passaging in mice, we detected virus-specific vRNA till day 8 post infection. Moreover, we detected virus-specific mRNA and de novo synthesized viral nucleoprotein (NP) and membrane protein (M1) in lungs of mice on day 2 and 4 after exposure to A/Duck. Virus-specific antibodies in sera of these mice were detectable by ELISA already after a single intranasal dose of A/Duck virus. Not only antibodies specific to the surface glycoprotein hemagglutinin (HA) were induced, but also antibodies specific to the NP and M1 of IAV were detected by Western blot and their titers increased after the second exposure of mice to this virus. Importantly, antibodies neutralizing virus A/Duck were proved in mouse immune sera after the second dose of virus and a slight increase of mRNA expression of immune mediators tumor necrosis factor alpha (TNF-α) and IP10 has been observed in lungs of these mice 48 hr after the infection. These observations correspond to the limited replication ability of the virus in mice and provided an important information about its ability to induce virus-specific antibodies, including those neutralizing virus, even without the previous virus adaptation to the new mammalian host. Such antibodies could consequently influence the immune potential of exposed individuals and their defensive capability against the newly emerged, even more virulent IAV.

Virus-neutralizing antibody response of mice to consecutive infection with human and avian influenza A viruses.

In this work we simulated in a mouse model a naturally occurring situation of humans, who overcame an infection with epidemic strains of influenza A, and were subsequently exposed to avian influenza A viruses (IAV). The antibody response to avian IAV in mice previously infected with human IAV was analyzed. We used two avian IAV (A/Duck/Czechoslovakia/1956 (H4N6) and the attenuated virus rA/Viet Nam/1203-2004 (H5N1)) as well as two human IAV isolates (virus A/Mississippi/1/1985 (H3N2) of medium virulence and A/Puerto Rico/8/1934 (H1N1) of high virulence). Two repeated doses of IAV of H4 or of H5 virus elicited virus-specific neutralizing antibodies in mice. Exposure of animals previously infected with human IAV (of H3 or H1 subtype) to IAV of H4 subtype led to the production of antibodies neutralizing H4 virus in a level comparable with the level of antibodies against the human IAV used for primary infection. In contrast, no measurable levels of virus-neutralizing (VN) antibodies specific to H5 virus were detected in mice infected with H5 virus following a previous infection with human IAV. In both cases the secondary infection with avian IAV led to a significant increase of the titer of VN antibodies specific to the corresponding human virus used for primary infection. Moreover, cross-reactive HA2-specific antibodies were also induced by sequential infection. By virtue of these results we suggest that the differences in the ability of avian IAV to induce specific antibodies inhibiting virus replication after previous infection of mice with human viruses can have an impact on the interspecies transmission and spread of avian IAV in the human population.

HA2 glycopolypeptide of influenza A virus and antiviral immunity.

Influenza A viruses (IAVs) cause acute respiratory infections in humans against which an effective prevention has not yet been developed due to their high variability and broad host specificity. The permanent threat of arising new influenza pandemic is represented by avian viruses which after their interspecies transmission can cause a disease with a devastating impact on humans lacking the specific immunity. Since the current vaccines inducing virus-neutralizing (VN) antibodies are targeted at a variable globular part of hemagglutinin (HA), their efficacy is limited and they need permanent updating. On the other hand, conserved IAV antigens such as proton channel M2, membrane protein M1 or nucleoprotein (NP) do not induce VN antibodies, but they do induce heterosubtypic protection resulting in the reduction of virus replication and an improved recovery from the disease. From this point of view recent attention has also been focused on the conserved part of HA, its HA2 glycoprotein (HA2). The main aspects revealing a contribution of HA2 gp to protective immunity are discussed in this review.

Two distinct regions of HA2 glycopolypeptide of influenza virus hemagglutinin elicit cross-protective immunity against influenza.

UNLABELLED: Currently, a new trend in development of vaccines against influenza with broader spectrum of efficacy is focused on conserved antigens of influenza virus. The HA2 glycopolypeptide (HA2 gp) is one of conserved antigens, potentially suitable as immunogens inducing cross-protection against influenza. We selected two distinct domains of HA2 gp originating from influenza A virus (IAV) of H3 subtype for induction of antiviral immune response: the ectodomain (EHA2) comprising aa 23-185 and the fusion peptide (FP) comprising N-terminal aa 1-38. BALB/c mice were immunized with three doses of EHA2 and FP, respectively, and subsequently challenged with 2 LD50 of IAV of homologous (H3) or heterologous (H7) HA subtype. Both peptides induced significant antibody response and protected mice against the lethal infection. The most efficient protection was achieved with EHA2 against homologous virus. KEYWORDS: influenza A virus; cross-protection; HA2 glycopolypeptide; HA2 ectodomain; fusion peptide; mice; vaccine.

Evaluation of anti-influenza efficiency of polyclonal IgG antibodies specific to the ectodomain of M2 protein of influenza A virus by passive immunization of mice.

We attempted to quantify the protective potential of polyclonal IgG antibodies specific to the ectodomain of M2 protein (eM2) of influenza A virus (IAV) against lethal influenza infection of mice. For this purpose, eM2 conjugated with keyhole limpet hemocyanin (KLH) or KLH alone were administered with Freund's adjuvant intraperitoneally (i.p.) to BALB/c mice. IgG antibodies specific to the KLH-eM2 conjugate (anti-KLH-eM2 IgGs) and KLH (anti-KLH IgGs), respectively, were purified from ascitic fluids. Analysis of the preparation of anti-KLH-eM2 IgGs by ELISA revealed that it contained about 25% of anti-eM2 IgGs and 75% of anti-KLH IgGs. Taking into account this finding mice were passively immunized by intravenous route with 320, 160, 80, and 40 µg of anti-eM2 IgGs per mouse, respectively, while 320 µg of anti-KLH IgGs were used in control. Following subsequent infection with 3 LD50 IAV the survival of mice was determined. An absolute protection (100% survival) was obtained with 320 µg of anti-eM2 IgGs, and a relatively strong significant protection (~80% survival, p = 0.024) with 160 µg. The amount 160 µg of IgGs represents approx. 100 µg IgGs per 1 ml of blood.

Heterosubtypic protective immunity against influenza A virus induced by fusion peptide of the hemagglutinin in comparison to ectodomain of M2 protein.

Several types of influenza vaccines are available, but due to the highly unpredictable variability of influenza virus surface antigens (hemagglutinin (HA) and neuraminidase) current vaccines are not sufficiently effective against broad spectrum of the influenza viruses. An innovative approach to extend the vaccine efficacy is based on the selection of conserved influenza proteins with a potential to induce inter-subtype protection against the influenza A viruses. A promising new candidate for the preparation of broadly protective vaccine may be a highly conserved N-terminal part of HA2 glycopolypeptide (HA2 gp) called fusion peptide. To study its capacity to induce a protective immune response, we immunized mice with the fusion peptide (aa 1-38 of HA2 gp). The protective ability of fusion peptide was compared with the ectodomain aa 2-23 of M2 protein (eM2) that is antigenically conserved and its immunogenic properties have already been well documented. Corresponding peptides (both derived from A/Mississippi/1/85 (H3N2) virus) were synthesized and conjugated to the keyhole limpet hemocyanin (KLH) and used for the immunization of mice. Both antigens induced a significant level of specific antibodies. Immunized mice were challenged with the lethal dose of homologous (H3N2) or heterologous A/PR/8/34 (H1N1) influenza A viruses. Immunization with the fusion peptide led to the 100% survival of mice infected with 1 LD50 of homologous as well as heterologous virus. Survival rate decreased when infectious dose was raised to 2 LD50. The immunization with eM2 induced effective cross-protection of mice infected even with 3 LD50 of both challenge viruses. The lower, but still effective protection induced by the fusion peptide of HA2 gp suggested that besides ectodomain of M2, fusion peptide could also be considered as a part of cross-protective influenza vaccine. To our knowledge, this is the first report demonstrating that active immunization with the conjugated fusion peptide of HA2 gp provided the effective production of antibodies, what contributed to the cross-protection against influenza infection.

Broadly cross-reactive monoclonal antibodies against HA2 glycopeptide of Influenza A virus hemagglutinin of H3 subtype reduce replication of influenza A viruses of human and avian origin.

The reactivity of monoclonal antibodies (MAbs) prepared to the HA2 glycopeptide (gp) of A/Dunedin/4/73 (H3N2) hemagglutinin was tested against influenza A viruses of H3, H4, and H7 subtypes. Only one (CF2) out of six MAbs reacted with influenza A viruses of all three subtypes (H3, H4 and H7). The inter-subtype reactivity of this MAb (CF2) is in accord with the highly conservative sequence in the previously defined MAb-binding site I, i.e. the aa 1-38 of N-terminus of HA2 gp. MAb CF2 as well as inter-subtype cross-reactive MAb IIF4, recognizing the binding site II of HA2 gp, were tested for their effect on replication of influenza A viruses. Both these MAbs reduced the number of plaques of viruses of homologous (H3) as well as heterologous (H4) virus subtypes, the latter less efficiently. The potential of these MAbs to influence in vivo replication of influenza A viruses of various subtypes is discussed.

Multiorgan distribution of human influenza A virus strains observed in a mouse model.

Multiorgan spread and pathogenesis of influenza infection with three human influenza A viruses was studied in mice. Mouse-adapted viruses A/Dunedin/4/73(H3N2), A/Mississippi/1/85(H3N2), and A/PR/8/34(H1N1) differed considerably in virulence (p.f.u./LD(50)): 79,000 p.f.u. for Dunedin, 5,000 p.f.u. for Mississippi, and 65 p.f.u. for PR/8, which qualified Dunedin as low virulent, Mississippi as intermediate, and PR/8 as highly virulent. All three viruses were detected in lungs, heart, and thymus by cultivation and RT-PCR. Moreover, vRNA of all viruses was found in liver and spleen, of Dunedin and PR/8 also in kidneys and that of Dunedin and Mississippi in blood. Only vRNA of Dunedin was demonstrated in brain. Lung damage accompanied by histopathological changes and thymus reduction were most extensive after infection with the highly virulent virus PR/8. We assume that the ability to spread to multiple organs may be a more common property of influenza viruses in mammalian hosts than previously believed.

Antibodies induced by the HA2 glycopolypeptide of influenza virus haemagglutinin improve recovery from influenza A virus infection.

The haemagglutinin (HA) of influenza A virus consists of two glycopolypeptides designated HA1 and HA2. Antibodies recognizing HA1 inhibit virus haemagglutination, neutralize virus infectivity and provide good protection against infection, but do not cross-react with the HA of other subtypes. Little is known regarding the biological activities of antibodies against HA2. To study the role of antibodies directed against HA2 during influenza virus infection, two vaccinia virus recombinants (rVVs) were used expressing chimeric molecules of HA, in which HA1 and HA2 were derived from different HA subtypes. The KG-11 recombinant expressed HA1 from A/PR/8/34 (H1N1) virus and HA2 from A/NT/60 (H3N2) virus, whilst KG-12 recombinant expressed HA1 from A/NT/60 virus and HA2 from A/PR/8/34 virus. Immunization of BALB/c mice with rVV expressing HA2 of the HA subtype homologous to the challenge virus [A/PR/8/34 (H1N1) or A/Mississippi/1/85 (H3N2)] did not prevent virus infection, but nevertheless resulted in an increase in mice survival and faster elimination of virus from the lungs. Passive immunization with antibodies purified from mice immunized with rVVs confirmed that antibodies against HA2 were responsible for the described effect on virus infection. Based on the facts that HA2 is a rather conserved part of the HA and that antibodies against HA2, as shown here, may moderate virus infection, future vaccine design should deal with the problem of how to increase the HA2 antibody response.

Antibodies specific to the HA2 glycopolypeptide of influenza A virus haemagglutinin with fusion-inhibition activity contribute to the protection of mice against lethal infection.

Four monoclonal antibodies (mAbs) recognizing distinct antigenic sites on the HA2 glycopolypeptide of influenza virus A/Dunedin/4/73 (H3N2) have been tested for in vivo protection. When applied intravenously before infection, three of them increased the survival of BALB/c mice infected with 1 LD50 homologous virus. The protection resulted simultaneously in 2 days earlier clearance of virus from the lungs. These three antibodies inhibited the fusion activity of virus in previous in vitro experiments. One of them, specific to N-terminal aa 1-38 of the HA2 glycopolypeptide, was also tested for protection against the heterologous virus A/Mississippi/1/85 (H3N2). Protection similar to that against the homologous virus was observed. The fourth mAb, without fusion-inhibition activity, did not protect mice. It is concluded that antibodies specific to the antigenically conserved HA2 glycopolypeptide that exhibit fusion-inhibition activity can contribute to the protection of infected mice and mediate more effective recovery from infection.

Evaluation of clinical specimens for influenza A virus positivity using various diagnostic methods.

The diagnostic method for Influenza A virus, utilizing the SERION ELISA Antigen kit (SERION EIA), if results were evaluated according to the manufacturer's instructions, has repeatedly failed to detect a great number of clinical samples positive by virus isolation and RT-PCR. Therefore we compared the SERION EIA with the one-step 44/107L-Px immunocapture enzyme immunoassay (44/107L-Px EIA), developed in our laboratory (Tkácová and Varecková, J. Virol. Methods 60, 65-71, 1996). Seventy-three clinical specimens, of which 65 were positive by virus isolation (used as reference method), were tested by both EIAs. By the SERION EIA, out of the 65 reference-positive samples only 8 (12%) were positive, 5 (8%) were ambiguous, and 52 (80%) were negative, which corresponded to the sensitivity of 12%. On the contrary, the sensitivity of the 44/107L-Px EIA was 74%. However, the calculation of cut-off values for the evaluation of positivity of clinical specimens in these two assays were not the same. If the evaluation procedure used for the 44/107L-Px EIA was applied to the SERION EIA, the sensitivity and the specificity of both EIAs became comparable, namely 71% and 100% for the SERION EIA and 74% and 100% for the 44/107L-Px EIA, respectively. From these results it follows that not the detection ability of the SERION EIA, but the evaluation procedure recommended by its manufacturer led to a loss of large number of positive specimens.

Induction of tumor necrosis factor alpha in murine macrophages with various strains of Coxiella burnetii and their lipopolysaccharides.

The ability of various strains of Coxiella burnetii (C.b.) and their phase I and II lipopolysaccharides (LPSs) to induce tumor necrosis factor alpha (TNF-alpha) in peritoneal Balb/c mouse macrophages in vitro was investigated. Considerable differences in the induction ability were observed in dependence on the strain applied. In a TNF-alpha bioassay, the most effective inducers were both corpuscles and LPSs of the strains Priscilla and Scurry, followed by Nine Mile, Luga, and Henzerling I. In contrast, in ELISA, the most effective inducers were LPSs of the strains Luga and Henzerling, followed by Nine Mile, Priscilla, and Scurry. The role of toll-like receptor 4 (TLR4) in the induction was confirmed by the use of C3H/HeJ mouse macrophages. Thus, the induction of TNF-alpha was much higher in Balb/c mouse macrophages than that in TLR4-deficient C3H/HeJ mouse macrophages. Differences in the results of the bioassay and those of ELISA suggest a role of another secreted factor(s) induced with C.b. in murine macrophages that could act synergically with TNF-alpha in L929 cells in the bioassay. The observed differences in TNF-alpha induction might play a role in the pathobiology of Q fever.

Differences in antibody responses of mice to intranasal or intraperitoneal immunization with influenza A virus and vaccination with subunit influenza vaccine.

Two antigenically related but different influenza A virus strains of H3N2 subtype, A/Dunedin/ 4/73 (H3N2) (Dunedin) and A/Mississippi/1/85 (H3N2) (Mississippi), were used for intranasal (i.n.) and intraperitoneal (i.p.) immunization of mice and respective antibody responses were compared. In ELISA, using purified influenza A virus as antigen, the highest titer of antiviral antibodies was observed after a repeated i.n. infection, in which the Dunedin strain was followed by the Mississippi strain and vice versa. Similarly, in virus neutralization (VN) test, the highest titer of VN antibodies was found after a repeated i.n. infection. The subunit vaccine INFLUVAC, when administered intramuscularly (i.m.), induced only a poor antibody response as assayed by ELISA. Moreover, the INFLUVAC vaccination elicited a 100-fold lower titer of VN antibodies than the i.n. infection and an approx a 10-fold lower titer than the i.p. immunization. A repeated INFLUVAC vaccination did not lead to a significant increase of VN antibody titer. Also the antibody response to HA2gp--a conserved part of influenza hemagglutinin (HA) that might contribute to the induction of specific antiviral antibodies--was followed. Similarly to the VN antibody response, the highest HA2 antibody titer was induced after a repeated i.n. infection, whereas the lowest HA2 antibody titer was observed after a single or repeated INFLUVAC vaccination. Overall, the HA2 antibody titers remarkably well corresponded to the VN potential of the examined sera.

Analysis of specificity of interaction between synthetic key saccharide components of lipopolysaccharides and monoclonal antibodies to Coxiella burnetii.

Two monoclonal antibodies (MAbs) against the lipopolysaccharides (LPSs) of Coxiella burnetii (C.b.) strains Priscilla and Nine Mile were prepared characterized by their interaction with synthetic glycoconjugates representing parts of LPSs of C.b. in virulent phase. Both MAbs were directed against immunodominant epitopes comprising core constituent of LPSs, Kdo (3-deoxy-alpha-D-manno-2-octulo-pyranosylonic acid). ELISA showed that the anti-Nine Mile MAb 4/11 bound preferably to disaccharides (alpha-Kdo (2 --> 4) alpha-Kdo and alpha-Kdo (2 --> 4) alpha-(5d) Kdo), while the anti-Priscilla MAb 1/4/H bound to all conjugates, though with various intensity. On the other hand, immunoelectron microscopy revealed a positive binding of only one glycoconjugate, namely the trisaccharide alpha-Kdo (2 --> 4) alpha-Kdo (2 --> 4) alpha-Kdo-BSA, to both MAbs. In competitive ELISA (cELISA), the anti-Priscilla MAb 1/4/H distinguished the strains Nine Mile and Priscilla, while the anti Nine Mile MAb 4/11 did not.

Semi-quantitative RT-PCR assessment of molecular markers in breast large-core needle biopsies.

Large-core needle biopsies are frequently used for the preoperative evaluation of the breast lesions. In addition to initial diagnostic information, they can show the status of molecular markers with predictive and prognostic value and further contribute to an optimal selection of treatment strategy. So far, the potential use of large-core needle biopsies in assessment of marker profile of the breast lesions was studied using the immunostaining approaches. In this work, we sought to determine whether analysis of the large-core needle biopsy by semi-quantitative reverse-transcription polymerase chain reaction reveals molecular data that correspond with the marker status of the subsequently removed tumor. Five molecular markers including ER, PR, c-erbB-2/HER-2/neu, c-erbB-3/HER-3, and CD44 were assessed on mRNA isolated from 23 large-core needle biopsies and corresponding surgical breast cancer specimens using beta2- microglobulin as an internal standard. Significant or highly significant correlation between core biopsies and excised tumors was observed for each marker when the mRNA expression status was scored as positive or negative, with the concordance of the data ranging from 73.9% to 86%. Using the dichotomous scoring, majority of the biopsies (75%) displayed molecular profile that was either identical to the profile of the related tumor specimen, or with the difference in one marker. However, no significant correlation was found when the levels of the markers were expressed as continuous variables, possibly due to intratumoral cell heterogeneity. These results suggest potential usefulness and reliability of semi-quantitative RT PCR in the evaluation of large-core needle biopsies with regard to marker positivity or negativity. On the other hand, the marker-related data expressed as continuous variables cannot be accurately assessed on the large- core needle specimens using this approach, indicating the need for methodological improvements.

Inhibition of fusion activity of influenza A haemagglutinin mediated by HA2-specific monoclonal antibodies.

The effect of seven monoclonal antibodies (MAbs) specific to the light chain (HA2) of influenza A haemagglutinin (HA) on its fusion activity was investigated. These MAbs, which are non-virus neutralizing, defined four distinct antigenic sites on HA2 glycopolypeptide and the corresponding epitopes were attributed to the sequence stretches on HA2. The accessibility of all seven HA2 epitopes significantly increased after trypsin cleavage and pH 5 treatment of the HA (X-31). The influence of anti-HA2 MAbs on the fusion process was followed by cell-cell fusion of CHO cells expressing precursor HA, virus-liposome fusion assay, and haemolysis mediated by virus. MAb CF2, which bound directly to the fusion peptide 1-35 of HA2, was positive in all three fusion-inhibition assays and was the only one inhibiting the polykaryon formation of CHO-X-31 cells. Two other MAbs belonging to the same antigenic site but not binding directly to the fusion peptide inhibited virus to liposome fusion (EB12) or inhibited haemolysis (BB8). Moreover, MAb IIF4 binding to distinct antigenic site within 125-175 HA2 inhibited haemolysis, too. Thus, fusion activity of HA may be inhibited by anti-HA2 MAbs, mainly those binding to or near the fusion peptide. These antibodies represent useful probes for studies of influenza virus to cell membrane fusion.