T helper cell-mediated interferon-gamma expression after human parvovirus B19 infection: persisting VP2-specific and transient VP1u-specific activity

Abstract

Human parvovirus B19 is a small non-enveloped DNA virus with an icosahedral capsid consisting of proteins of only two species, the major protein VP2 and the minor protein VP1. VP2 is contained within VP1, which has an additional unique portion (VP1u) of 227 amino acids. We determined the ability of eukaryotically expressed parvovirus B19 virus-like particles consisting of VP1 and VP2 in the ratio recommended for vaccine use, or of VP2 alone, to stimulate, in an HLA class II restricted manner, peripheral blood mononuclear cells (PBMC) to proliferate and to secrete interferon gamma (IFN-γ) and interleukin (IL)-10 cytokines among recently and remotely B19 infected subjects. PBMC reactivity with VP1u was determined specifically with a prokaryotically expressed VP1u antigen. In general, B19-specific IFN-γ responses were stronger than IL-10 responses in both recent and remote infection; however, IL-10 responses were readily detectable among both groups, with the exception of patients with relapsed or persisting symptoms who showed strikingly low IL-10 responses. Whereas VP1u-specific IFN-γ responses were very strong among the recently infected subjects, the VP1u-specific IFN-γ and IL-10 responses were virtually absent among the remotely infected subjects. The disappearance of VP1u-specific IFN-γ expression is surprising, as B-cell immunity against VP1u is well maintained.

Introduction

Human parvovirus B19 is a small non-enveloped DNA virus with an icosahedral capsid of two protein species, the major viral protein VP2 (58 kDa) and the minor viral protein VP1 (83 kDa). VP2 is contained within VP1, which has an additional unique portion (VP1u) of 227 amino acids [1]. Of the 60 capsid protein molecules in the native virus particle, VP2 makes up ∼ 95%[2]. Although B19 is a significant human pathogen with a wide illness spectrum, subclinical infections are also common [3,4]. Symptomatic B19 infection is associated with HLA-DRB1 and HLA-B49 alleles [5]. The typical clinical manifestation is fifth disease (erythema infectiosum), a benign febrile rash with or without arthropathy [6]. The joint symptoms occasionally persist for years and may mimic rheumatoid arthritis [7]. In patients with haemolytic disposition B19 infection may cause transient aplastic crisis (TAC) [8]. Among immune compromised [9] and, rarely, immunocompetent individuals [10,11] B19 infection may remain chronic. Chronic infection has been proposed to be the aetiology of persistent B19-arthropathy [12,13]. During pregnancy B19 infection sometimes leads to fetal hydrops and death [14,15]. Occasionally, B19 infection may result in thrombocytopenia, neutropenia, myocarditis, hepatitis or various neurological complications [3]. Pneumonia, pleuritis [16,17] or generalized oedema in adults with signs of congestive heart failure [18] may furthermore result from acute B19 infection.

Due to the important morbidity, recombinant vaccines for the B19 virus are being developed [1,19]. For efficient protection, the vaccine should elicit neutralizing antibodies [9]. Empty capsids consisting of both VP1 and VP2 have fulfilled this criterion in both animal [1] and human models [19].

We have shown previously that such a recombinant antigen elicits vigorous T helper (Th) lymphocyte proliferation among both recently and remotely B19 infected subjects [17] and among the latter, similar proliferation, interferon (IFN)-γ and interleukin (IL)-10 responses were found with VP1/2 and VP2 capsid antigens [20]. IFN-γ is a proinflammatory, Th1-type cytokine stimulating intracellular killing of microbes [21]. IFN-γ also stimulates antigen presentation to helper (CD4⁺) and cytotoxic (CD8⁺) T cells [21] and has antiviral activity [22]. IL-10, in turn, is a potent anti-inflammatory Th2 type cytokine, suppressing intracellular killing [21]. It also inhibits antigen presentation, leading to T cell anergy [21], and it directly inhibits human Th cell proliferation and IL-2 production [23]. Furthermore, IL-10 is a class-switch factor for IgG1 [24], the predominant IgG subclass for B19 [25,26]. IFN-γ, in turn, down-regulates IgG1 secretion [27].

VP2, the major structural protein of human parvovirus B19, appears to provide the major target for B19-specific Th among remotely infected subjects [20]. The main goal of the present study was to investigate whether any Th cell reactivity could be found within VP1u. We studied recently as well as remotely B19-infected adults, by using isolated VP1u antigen. Due to the variation of the clinical pictures from transient rash to persistent arthralgia, we could also determine the relation of the B19-specific Th cell responses with the different clinical manifestations. Furthermore, we had the opportunity of studying the influence of pregnancy and immunosuppressive corticosteroid medication on B19-specific Th cell immunity.

Materials and methods

Patients with recent B19 infection

Donors of fresh peripheral blood mononuclear cells (PBMC)

Twelve recently infected, constitutionally healthy patients (designated C1–C12) were studied 25–70 days after onset of symptoms of serologically documented B19 infection. Patients C1 and C2 have been described elsewhere [17] as R5 and R6, respectively. Patients C11 and C12 were pregnant, and patient C11 underwent a fetal loss due to B19 infection (Table 1). For comparison, PBMC were collected from patient G 2 months after delivery. She had acute B19 infection on gestation week 24, and delivered a healthy baby at full term (Table 1). Patient L1, described elsewhere [17], had prolonged post-infectious arthropathy (rash and arthralgia for > 6 months after B19 infection). For comparison, patient R1 [17], who recovered from acute B19 infection without complications, was studied 515 days after onset of symptoms (Table 1).

Table 1. Clinical pictures of the B19-infected patients.

Subject, gender, age (years)	Days after onset of symptoms	Symptoms at onset	Other clinical data
C1, f, 31	30	A, R, F
C1 follow-up	155	-
C2, f, 54	40	A, R, F	Pneumonia and pleuritis, cured by short-term prednisone treatment (20 mg prednisone daily for 7 days with subsequently tapered dose. Treatment commenced 3 weeks before PBMC collection)
C2 follow-up	150	A (severe) relapseda
C10, m, 37	30	R, F	Thrombocytopenia (peripheral destruction of platelets), 14th day of prednisolone treatment, currently 60 mg daily
C10 follow-up	230	A (mild)a	Thrombocytopenia cured. Prednisolone tapered to 5 mg every other day
C11, f, 36	30	-	Fetal death (gestation week 19)
C11 follow-up	60	-
C12, f, 39	25	A, R, F	Symptoms during gestation week 27. No antiviral or transfusion therapy
			Full-term pregnancy and healthy baby
			Symptoms during gestation week 24. No antiviral or transfusion therapy
G, f, 36	180	R	Full-term pregnancy and healthy baby
R1, f, 33	515	A, R, F
L1, f, 30	180	A, R, Fa	Persisting arthritis and rash

f, Female; m, male; A, arthralgia; R, rash; F, fever.

^aSymptoms present during the study.

Control subjects

We also studied 23 remotely B19 infected subjects (10 men and 13 women; age range 21–50 years), designated S1–S23. S1 and S2 became pregnant during this study, and both delivered at full term. In addition, we studied 16 healthy B19-seronegative laboratory workers (three men and 13 women, age range 20–46 years). The members of this non-immune group were designated Ni1–Ni16.

Donors of cryopreserved PBMC

PBMC were collected from nine recently infected patients (designated A1–A9) and stored in liquid nitrogen. PBMC from patients A1–A9 were obtained 10–90 days after onset of symptoms of serologically documented B19-infection. Cryopreserved PBMC were used for studying of Th cell responses against unique portion of VP1 (VP1u) in recent infection, as this antigen was not in our hands at the time when fresh PBMC from recently infected subjects were obtained.

Cryopreserved control samples for recently infected patients A1–A9 were obtained from seven remotely B19 infected subjects (designated Sc1–Sc7) and five B19-seronegative subjects (designated Ni1–Ni5).

Informed consent was obtained from the study subjects. This study was approved the Ethical Committee of the Health Department of the City of Helsinki.

Antibody assays

B19-IgG and IgM were measured by EIA employing virus-like VP2 particles as antigen; B19 IgG was measured by EIA developed in house [17,28], whereas B19 IgM was measured by a commercial EIA (Biotrin, Dublin, Ireland). For exclusion or verification of recent B19 infection, all samples were studied further for epitope-type specificity of VP2-IgG [28].

Antigens

Virus-like particles

Recombinant VP1/2 and VP2 capsids were expressed, purified and sterilized exactly as described [17,20]. According to densitometry of silverstained gels [17,20], our VP1/2 capsid preparations contained approximately 66% VP2 and 33% VP1, the ratio recommended for vaccine use [1].

Recombinant unique portion of VP1 (VP1u)

The 227-aa VP1u was expressed prokaryotically and purified as described [29]. After extensive dialysis against phosphate buffered saline (PBS), endotoxins were removed by passage four times through endotoxin removal columns (Detoxi-Gel AffinityPak; Pierce, Rockford, IL, USA). The VP1u protein was then sterile-filtered by using 0·2 -µm filters (Anotop 10 Plus; Whatman, Maidstone, UK).

Tetanus toxoid (TT) and Candida albicans antigens

Tetanus toxoid (TT) and Candida albicans antigens [17] were used as control antigens.

Endotoxin was determined as before [17,20] in all preparations, and was < 0·0015 EU/µg with VP1/2 and VP2 proteins and 0·013 EU/µg with VP1u protein (>> 1·5 EU/µg in a preparation designated VP1u^e, from which the endotoxin removal step was omitted).

Isolation of PBMC

Blood was drawn to mononuclear cell separation tubes (Vacutainer CPT, Becton Dickinson, Franklin Lakes, NJ, USA) containing 0·45 ml sodium citrate. Cells were spun for 30 min at 1700 g, washed twice in PBS and were prepared within 2 h of sampling.

Cryopreservation of PBMC

If not used immediately, PBMC were cryopreserved in 10% dimethyl sulphoxide (DMSO) and 90% fetal calf serum (FCS) and stored in liquid nitrogen. For use, the PBMC were thawed, washed twice with complete RPMI and cultured identically as fresh PBMC [17,20].

Lymphocyte culture

Isolated PBMC were resuspended in complete RPMI-1640 containing 20 mM HEPES, 2 mMl-glutamine, streptomycin (100 µg/ml), penicillin (100 U/ml), 50 µM2-mercapto-ethanol (2-ME) and 10% heat-inactivated human AB serum (Finnish Red Cross Blood Transfusion Service) containing B19-IgG. Manually counted PBMC (200 000/well) and the antigens were cultured in 96-well U-bottomed plates (Costar, Corning Inc., Corning, NY, USA) in a humidified incubator (37°C and 5% CO₂). The B19 VP1/2 and VP2 capsids and the VP1u and VP1u^e proteins were used at 1·5 µg/ml, whereas the control antigens tetanus toxoid and C. albicans were used at 5 µg/ml at 2·5 µg/ml, respectively.

Proliferation assay

Manually counted PBMC and the antigens in triplicate were cultured for 6 days (37°C and 5% CO₂) and pulsed for the last 16 h with 1 µCi of tritiated thymidine (specific activity 50 Ci/mmol, Nycomed Amersham, Buckinghamshire, UK). Thymidine incorporation was measured in a liquid scintillation counter (Microbeta, Wallac Ltd, Turku, Finland). The data were expressed as counts per minute (Δ cpm): Δ cpm = mean cpm (test antigen) – mean cpm (media).

Cytokine assays

Cytokine assays were carried out as described [20]. Briefly, PBMC culture supernatants were harvested on day 3 for IFN-γ and on day 5 for IL-10, and were stored at −20°C. Cytokine production was analysed by using specific IFN-γ and IL-10 EIAs (BD PharMingen, San Diego, CA, USA) according to the manufacturer's instructions. Background cytokine production was subtracted from the total to yield antigen-specific cytokine production. The detection limits were 5 pg/ml for IFN-γ and 8 pg/ml IL-10.

Antibody blocking assays

Class restriction of the T cell IFN-γ and IL-10 responses were studied by using HLA class I (IgG2a, clone W6/32; Dako, Glostrup, Denmark) and class II-specific (HLA-DR) (IgG2a, clone L243; Becton Dickinson, San Jose, CA, USA) monoclonal antibodies at 1·5 µg/ml [17].

Statistical methods

Proliferation and cytokine responses were evaluated statistically using the Mann–Whitney U-test and the paired t-test. P-values < 0·05 were considered significant.

Results

VP1/2 and VP2 antigen-specific responses in recent versus. remote B19 immunity

The VP1/2 and VP2 antigen-specific proliferation, IFN-γ and IL-10 responses of the recently infected, immunocompetent subjects C1–9 were stronger (Table 2a and b) than those of the remotely infected subjects. However, statistical significance was achieved only with the proliferation and IFN-γ responses (P = 0·003), not with IL-10 (P= 0·52). Interestingly, with the control antigen TT, the recently infected patients showed weaker (P = 0·013) proliferation, IFN-γ and IL-10 responses (Table 2a and b). In the corresponding background responses, no statistically significant differences (P = 0·096) were found between the recently and the remotely infected subjects (Table 2a and b). The remotely infected subjects showed considerably stronger (P = 0·0001) B19-antigen specific proliferation, IFN-γ and IL-10 responses than did the seronegative subjects, whereas the control-antigen (TT) specific responses were similar (P = 0·77) (Table 2a and b).

Table 2a. Peripheral blood mononuclear cells (PBMC) proliferation responses to VP1/2, VP2 and to tetanus toxoid (TT) among recently (C1–9) and remotely (S1–23) parvovirus B19 infected subjects compared with responses among non-immune (Ni 1–16) subjects.

	Δ CPMa±s.d.
Subjects	VP1/2	VP2	TT
C1-9	54 025 ± 24 186	49 904 ± 25 806	11 108 ± 19 240
S1-23	20 001 ± 17 243	21 967 ± 18 888	29 323 ± 25 574
Ni 1–16	2 009 ± 2 632	1 961 ± 3 427	30 566 ± 25 975

^aΔcpm = antigen-specific counts per minute (cpm) minus background counts per minute (cpm). Background proliferation responses ranges were 499–3772 (mean 1436), 309–3277 (mean 1178) and 288–4752 (mean 1683) cpm among recently and remotely infected and seronegative subjects, respectively.

Table 2b. Peripheral blood mononuclear cells (PBMC), interferon (IFN)-γ and interleukin (IL)-10 responses to VP1/2, VP2 and to tetanus toxoid (TT) among recently (C1–9) and remotely (S1–23) parvovirus B19 infected subjects compared with responses among non-immune (Ni1–16) subjects.

	Δ CPMa (pg/ml) ± s.d.			Δ CPMb (pg/ml) ± s.d.
Subjects	VP1/2	VP2	TT	VP1/2	VP2	TT
C1-9	568 ± 456	499 ± 473	101 ± 211	50 ± 44	55 ± 65	14 ± 23
S1-23	138 ± 210	164 ± 239	158 ± 200	35 ± 28	34 ± 29	74 ± 63
Ni 1–16	0	1 ± 3	188 ± 25 975	4 ± 6	5 ± 7	82 ± 73

^aIFN-γ measured from tissue culture supernatant.

^bIL-10 measured from tissue culture supernatant. Background cytokine production was subtracted from total to yield antigen-specific cytokine production. Background IFN-γ responses ranged 0–0, 0–26 (mean 3) and 0–12 (mean 2) pg/ml among recently and remotely infected and seronegative subjects, respectively, and the corresponding background IL-10 responses ranges were 0–40 (mean 14), 0–22 (mean 7) and 0–27 (mean 9) pg/ml, respectively.

VP1/2 versus VP2 antigen-specific proliferation, IFN-γ and IL-10 responses among recently and remotely B19-infected subjects

The mean proliferation and IL-10 responses among the recently infected subjects C1–9 were similar (P = 0·38) in comparison of the VP1/2 and VP2 antigens (Table 2a and 2b). The mean IFN-γ responses appeared higher with the VP1/2 capsids than with the VP2 capsids, but statistical significance was not reached (P = 0·14). In magnitude the B19-specific proliferation, IFN-γ and IL-10 responses varied strongly from patient to patient, giving rise to large standard deviations (s.d.). The dominant B19-specific cytokine was usually IFN-γ, yet most patients also showed readily detectable IL-10 responses (Table 2b).

When VP1/2 and VP2 antigen-specific proliferation, IFN-γ and IL-10 responses within the remotely B19-infected subjects were compared, no statistically significant differences between the two B19 antigens (P = 0·071) were found (Table 2a and b), confirming our previous results [20].

VP1u versus VP2-specific IFN-γ and IL-10 responses among remotely and recently B19-infected subjects

We next determined the ability of Th cells to recognize isolated VP1u antigen. First, we used fresh PBMC from remotely infected subjects: stronger IFN-γ responses (P = 0·015) and IL-10 responses (P = 0·037) were found with VP2 capsids than with VP1u (Table 3). Proliferation responses (mean Δcpm ± s.d.) were also stronger with VP2 (12 282 ± 16 338) than with VP1u (328 ± 593) (P = 0·010), confirming that PBMC from remotely B19-infected subjects poorly recognize VP1u (Table 3).

Table 3. Peripheral blood mononuclear cells (PBMC) interferon (IFN)-γ and interleukin (IL)-10 responses to VP1u, VP2 and to control (CTRL) antigens among recently and remotely parvovirus B19 infected subjects compared with responses among non-immune subjects.

	IFN-γ (pg/ml) ± s.d.a			IL-10 (pg/ml) ± s.d.b
Subjects Average responses for groups and subgroups	VP1u	VP2	CTRL	VP1u	VP2	CTRL
Fresh PBMC
Remotely infected patients, S1–16	3 ± 4	57 ± 79	100 ± 156c	13 ± 19	33 ± 37	58 ± 69c
Non-immune subjects, Ni 1–9	14 ± 27	2 ± 5	107 ± 128c	7 ± 12	5 ± 11	60 ± 54c
Cryopreserved PBMC
Recently infected patients, A1–9	907 ± 622	301 ± 222	>600d	35 ± 43	20 ± 14	82 ± 70d
Remotely infected staff, Sc1–7	30 ± 48	197 ± 133	536 ± 85	30 ± 75	44 ± 83	140 ± 194d
Non-immune subjects, Ni 1–5	0	0	>600d	0	0	177 ± 92d

^aInterferon gamma measured from tissue culture supernatant.

^bInterleukin 10 measured from tissue culture supernatant.

^cTetanus toxoid was used as control antigen with fresh PBMC.

^dCandida albicans was used as control antigen with cryopreserved PBMC. Background cytokine production was subtracted from total to yield antigen-specific cytokine production. With fresh PBMC, background IFN-γ responses ranges were 0–12 (mean 3) and 0–9 (mean 1) pg/ml among remotely infected and seronegative subjects, respectively, and the corresponding background IL-10 responses ranges were 0–48 (mean 8) and 0–15 (mean 3) pg/ml. With cryopreserved PBMC, background IFN-γ responses ranges were 0–448 (mean 81), 0–174 (mean 28) and 0–0 pg/ml among recently and remotely infected and seronegative subjects, respectively, and the corresponding background IL-10 responses ranged 0–386 (mean 99), 28–620 (mean 99) and 62–242 (mean 120) pg/ml.

We then compared the ability of cryopreserved PBMC from recently and remotely infected subjects to recognize the VP1u and VP2 antigens (Table 3). The B19 antigen-specific responses of the recently infected patients differed strongly from the responses described above: the VP1u antigen elicited stronger IFN-γ and IL-10 responses than did the VP2 antigen (Table 3). However, statistical significance was reached only with IFN-γ responses (P= 0·017), not with IL-10 responses (P= 0·37).

With the remotely infected subjects, the B19 antigen-specific responses first determined with fresh PBMC were reproduced with cryopreserved PBMC, as stronger IFN-γ and IL-10 responses were detected with the VP2 antigen than with the VP1u antigen (Table 3). However, statistical significance was reached only with IFN-γ responses (P= 0·024), not with IL-10 (P= 0·37).

Of note, the recently infected patients showed clearly higher IFN-γ background responses with cryopreserved PBMC than with fresh PBMC (Tables 2b and 3). It is likely that factors related to recent B19 infection have had a major role in this enhanced spontaneous IFN-γ secretion, whereas cryopreservation possibly had a minor role, as the remotely infected subjects also showed somewhat higher IFN-γ background responses with cryopreserved PBMC than with fresh PBMC (Tables 2b and 3).

On the other hand, cryopreservation had a significant effect on spontaneous IL-10 secretion, as all subject groups showed clearly higher IL-10 background responses with cryopreserved PBMC than with fresh PBMC (Tables 2b and 3). The reasons for this increased IL-10 secretion are unknown. Possibly the high proportion (90%) of FCS in our cryopreservation media activated PBMC for IL-10 secretion upon freezing and/or thawing.

The effect of endotoxin contamination of VP1u antigen on IL-10 responses

The importance of endotoxin removal was elucidated by studying PBMC-mediated IL-10 responses with VP1u^e antigen from which endotoxins were not removed. Fresh PBMC were used. The IL-10 responses (mean ± s.d.) among 15 remotely infected seropositive and nine seronegative subjects were as high as 428 ± 262 and 394 ± 211 pg/ml (P = 0·68), respectively. These high and non-specific responses were expected, as endotoxins are known to activate monocytes to produce high levels of IL-10 [30].

Comparison of acute-phase and convalescent-phase T cell function

As shown in Table 4, during follow-up after primary infection, the B19 specific IFN-γ responses decreased and IL-10 responses strongly increased in patient C1, who recovered without complications. Patient C2, with a preceding short-term low-dose corticosteroid course (Table 1), had a very different response pattern: she showed a concomitant increase in B19-specific IFN-γ responses and disappearance of IL-10 responses during follow-up, at which time she suffered from a relapsed, severe arthralgia (Table 1).

Table 4. Individual peripheral blood mononuclear cells (PBMC), interferon (IFN)-γ and interleukin (IL)-10 and proliferation responses to VP1/2, VP2 and to tetanus toxoid (TT) among recently parvovirus B19 infected subjects compared with responses of remotely infected subjects (R1, S1 and S2) and a patient with persisting symptoms, L1.

	Δ cpma			IFN-γ (pg/ml)b			IL-10 (pg/ml)c
Subject	VP1/2	VP2	TT	VP1/2	VP2	TT	VP1/2	VP2	TT
C1	40 593	46 005	0	450	450	0	24	4 128	0
C1 follow-up	78 024	64 916	0	304	104	0	90	2
C2	56 920	62 587	7 532	600	280	0	21	10	0
C2 follow-up	36 152	35 015	7 575	1200	712	33	0	0	0
L1	63 675	66 383	54 360	1200	1168	312	25	13	15
R1	79 480	72 602	66 876	1200	1032	600	176	76	88
C10	6 632	14 055	5 983	15	15	0	60	55	20
C10 follow-up	19 100	22 554	38 504	9	11	138	9	11	120
C11d	17 283	13 251	13 431	25	8	8	6	10	12
C11 follow-up	12 040	7 808	1 969	30	14	10	48	20	34
C12d	6 583	8 980	820	21	39	10	6	14	0
Gd 2 months after delivery	6 750	9 801	10 951	68	53	17	0	0	0
S1 not pregnante	45 815	51 573	48 323	257	189	200	68	66	87
S1 gestation week 18	12 968	14 687	22 341	7	17	78	54	62	49
S1 gestation week 24	11 569	10 590	14 873	0	0	54	6	9	110
S1 5 months after delivery	9 882	12 048	14 160	36	78	36	78	102	248
S2 not pregnante	49 096	47 655	22 075	458	768	125	94	79	55
S2 gestation week 15	18 033	24 626	5 441	130	264	27	36	47	22
S2 gestation week 35	6 495	n.d.	3 201	200	n.d.	40	46	n.d.	40

^aΔcpm = antigen-specific cpm minus background cpm.

^bIFN-γ measured from tissue culture supernatant.

^cIL-10 measured from tissue culture supernatant. Background cytokine production was subtracted from total to yield antigen-specific cytokine production.

^dAcute B19 infection during pregnancy.

^eAverage results from three experiments before pregnancy; n.d., not done.

Patient C10, with a long-term and (initially) high-dose corticosteroid course (Table 1) for B19-associated thrombocytopenia, showed an initially strong IL-10 response with the B19 antigens, whereas the corresponding proliferation and IFN-γ responses were low (Table 4). At follow-up, his B19-specific IFN-γ responses remained low and IL-10 responses were profoundly reduced, whereas his proliferation responses increased with both B19 antigens. Interestingly, patient C10 showed much stronger proliferation, IFN-γ and IL-10 responses with the control antigen TT during follow-up than during onset (Table 4).

Comparison of patients with self-limiting or persistent symptoms

The B19-specific PBMC proliferation and IFN-γ responses during follow-up of patient R1 who recovered without complications, and patient L1 with rash and arthralgia persisting over 180 days after B19 infection were equally strong. By contrast, the B19-specific IL-10 response was very strong in patient R1 but minimal in patient L1 (Table 4).

PBMC responses in patients with B19 infection during pregnancy

Three subjects (patients C11, C12 and G) with B19 infection during pregnancy were studied (Table 1). Their B19-specific PBMC proliferation, IFN-γ and, importantly, IL-10 responses were much weaker than were those of corresponding recently infected non-pregnant patients. In all, strong B19-specific IL-10 responses were not encountered among patients with B19 infection during pregnancy. Patients C12 and G, who had symptomatic B19 infection and successful pregnancy, showed similar IFN-γ responses with the VP1/2 and VP2 antigens. Patient C11 with asymptomatic infection and fetal death, in turn, showed stronger IFN-γ responses with the VP1/2 antigen than with the VP2 antigen (Table 4).

Influence of pregnancy on established Th cell immunity

Before pregnancy, staff members S1 and S2 had shown particularly strong proliferation responses with B19 capsids. Also, their B19-specific IFN-γ and IL-10 responses were repeatedly strong and comparable to those seen among the recently infected patients C1-C9 (Table 4). During pregnancy, their proliferation, IFN-γ and, importantly, IL-10 responses were reduced with either B19 antigen. The mean background IL-10 response with S1 was 5 pg/ml before pregnancy; and only slight increases to 16 and 10 pg/ml were observed during gestation weeks 18 and 24, respectively. The mean background IL-10 response with S2 was 4 pg/ml before pregnancy; and a decline to 0 pg/ml was found during gestation weeks 15 and 35. The background IFN-γ responses with S1 and S2 before or during pregnancy were 0 pg/ml. After delivery, staff member S1 showed a slight increase in B19-specific IFN-γ responses and a strong increase in IL-10 responses, whereas her proliferation responses remained low (Table 4).

HLA restriction of IFN-γ and IL-10 secreting cells

HLA class restriction of the IFN-γ (Fig. 1a) and IL-10 (Fig. 1b) responses were studied by using class I- and class II-specific monoclonal antibodies. Blocking of antigen presentation via HLA class I by using monoclonal antibody W6/32 showed little, if any, inhibitory effect on B19-specific IFN-γ or IL-10 responses, whereas blocking of antigen presentation via HLA class II by using monoclonal antibody L243 strongly inhibited IFN-γ and IL-10 responses among all subjects.

Fig. 1.

Effect of HLA class I-specific (W6/32) and class II-specific (L243) monoclonal antibodies (mAbs) on B19-specific interferon (IFN)-γ (a) and interleukin (IL)-10 (b) responses. Background cytokine production (in the presence of corresponding blocking antibody) was subtracted from total to yield antigen-specific cytokine production. Subject groups: S3 is a healthy B19-seropositive, remotely infected ‘top responder’ having strong B19-specific T cell immunity. ‘C’ indicates recently infected patients. Note: patient C11 was pregnant, and had asymptomatic B19 infection and fetal death. Background IFN-γ responses were 0 pg/ml with or without mAbs, whereas backround IL-10 responses ranged 0–10, 0–8 and 0–7 pg/ml with ‘No Mab’, with W6/32 and with L243, respectively.

Discussion

To date, most knowledge of the possible role of B19 virus-specific T cell functions in clinical manifestations is based on indirect data, obtained by measuring cytokine mRNA in PBMC (without ex vivo stimulation with B19 antigens) or circulating cytokines. In these studies, IL-4 and −10 [31] and IFN-γ[31,32] mRNA expression has been detected in PBMC from recently B19 infected patients, and circulating cytokines tumour necrosis factor (ΤΝF)-α and IFN-γ have been detected among patients with B19-associated prolonged fatigue [33], meningoencephalitis [34] or myocarditis [35].

Thus, only few Th cell studies using B19 specific in vitro assays and recently infected patients have been published. Mitchell et al. concluded NS1-specific lymphocyte proliferation to correlate with the time of B19 infection rather than with the development of B19 arthropathy [36]. We have previously shown vigorous VP1/2 capsid protein-specific Th cell proliferation not only in recently infected adults, but also in several remotely infected, healthy subjects [17]. Among children with fifth disease, Corcoran et al. detected stronger proliferation responses with VP1 than with VP2 antigen, and weak IFN-γ responses with both B19 structural proteins [26]. They concluded that VP1u constitutes the major target for Th cells, and that children with recent infection have defective B19-specific IFN-γ responses [26].

Ours is the first in vitro study on B19-specific IFN-γ and IL-10 responses among recently B19-infected adults. CD4⁺ Th cells appeared to be the main responding cells among PBMC, as B19-specific responses were restricted among seropositive subjects, indicating that the responding cells are able to establish memory. Furthermore, HLA class II-specific antibodies blocked B19-specific IFN-γ and IL-10 responses in this study, whereas depletion of CD4⁺ cells abrogated them in our previous study [20]. IFN-γ emerged as the dominant B19-specific cytokine in both recent and remote infection; yet B19-specific IL-10 responses were readily detectable among asymptomatic, recently or remotely infected subjects, consistent with a role in restoration of immune system homeostasis upon infection clearance [21]. Only one patient, C10, who was treated by high-dose corticosteroids, showed B19-specific PBMC responses with IL-10 as the dominant cytokine. This supports the finding that corticosteroids, when present in high doses during priming, favour generation of effector Th cells producing mainly IL-10 [37].

To date, most studies concerning the pathogenesis of prolonged or relapsing B19 arthropathy suggest that persisting B19 infection is essential in pathogenesis of chronic arthropathy [13,38–40], whereas one report considers that virus persistence might not be necessary, as an autoimmune response could be triggered by the virus in predisposed individuals, this response being self-maintained also after virus clearance [41].

We studied two patients with relapsing or persisting symptoms. PBMC from these patients showed strikingly low IL-10 responses. Hence, if B19-specific Th cells recognizing persisting viral antigens or cross-reactive self-antigens play a role in pathogenesis, their insufficient regulation via IL-10 could lead to prolongation of symptoms. This should be studied further with more patients.

We also studied B19-specific PBMC responses during pregnancy. In that condition, down-regulation of cellular immunity is important, as the effector cells and/or Th 1 type cytokines such as IL-2, IFN-γ and ΤΝF-α are harmful to the conceptus [42]. Elevated levels of cytokines, hormones and other molecules are likely to play critical roles in suppressing Th 1 type immunity [42]. We found attenuated proliferation and cytokine responses both at the acute-phase and in long-term immunity. Attenuation of B19-specific IFN-γ responses during pregnancy has been suggested in a recent study lacking, however, internal control subjects [43]. In our study pregnancy strongly suppressed B19-specific proliferation and IFN-γ responses among both recently and remotely infected subjects yet interestingly, B19-specific IL-10 responses were also suppressed. We expected to see high B19-specific IL-10 responses among the pregnant subjects, as earlier studies using mitogen-activated PBMC have shown higher IL-10 responses among pregnant than non-pregnant women [42,44]. In our set-up, however, pregnancy suppressed all aspects of B19-specific PBMC responses. This suggests that pregnancy may suppress recall antigen-specific responses more strongly than mitogen-specific responses.

TT-specific responses were surprisingly low among the recently infected patients. The reasons for this are unknown, but a mild immunosuppression accompanied with acute infection has to be considered. One possible mechanism might be direct lymphocyte infection, known to occur in some animal parvovirus infections such as rat virus [45]. Indeed, B19 has been detected in lymphocytes [38]. Another mechanism might be higher spontaneous secretion of immunosuppressive cytokines, such as IL-10, by PBMC of recently infected subjects.

We initiated this study by using VP2 capsids and VP1/2 capsids containing ∼ 33% VP1 [17], a proportion higher than the ∼ 5% of the natural virus [2]. As statistically significant differences in T cell reactivity were not seen between the VP1/2 and VP2 capsids, it appears likely that with a natural B19 virus capsid, most Th cell reactivity targets epitopes within the major capsid protein VP2. With isolated VP1u antigen, the proliferation, IFN-γ and IL-10 responses were virtually absent in remote infection. However, strong VP1u-specific IFN-γ responses were found in recent B19 infection. We consider this the key finding of the present study, particularly as the opposite pattern exists for B19-specific B-cell immunity; during late convalescence, IgG for VP2 linear epitopes disappear, whereas IgG for VP1u persist [9,46]. Furthermore, these two phenomena might be linked: as bound antibodies influence the presentation of Th cell epitopes [47], the VP2-IgG could favour the presentation of Th cell epitopes within VP1u, leading to IFN-γ orientated Th cell ‘help’ inhibiting IgG secretion [27]. These VP1u-specific Th cells could even be cytotoxic to the B-cells specific for the primary structure of VP2 [48].

The reason why the strong IFN-γ responses with VP1u are confined within recent infection is currently unknown and deserves further study. One possible mechanism is too extensive activation-induced cell death (AICD) in the (late) convalescent phase [49]. Alternatively, the VP1u-specific Th cells could become suppressed by Τh3 cells secreting transforming growth factor β or by direct CD4⁺CD25⁺ cell contact [50].