Surface antigen-negative hepatitis B virus infection in Dutch blood donors.

Hepatitis B virus (HBV) surface antigen (HBsAg) is a reliable marker for HBV infection, but HBsAg-negative forms of HBV infection occur. The introduction of HBV DNA screening of Dutch blood donors, which were not preselected for absence of HBV core antibodies, enabled the characterization of HBsAg-negative HBV infection in healthy persons and a comparison of the HBV genomes involved. The screening of 4.4 million Dutch blood donations identified 23 HBsAg-negative, HBV DNA-positive persons. Serological testing of the index donations, follow-up samples and archived earlier samples was performed to determine the nature of each HBV DNA-only case. Despite low viral loads HBV DNA could be sequenced in 14 out of 23 donors, allowing HBV genotyping and the analysis of mutations in the HBV surface gene. Four types of HBsAg-negative HBV infection were detected: infection in the early stage before occurrence of HBsAg; suppressed infection after vaccination; HBV genotype G infection with decreased HBsAg production; and chronic occult (HBsAg negative) HBV infection. In the donors with occult HBV genotype D infection the HBV surface gene showed multiple "escape" mutations in the HBsAg a-determinant and CTL epitopes, while in an occult genotype A case the surface gene showed no mutations. HBsAg-negative forms of HBV infection in healthy blood donors explain the ongoing transmission of HBV via blood transfusion, if donor screening is limited to HBsAg. The screening of blood donors for HBV DNA and HBV core antibodies seems to cover all stages and variants of HBV infection.

Real-time polymerase chain reaction detection of parvovirus B19 DNA in blood donations using a commercial and an in-house assay.

BACKGROUND: European regulations require testing of manufacturing plasma for parvovirus B19 (B19) DNA to limit the load of this virus to a maximum acceptable level of 10 IU/µL. To meet this requirement, most manufacturers introduced a test algorithm to identify and eliminate high-load donations before making large manufacturing pools of plasma units. Sanquin screens all donations using a commercial assay from Roche and an in-house assay. STUDY DESIGN AND METHODS: Between 2006 and 2009, 6.2 million donations were screened using two different polymerase chain reaction (PCR) assays targeting B19 DNA. Donations with B19 DNA loads of greater than 1 × 10(6) IU/mL showing significant differences in viral load between the two assays were further analyzed by sequencing analysis. RESULTS: A total of 396 donations with B19 DNA loads of greater than 1 × 10(6) IU/mL were identified. Fifteen samples (3.8%) had discordant test results; 10 samples (2.5%) were underquantified by the Roche assay, two samples (0.5%) were underquantified by the in-house assay, and three samples (0.8%) were not detected by the Roche assay. Sequencing analysis revealed mismatches in primer and probe-binding regions. Phylogenetic analysis showed that 12 samples were B19 Genotype 1. The three samples not detected by the Roche assay were B19 Genotype 2. CONCLUSION: This study shows that 3.8% of the viremic B19 DNA-positive donations are not quantified correctly by the Roche or in-house B19 DNA assays. B19 Genotype 1 isolates showing incorrect test results are more common than B19 Genotype 2 or 3 isolates. Newly designed B19 PCR assays for blood screening should preferably have multiplexed formats targeting multiple regions of the B19 genome.

HBsAg-negative mono-infection with hepatitis B virus genotype G.

Infection with a genotype G strain of hepatitis B virus (HBV-G) often occurs as a co-infection with HBV genotype A. In mono-infection with HBV-G, the production of hepatitis B surface antigen (HBsAg), HBe antigen and anti-HBe seems diminished, hampering the serological diagnosis of HBV-G mono-infection. To corroborate this notion, we studied in detail a series of samples of a blood donor with transient HBV-G infection. In this donor, during the temporary presence of HBV DNA and the seroconversion to HBcore antibodies (anti-HBc), no HBsAg or hepatitis B e antigen was detected. During follow-up, no anti-HBe appeared. Multiple resistance mutations to lamivudine were present, demonstrating primary infection with a resistant HBV strain. Cloning and sequencing indicated that no other HBV genotype but genotype G was present. Like other HBV-G isolates, the DNA sequence of the HBsAg a-determinant showed no mutations that could explain the failure to detect HBsAg. Our findings demonstrate that HBV genotype G mono-infection occurs and that routine serology is unsuitable for its detection.

Cellular proviral HIV type 1 DNA load persists after long-term RT-inhibitor therapy in HIV type 1 infected persons.

In a set of 42 antiretroviral naive HIV-1 infected persons who were treated with either Zidovudine (AZT) monotherapy, or a combination of AZT + ddC (Zalcitabine) or AZT + ddI (Didanosine), the HIV-1 DNA load was measured by competitive polymerase chain reaction (PCR) and related to the HIV-1 RNA load in plasma, the CD4+ counts and to clinical markers. The question was whether a reduction in the cellular HIV-1 DNA level contributes to clinical benefit, as predicted by a lasting response in HIV-1 RNA levels in plasma. No significant decline relative to baseline in HIV-1 DNA load was found in the AZT monotherapy arm. In this arm the differences from baseline for both HIV-1 RNA load and CD4+ T cell counts were small and transient. In both combination therapy arms, the maximum mean decline in HIV-1 DNA load was 0.6 log and it never differed significantly from baseline. This is in contrast to plasma HIV-1 RNA load that declined earlier and steeper (mean of 1.5 and 1.9 log for AZT + ddC and AZT + ddI, respectively) and that remained significantly below baseline for 80 weeks. Although 9 of 42 (32%) of the patients under combination therapy had prolonged decreased plasma RNA levels, the proviral HIV-1 DNA remained present in the cells throughout the total follow-up of 144 weeks. In conclusion, combination therapy showed better laboratory parameter responses than AZT monotherapy, in agreement with the clinical data. The HIV-1 DNA sequences did not disappear in any of the patients, heralding renewed active infection after cessation of therapy.

Stability of HIV-1 RNA in blood during specimen handling and storage prior to amplification by NASBA-QT.

The influence of different storage temperatures and anticoagulation conditions on the HIV-1 RNA load as measured by NASBA-QT was examined. Blood specimens from 14 HIV-1 infected individuals were processed within 2 h after collection. The HIV-1 RNA load remained stable for at least 6 months when samples were frozen directly at -70 degrees C in lysis buffer. This lysis buffer fully inactivated the virus. When whole EDTA blood was stored, the HIV-1 RNA load was stable for 72 h at 25 degrees C, but it declined within 24 h at 4 degrees C. The HIV-1 RNA load in whole heparinized blood declined significantly after 24 h at both 4 and 25 degrees C. It was slightly lower (average of 0.18 log ml-1) than in whole EDTA blood. At 4 degrees C, the HIV-RNA load in serum and EDTA-plasma stored with lysis buffer did not decline up to 14 days. At + 30 degrees C, however, the load declined significantly after 2 days. Of clinical significance, the mean load in EDTA plasma was 0.5 log ml-1 higher than in serum. This difference was patient dependent (range 0.1-0.7 log ml-1). We thus recommend, for quantifying HIV-1 RNA by NASBA, to use preferably EDTA blood which is kept at room temperature until plasma separation. When using heparinized blood, the plasma should be stored frozen within 8 h.

Prospective longitudinal analysis of viral load and surrogate markers in relation to clinical progression in HIV type 1-infected persons.

The temporal relationship between viral and surrogate markers and clinical status was analyzed prospectively every 8 weeks in 34 asymptomatic HIV-1-infected persons. After 3 years, 25 persons remained clinically healthy whereas 9 persons showed clinical progression. In accordance with other reports we found that at study entry HIV-RNA load was predictive of clinical progression. All markers tested evolved significantly in time in both progressors and nonprogressors. The HIV RNA load in plasma and HIV DNA load in T cells were linearly related only in nonprogressors. In addition, the RNA/DNA ratio during follow-up was significantly higher in progressors, indicating a higher replication rate in progressors. The HIV DNA load correlated inversely with CD4+ T cell counts and positively with p24 antigenemia in both nonprogressors and progressors. A significant correlation of HIV DNA load with SI phenotype occurred in progressors only. HIV RNA levels correlated with beta 2-microglobulin level and with p24 antigenemia but not with SI phenotype. These three markers can all routinely be measured in plasma; however, only the HIV RNA levels appear to be informative for clinical progression. Six to 8 months before clinical progression, an SI phenotype switch, increased HIV RNA in plasma, and decreased CD4+ T cell counts were all indicative of an impending clinical event.

Increased production of nitric oxide correlates with viral load and activation of mononuclear phagocytes in HIV-infected patients.

The objective of our study was to determine the production of nitric oxide (NO) in patients infected with human immunodeficiency virus (HIV) and its relation to cellular immunity, activation of mononuclear phagocytes and the amount of circulating virus. Therefore, serum nitrate, the stable metabolite of NO, the number of peripheral CD4+ T-lmphocytes, serum neopterin, plasma HIV-RNA and HIV-DNA in peripheral blood mononuclear cells of afebrile HIV-infected patients were determined. Serum nitrate levels were significantly (p = 0.002) increased in HIV-infected patients (median 37 microM, range 13-137 microM, n = 77) in comparison to healthy subjects (median 28 microM, range 21-40 microM, n = 17). Serum nitrate levels did not correlate with the number of CD4+ T-lymphocytes (r = 0.05, p > 0.05). Serum nitrate levels were positively correlated with neopterin (r = 0.36, p = 0.05, n = 30), the amount of HIV-DNA in peripheral blood mononuclear cells (r = 0.63, p < 0.001, n =27) and plasma HIV-RNA levels (r = 0.35, p = 0.08, n = 27). A possible explanation of our findings is that HIV induces the production of NO by means of activated mononuclear phagocytes.

Differences between specificities of IgE and IgG4 antibodies: studies using recombinant chain 1 and chain 2 of the major cat allergen Felis domesticus (Fel d) I.

IgE- and IgG4 antibodies were compared for reactivity with recombinant chain 1 and chain 2 of the cat allergen Felis domesticus (Fel d) I. Recombinant chain 1 and chain 2 were coupled to sepharose and tested in IgE- and IgG4 radioallergosorbent test (RAST) experiments. Substantial IgE- and IgG4 binding was found. The fraction of Fel d I-specific antibody that bound to the recombinant chains was calculated. For chain 1, the mean value of this fraction was 0.30 for IgE and 0.23 for IgG4 (P = 0.05). For chain 2, the mean value of this fraction was 0.19 for IgE and 0.13 for IgG4 (P = 0.02). These results indicate that differences in fine specificity exist between IgE and IgG4 antibodies. Moreover, these findings support our results with chemically prepared peptides derived from these two chains and suggest that the B cells producing IgE antibodies are more likely to recognize a less 'native' form of Fel d I, compared with IgG4.

The role of the finger domain of tissue-type plasminogen activator (t-PA) and plasminogen activator inhibitor 1 (PAI-1) in initiation and progression of thrombolysis.

We further investigated the role of the finger (F) and the kringle-2 (K2) domains of tissue-type plasminogen activator (t-PA) in fibrin-stimulated plasminogen activation. To that end, the action of purified (wt) t-PA or of variants lacking F (del.F) or K2 (del.K2) was assessed either in a static, human whole blood clot-lysis system or in whole blood thrombi generated in the "Chandler loop". In both clot-lysis systems, significant differences were observed for the initiation of thrombolysis with equimolar concentrations of the t-PA variants. A relatively minor "lag phase" occurred in thrombolysis mediated by wt t-PA, whereas a 6.4-fold and 1.6-fold extension is found for del.F and del.K2, respectively. We observed identical lag-times, characteristic for each t-PA variant, in platelet-rich heads and in platelet-poor tails of thrombi. Since plasminogen activator inhibitor 1 (PAI-1) is preferentially retained in the platelet-rich heads, we conclude that the inhibitor does not interfere with the initial stage of thrombolysis but exerts its action in later stages, resulting in a reduction of the rate of clot lysis. A complementation clot-lysis assay was devised to study a potential interplay of del.F and del.K2. Accordingly, clot lysis was determined with combinations of del.F and del.K2 that were inversely varied in relation to equipotent dosage to distinguish between additive, antagonistic or synergistic effects of these variants. The isobole for combinations of del.F and del.K2 shows an independent, additive action of del.F and del.K2 in clot lysis.(ABSTRACT TRUNCATED AT 250 WORDS)

Plasminogen activator inhibitor-1 released from activated platelets plays a key role in thrombolysis resistance. Studies with thrombi generated in the Chandler loop.

To investigate the potential role of plasminogen activator inhibitor-1 (PAI-1), which is released from the alpha-granules of activated platelets, in thrombolysis resistance, we employed a model (the "Chandler loop") that mimics the formation of arterial thrombi in vivo and that can be manipulated in terms of rheological parameters and composition of blood cells. Light and electron microscopy revealed that the distribution of blood cells in Chandler thrombi is polarized, as it is in arterial thrombi, resulting in platelet-rich "white heads" and red blood cell-rich "red tails.". Resistance toward tissue-type plasminogen activator (TPA)-mediated thrombolysis parallels the presence of platelets that are fully activated in this system. We demonstrate that the PAI-1 released by the alpha-granules is preferentially retained within the thrombus and that the concentration of PAI-1 antigen is higher in the head than in the tail of the thrombus. The relative thrombolysis resistance of the heads of Chandler thrombi can be largely abolished by inclusion of an anti-PAI-1 monoclonal antibody that blocks that inhibitory activity of PAI-1 toward TPA. We propose that PAI-1, released from activated platelets, plays a key role in thrombolysis resistance and/or reocclusion after thrombolytic therapy. This is due to binding of PAI-1 to polymerized fibrin within the thrombus, followed by inhibition of TPA-mediated fibrinolysis.

Structure of the major cat allergen Fel d I in different allergen sources: an immunoblotting analysis with monoclonal antibodies against denatured Fel d I and human IgE.

In this paper we show the reactivity of monoclonal antibodies (mAbs) and human IgE with Fel d I from different allergen sources in reduced SDS-PAGE immunoblots. By SDS-PAGE analysis of affinity-purified 125I-Fel d I, a 14- to 20-kD band was found, which dissociated under reducing conditions into a 4- to 5-kD chain (chain 1) and a 11- to 15-kD chain (chain 2). In initial immunoblotting experiments with mAbs against Fel d I however, only chain 1 was detected, while the mAbs lost activity upon reduction of Fel d I. Therefore mAbs were raised against reduced and alkylated Fel d I. Two of the four mAbs to 'denatured' Fel d I that were obtained did react with chain 2 on an immunoblot under reducing conditions; the other two reacted with chain 1. The mAbs did not react with native Fel d I. With these mAbs and human IgE, differences between allergen source materials in blot patterns of Fel d I were detected. A variable molecular weight for the protein stained with mAb antichain 2 was found, and occasionally the presence of a 12-kD band stained with mAb antichain 1. Human IgE strongly bound to chain 1 of Fel d I, while only 2 out of 6 sera gave a strong reaction with chain 2. The additional 12-kD band was also recognized by human IgE. In a competitive radioimmunoassay with mAb antichain 1, differences in levels of 'denatured' Fel d I between commercial extracts were quantitated. In vitro 'denatured' Fel d I was generated under high pH conditions. The reactivity of human IgE with this 'denatured' Fel d I was demonstrated in indirect RAST experiments with mAb antichain 1. We conclude that mAb antichain 1 recognizes a form of Fel d I that is not detected by mAb antinative Fel d I, but does react with human IgE.

Affinity purification of a major and a minor allergen from dog extract: serologic activity of affinity-purified Can f I and of Can f I-depleted extract.

Most dog-allergic patients react to a major 25 kd component on sodium dodecyl sulfate blots, Can f I (Ag 13). We initially raised monoclonal antibodies (Cf-3 and Cf-2) reactive with IgE-binding components distinct from Can f I. After a slight modification, we immunized other strains of mice and produced monoclonal antibodies coded Cf-1a and Cf-1b reactive with Can f 1. We affinity purified the allergens, Can f I and "dog allergen 2" with Cf-1a and Cf-2 ascites, respectively, and house dust-rich dog dander. Comparison of purified Can f I with dog saliva in RAST demonstrated that Can f I is a potent allergen for most dog-allergic patients (average response, 70%). After depletion of dog saliva of Can f I, a slightly lower contribution for Can f I was found, but the overall results supported the conclusion that Can f I is a major allergen in dog saliva. Comparison of purified dog allergen 2 with dog dander in RAST demonstrated that dog allergen 2 is less important for dog-allergic patients (average response, 23%). We radiolabeled the purified allergens and developed assays to measure Can f I and dog allergen 2 in allergen extracts and dust samples. Dog saliva was a strong allergen source, dog urine and feces contained very little of the allergens, and both allergens were found to a variable degree in the nine dog breeds tested.

Evidence for a Fel d I-like molecule in the "big cats" (Felidae species).

In this study, we investigated the cross-reactivity pattern of IgE and IgG4 antibodies to the major feline allergen, Fel d I. We studied the IgE and IgG4 response of 11 cat-allergic patients against Fel d I-like structures in eight members of the Felidae family: ocelot, puma, serval, siberian tiger, lion, jaguar, snow leopard, and caracal. Hair from these "big cats" was collected, extracted, and used in a RAST system and histamine-release test. By means of a RAST-inhibition assay with affinity-purified Fel d I from cat dander, it was established that, in the Felidae species, a Fel d I equivalent is present that reacts with IgE and IgG4 antibodies. We found that all patients had cross-reacting IgE antibodies to seven of the Felidae tested; no IgE antibodies reactive with the caracal were found. Eight of 10 patients with IgG4 antibodies directed to cat dander also had IgG4 antibodies directed to several Felidae species, including the caracal. However, the correlation between the IgE and the IgG4 antibody specificity was low, indicating that, in the case of Fel d I IgE and IgG4, antibodies do not necessarily have the same specificity.

Monoclonal antibodies to the major feline allergen Fel d I. I. Serologic and biologic activity of affinity-purified Fel d I and of Fel d I-depleted extract.

Monoclonal antibodies were raised against a major feline allergen, Fel d I. The specificity of the antibodies for Fel d I was demonstrated by a modified crossed immunoelectrophoretic procedure. These antibodies were used in affinity purification and depletion. Fed d I was eluted from the affinity matrix at a low pH. It was tested in serologic and biologic assays. Results of both the RAST and the histamine release test with affinity-purified Fel d I confirmed that it is a very potent allergen for the majority of patients allergic to cats. In the extract depleted with the monoclonal antibody, the ratio Fel d I/cat albumin was reduced by a factor of 20. With this depleted extract, 18 of 25 patients showed a reduction greater than 50% of the bound anti-IgE in the RAST compared with the RAST with crude cat extract. In the histamine release test with cells from two patients, the activity of the depleted extract was reduced by factors of 30 and 40, respectively. Depletion of Fel d I by polyclonal antibodies resulted in a reduction of the ratio Fel d I/cat albumin by a factor of 330. In the RAST, a reduction of the IgE response greater than 50% was found in 23 of 25 patients; in 12 patients the reduction was greater than 90%. In the histamine release test, the polyclonally depleted extract was 200 to 300 times less potent than the crude cat-dander extract. These results indicate that a very large proportion of the allergenic activity of cat-dander extract is caused by a single molecule: Fel d I.

Discrepancies between the skin test and IgE antibody assays: study of histamine release, complement activation in vitro, and occurrence of allergen-specific IgG.

Intracutaneous skin tests (STs) and RAST with the common allergens, grass pollen, house dust mite, and cat dander, were performed on 660 adult patients. In 117 patients (18%), we found 140 discordances (7%) in a total number of 1980 ST and RAST combinations. In agreement with studies in the literature, greater than 80% of the discordances consisted of positive skin reactions without detectable allergen-specific IgE antibodies in serum. The percentages of discordant results were similar for the three allergens. Reproducibility of both the RAST and the ST was evaluated in the discordant group. Repetition of the routine RAST procedure elicited results similar to those in the first test in 81% (105/130). A second ST elicited identical results in 89% (47/53). In addition to the routine IgE antibody assay, sera of patients with a positive ST but without detectable IgE antibodies were tested in two other RAST systems: (1) a RAST with allergen extracts from the same production batch as the ST reagents, and (2) the Pharmacia RAST. In spite of having a clearly positive ST, sera from 68 (80%) of 85 patients remained completely negative in all three RAST systems. Histamine release (HR) in vitro from washed leukocytes was studied in 35 patients with a reproducible positive ST and negative RAST results with serum. Interpretation of this test was possible in 28 patients. In 82% (23/28) of these patients, clearly detectable HR was found with the relevant allergen extract. A role of IgE in the skin reactions and HR tests was confirmed by positive RAST results with IgE that was affinity purified from serum of seven of these patients. Allergen-specific IgG4 antibodies are unlikely to be implicated, since no antibodies against grass pollen and house dust mite were detectable in sera of these patients. Only 18% of the patients with an unexplained skin reaction with cat dander have detectable IgG4 antibodies, but these antibodies were found in a similar frequency in a nonallergic, ST negative control group. Low total IgG responses precluded false negative RAST results caused by competition of IgG antibodies with IgE antibodies. There were no significant differences in the degree of complement activation in vitro by house-dust extracts between healthy control subjects, nonallergic patients, and patients with unexplained skin reactivity. It is concluded that a high proportion of the positive skin reactions with common inhalant allergens, which are not accompanied by a positive RAST, are probably caused by IgE antibodies that are not detectable in serum with any of the RAST procedures.(ABSTRACT TRUNCATED AT 400 WORDS)

Skin tests and histamine release with P1-depleted Dermatophagoides pteronyssinus body extracts and purified P1.

Monoclonal antibodies were raised against P1 and Dp X, two major allergens in Dermatophagoides pteronyssinus extracts. The concentrations of IgE antibodies to P1 and Dp X in sera of mite-sensitive patients were determined in RAST with immunosorbent-bound monoclonal antibodies used for insolubilization of the allergens. The major allergen-specific IgE titers were compared with the total IgE response against D. pteronyssinus. The results of these serologic assays confirmed studies in the literature that P1 and Dp X are major allergens. The contribution of IgE anti-P1 to the total antimite response frequently exceeded 50% and, in general, appeared to be higher than the contribution of IgE anti-Dp X. Twenty percent of the mite-sensitive patients had no detectable IgE to either P1 or Dp X. The contribution of P1 to the biologic activity of D. pteronyssinus body extracts was derived from the effect of P1 depletion on the reactivity in the histamine-release test and skin test. This technique was preferred to the study of purified allergen because biologic activity of the nonabsorbed components is not affected. Immunization of rabbits with affinity-purified P1 yielded monospecific polyclonal antisera. Mite extracts depleted with either monoclonal or polyclonal anti-P1 were applied in the histamine-release test. The skin test was performed with extracts depleted with polyclonal anti-P1. In addition, the activity of affinity-purified P1 was investigated in these tests. The results indicated that P1 depletion of D. pteronyssinus body extracts had no detectable effect on the activity in most patients, namely, at least 70% of the activity was retained in the depleted extract. There was a considerable variation between patients in the sensitivity for purified P1, as compared to the sensitivity for whole D. pteronyssinus extracts. In the histamine-release test, the activity of purified P1 was up to 35% of the activity of the D. pteronyssinus body extract but did not exceed 10% in most patients. This was in agreement with the relative activity of purified P1, as found in the skin test. Therefore, the contribution of P1 to the biologic activity of D. pteronyssinus body extracts, as measured by end point titration, appeared to be less than expected on the basis of the serologic studies and articles in literature. The depletion experiments stress that there is still much uncertainty as to the biologic activity of house dust-mite extracts.

Activation of the classical pathway of human complement in vitro by house-dust extracts is caused by IgM antibodies to polysaccharide antigen(s) and is not related to atopy.

The mechanism of activation of the classical pathway of human complement by house-dust extracts and its relevance to atopic disease was studied. Our results confirm that for most sera of adults, house-dust extract is, on a weight basis, a more potent C-activator than aggregated human IgG or lipopolysaccharide (endotoxin). Naturally occurring IgM antibodies directed against ubiquitous polysaccharides appeared to be the dominant factor in the C1 activation by house-dust extracts in human sera. Large variations were found between sera with respect to the concns of these IgM antibodies as measured by C1 activation or fixation of haemolytic complement. The IgM antibody titre was, however, not associated with atopic disease. Consequently, we do not support the hypothesis put forward by Berrens et al. (1978) (Allergol. Immunopath. 6, 45-54) that there might be a relation between atopy and enhanced reactivity of serum complement with allergenic extracts. More than 90% of the C-activating potential of allergen extracts like house dust was found in the fractions with high mol. wt material (mol. wt greater than 100 K). Therefore, these antigens are easily separated from the known IgE-binding major allergens of house-dust mite and cat dander.

Serologic aspects of IgG4 antibodies. II. IgG4 antibodies form small, nonprecipitating immune complexes due to functional monovalency.

Human IgG4 antibodies directed against phospholipase A, the P1 antigen from Dermatophagoïdes pteronyssinus extracts, and cat albumin were found unable to cross-link antigen. Previously, it was demonstrated that IgG4 antibodies, in contrast to IgG1 antibodies, did not cross-link Sepharose-bound antigen and antigen added in solution. To eliminate the possibility that this phenomenon was caused by preferential binding of both IgG4 Fab fragments to the solid-phase-bound antigen, cross-linking of antigen was studied in a fluid-phase system. In this test, incapability of IgG4 antibodies to bridge two antigens was also found. As a result of such a phenomenon, it is expected that immune complexes formed by IgG4 antibodies will be considerably smaller than complexes formed by IgG1. This was confirmed by analysis of the molecular size profiles of IgG1- and IgG4-containing immune complexes in sucrose-density gradients. Moreover, IgG1 was able to precipitate antigen in a radioimmunoprecipitation test, whereas precipitation was not demonstrable by the same amount of IgG4 antibodies. Even 3% polyethylene glycol 8,000 did not precipitate the small IgG4-containing immune complexes efficiently. The antibodies studied were of a high-affinity type, and there was no significant difference in association constants between IgG1 and IgG4 antibodies. Therefore, we were not able to confirm observations reported in the literature that the IgG4 subclass is associated with a low-affinity antibody response; probably, the affinity of the IgG4 antibodies was underestimated by other investigators because of the polyethylene glycol precipitation technique used to separate antibody-bound and free antigen. Our findings stress the point that IgG4 antibodies take a special place in the immune response upon chronic exposure to antigen.