Spleen-Dependent Turnover of CD11b Peripheral Blood B Lymphocytes in Bovine Leukemia Virus-Infected Sheep

Abstract

Lymphocyte homeostasis is determined by a critical balance between cell proliferation and death, an equilibrium which is deregulated in bovine leukemia virus (BLV)-infected sheep. We have previously shown that an excess of proliferation occurs in lymphoid tissues and that the peripheral blood population is prone to increased cell death. To further understand the mechanisms involved, we evaluated the physiological role of the spleen in this accelerated turnover. To this end, B lymphocytes were labeled in vivo using a fluorescent marker (carboxyfluorescein diacetate succinimidyl ester), and the cell kinetic parameters (proliferation and death rates) of animals before and after splenectomy were compared. We show that the enhanced cell death observed in BLV-infected sheep is abrogated after splenectomy, revealing a key role of the spleen in B-lymphocyte dynamics.

Among deltaretroviruses, human T-cell leukemia virus type I (HTLV-1) and bovine leukemia virus (BLV) are etiologic agents for lymphoproliferative diseases possibly leading to leukemia (5, 6, 18, 44, 54). With high frequencies of tumor development and reduced latency periods, the experimental infection of sheep with BLV is a model for the study of a mechanism of transformation. In BLV-infected sheep, B-cell lymphocytosis essentially results from expansion of CD11b⁺ B lymphocytes (7), whereas CD4⁺ T lymphocytes are the main targets for HTLV-1 (35). Despite marked differences between the two viral systems, the BLV model might be informative in understanding HTLV-1-induced leukemogenesis, essentially on the basis of the numerous structural and functional homologies between the two viruses (50). In this context, we have been particularly interested in the dynamic parameters that govern the accumulation of infected cells.

Lymphocyte homeostasis results from a subtle equilibrium between different parameters, including cell proliferation, differentiation, death, and recirculation between peripheral blood and secondary lymphoid organs. We have previously used different approaches to analyze these mechanisms directly in studies with BLV-infected sheep. First, the proliferation rates of B lymphocytes in BLV-infected and control sheep were compared via a method based on intravenous injection of bromodeoxyuridine (BrdU). This nucleoside analog of thymidine incorporates into the nascent DNA strand of dividing cells and subsequently can be detected by flow cytometry. By this approach, we showed that B cells of infected sheep proliferate significantly faster than those of uninfected controls. This difference in proliferation capacities was even further increased at the terminal neoplastic stage of the disease. In contrast, the death rates of BrdU-positive cells were similar between infected and control sheep (9). Importantly, these death parameters pertain to the cells that incorporated BrdU and not to the entire B-lymphocyte population. However, the net increase in proliferation in the absence of compensating cell death theoretically creates a very fast doubling of the lymphocyte population, a phenomenon that is not observed in vivo.

To resolve this apparent discrepancy, another approach was designed to specifically study the kinetics of B lymphocytes located within the peripheral blood. The principle of the technique is based on snapshot blood labeling with carboxyfluorescein diacetate succinimidyl ester (CFSE) (38). Direct intravenous injection of this cell-permeant fluorescent dye leads within seconds to the labeling of more than 98% of peripheral blood cells. Moreover, NH₂ labeling of target proteins ends within a few minutes, most likely due to the instability of the succinimidyl ester moiety. Providing that equal amounts of proteins are distributed among daughter cells, the number of cell divisions undergone since labeling can be estimated from flow cytometry data. Indeed, combined with the percentages of CFSE-positive cells, analysis of the CFSE mean fluorescence intensity allows the estimation of peripheral blood cell death and proliferation rates in vivo (2). Using this approach, it has been shown that the proportion of B lymphocytes labeled with CFSE decreased faster in BLV-infected sheep than in controls, a difference that was due to an increased cell death of peripheral blood B lymphocytes (10).

Stable CFSE labeling of peripheral blood B cells also permits the tracing of these cells while they recirculate through lymphoid organs. Hence, the recirculation of B cells to lymph nodes was assessed by the surgical establishment of cannulae in different efferent lymphatic vessels, allowing the sampling of lymph (10). These experiments led to the conclusion that B lymphocytes from BLV-infected and control sheep recirculate with similar efficiencies.

The most likely model that is consistent with all results of these kinetic experiments is that the excess of proliferation in lymphoid organs is balanced by increased cell death of peripheral blood B cells. Importantly, the increase in cell death observed for BLV-infected sheep is almost exclusively due to the CD11b-positive B subpopulation (10). In sheep, these cells are essentially absent from lymph, lymph nodes, and ileal Peyer's patches but are located in the peripheral blood and in the marginal zone of the spleen (16), whereas the B-lymphocyte population recirculating from blood through lymph and lymph nodes has a B⁺ CD11b⁻ phenotype (34, 55). We therefore hypothesized that the spleen plays a major role in the increased B-cell turnover, and we determined the CFSE kinetic profile (e.g., the evolution of the percentage of cells labeled with CFSE over time) for samples from splenectomized BLV-infected sheep.

MATERIALS AND METHODS

Experimental animals and surgical procedures.

Fifteen sheep were maintained under restricted conditions (L2) at the Study Center of Animal Productions (Gembloux, Belgium). Five animals (sheep no. 2147, 3004, 4164, 4533, and 4534) were used as uninfected controls, whereas sheep no. 2091, 2672, 3002, 4270, 4535, and 4536 were experimentally infected either with a BLV wild-type cloned provirus (strain 344) as previously described (52) or with blood from an infected sheep. Plasmid pBLVIG4, containing a mutant provirus harboring a stop codon in the G4 open reading frame [51]) was injected into sheep no. 1071, 1077, 4537, and 4538. The leukocyte and lymphocyte counts, which were regularly determined using an MS 4-5 vet automated cell counter (Mellet Schloesing Laboratories), did not vary during the course of the CFSE kinetics experiments. This study was performed between 5 and 40 months after BLV challenge. All sheep used during the course of this experiment were nonlymphocytic.

Handling of animals and experimental procedures were approved by the ethics committee and were conducted in accordance with institutional and national guidelines for animal care and use. Sheep were anesthetized by intravenous injection of approximately 10 ml of Nembutal (50 mg/ml; Abbott Laboratories), followed by a perfusion of 2 ml/hour during the surgery. Blood was collected under aseptic conditions from the splenic arteries and veins of sheep no. 2672, 4533, 4534, and 4535. The spleen was then removed after ligature of the splenic veins and arteries of sheep no. 1077, 3002, 3004, 4164, 4270, 4534, 4535, and 4538.

In vivo CFSE labeling, cell isolation, and ex vivo short-term culturing of PBMCs.

The labeling procedures were performed essentially as previously described (38). Briefly, 25 mg of CFSE dissolved in 4 ml of dimethyl sulfoxide containing 40 μl of heparin (1,000 U/ml) was injected into the jugular vein. At regular time intervals, blood was collected by jugular venipuncture and mixed with 0.3% (wt/vol) of EDTA used as an anticoagulant. Red blood cells were lysed with 1× fluorescence-activated cell sorter lysing solution (Becton Dickinson Immunocytometry Systems) for 15 min at room temperature, and the leukocytes were washed twice with phosphate-buffered saline (PBS). Peripheral blood mononuclear cells (PBMCs) were isolated by Percoll density gradient centrifugation (GE Healthcare) and washed twice with PBS-0.075% EDTA and at least three times with PBS alone. After estimation of their viability by trypan blue dye exclusion, 2.10⁶ cells were cultivated for 16 h in a 5% CO₂-air atmosphere at 37°C in 1 ml of complete RPMI 1640 medium (i.e., supplemented with 10% fetal calf serum, 2 mM l-glutamine, 100 U of penicillin, and 100 μg of streptomycin per ml; Invitrogen).

Immunophenotyping and analysis of viral expression.

Cells were labeled with monoclonal antibodies directed against immunoglobulin M (IgM) (1H4, mouse IgG1, used for single labeling and Pig45A2, mouse IgG2b, used for B and CD11b, B and CD5, B and L-selectin, and B and CD21 dual labeling), CD4 (ST4, mouse IgG1), CD8 (CACT80C, mouse IgG1), CD11b (CC125, mouse IgG1), and γδ T (86D, mouse IgG1) cells provided by K. Walravens (CERVA, Uccle, Belgium) and by one of us or obtained from VMRD Inc. Cells were then labeled with a rat anti-mouse IgG1 phycoerythrin (PE) antibody or with a rat anti-mouse IgG2a+b peridinin chlorophyll protein (PerCp) antibody (Becton Dickinson Immunocytometry Systems).

For detection of cells expressing viral proteins, PBMCs were cultivated for 16 h, washed twice in PBS-10% fetal calf serum and incubated for 15 min at room temperature with the fixative Intrastain Reagent A (DakoCytomation). After being washed, the cells were permeabilized for 10 min at room temperature with fluorescence-activated cell sorter permeabilizing solution (Becton Dickinson Biosciences). Cells were then incubated with a monoclonal antibody, 4′G9, directed against the major viral capsid protein p24 (4′G9 and mouse IgG1 were provided by D. Portetelle, FUSAGx, Gembloux, Belgium) and with a rat anti-mouse IgG PE antibody (Becton Dickinson Immunocytometry Systems). Finally, the labeled cells were resuspended in PBS and analyzed by flow cytometry using a Becton Dickinson FACScan instrument (Becton Dickinson Immunocytometry Systems).

Kinetics profiles obtained after in vivo CFSE labeling were analyzed with a mathematical model which is described in reference 2. This model basically uses two sets of data from the flow cytometry analyses: the proportion of CFSE-labeled cells and the mean fluorescence intensities. Compilation of these two types of experimental data allows for the calculation of the proliferation and death rates, with the latter parameter also including the disappearance of the cells out of the blood.

Determination of the proviral loads.

Genomic DNA was extracted either from 300 μl of blood containing 0.3% EDTA or from Percoll gradient-purified PBMCs. DNA preparation from blood required prior disruption of erythrocytes with a threefold excess of cell lysis buffer (Promega). Samples were then incubated overnight at 37°C with RNase A (0.1 mg/ml) and proteinase K (0.2 mg/ml) in a buffer containing 10 mM Tris-HCl, pH 8, 150 mM NaCl, 10 mM EDTA, and 0.5% sodium dodecyl sulfate. The DNA was next purified by a phenol-chloroform extraction and by ethanol precipitation. One hundred nanograms of genomic DNA was used for real-time PCR amplification of BLV proviral sequences essentially as described in reference 27. Briefly, a segment corresponding to the pol gene (nucleotides 3994 to 4060, according to the BLV GenBank entry with accession number K02120) was amplified with a pair of primers (5′-GAAACTCCAGAGCAATGGCATAA-3′ and 5′-GGTTCGGCCATCGAGACA-3′ at a 900 nM final concentration) and revealed with a minor groove binder (MGB) fluorescent probe (250 nM of 6-carboxyfluorescein-CTCACCCACTGCAAC-MGB) by use of the TaqMan PCR universal master mix on a GeneAmp 5700 apparatus (Applied Biosystems). A standard curve was generated after amplification of defined viral copy numbers (from 1 to 10⁷ of plasmid pBLV344) with 100 ng of control genomic DNA. In order to correct for differences in DNA concentrations and amplification efficiencies between samples, the 18S ribosomal DNA was quantified in parallel using probe 6-carboxyfluorescein-CATGCCGACGGGCGCTGA-MGB at 250 nM and a pair of primers at 900 nM (5′-TTGGATAACTGTGGTAATTCTAGAAGCTAA-3′ and 5′-CGGGTTGGTTTTGATCTGATAAAT-3′). The proviral loads were finally normalized to the numbers of PBMCs (of note is the fact that when the DNA was extracted from blood, the proviral loads were normalized to the numbers of leukocytes).

RESULTS

B lymphocytes in the splenic artery and vein of BLV-infected sheep display similar phenotypes.

We previously demonstrated that, compared to uninfected controls, CFSE-labeled B lymphocytes disappear quickly from the peripheral blood of BLV-infected sheep (10). This difference in kinetics is mainly due to increased cell death in the CD11b-positive B subpopulation. These cells are found in the marginal zone of the spleen but are excluded from other lymphoid compartments (lymph, lymph nodes, and ileal Peyer's patches) (16).

In order to evaluate a potential role of the spleen in these differential B-lymphocyte dynamics, we first compared the phenotypes of cells isolated from the artery and the vein that irrigate the spleen. To this end, samples were collected from these two blood vessels as well as from the jugular veins of two BLV-infected sheep (no. 4535 and 2672) and of two controls (no. 4533 and 4534). After purification by gradient density centrifugation, PBMCs were immunolabeled as described in Materials and Methods and analyzed by flow cytometry (Table 1; each set of data results from three independent labeling experiments). Within a given animal, the percentages of B cells in the total PBMC population did not differ significantly (unpaired Student t test) for the splenic artery, the splenic vein, and the jugular vein, although the relative proportions varied among animals. A similar conclusion holds true for the percentages of CD11b-positive cells within the B-lymphocyte population (Table 1).

TABLE 1.

Phenotype of cells isolated from the splenic artery and vein

Sheep	% of B lymphocytes in PBMC population from:			% of B+ CD11b+ cells in B-lymphocyte population from:			% Cells expressing p24 in PBMC population from:			Avg viral copy no. per 100 cells
	Jugular vein	Splenic artery	Splenic vein	Jugular vein	Splenic artery	Splenic vein	Jugular vein	Splenic artery	Splenic vein	Jugular vein	Splenic artery	Splenic vein
Controls
4533	28.1 ± 1.3	31.2 ± 3.6	33.0 ± 3.0	80.0 ± 3.1	83.9 ± 2.5	87.6 ± 2.2
4534	16.5 ± 3.3	18.9 ± 4.1	23.4 ± 1.9	75.3 ± 4.0	79.3 ± 0.6	80.8 ± 3.2
BLV-infected
4535	52.8 ± 5.8	51.4 ± 6.2	56.0 ± 5.3	68.1 ± 2.1	67.5 ± 4.1	73.0 ± 4.6	22.4 ± 1.2	22.1 ± 1.8	22.7 ± 1.8	59 ± 24	61 ± 12	68 ± 10
2672	25.6 ± 4.2	18.4 ± 3.1	22.3 ± 2.8	78.9 ± 2.2	81.5 ± 2.7	83.0 ± 1.2	1.0 ± 0.3	1.1 ± 0.3	1.1 ± 0.2	2.0 ± 0.3	2.5 ± 1.1	3.2 ± 1.3

To further assess the capacity of BLV-infected cells to bypass the spleen, two types of assays were performed. First, the proviral loads were measured in blood samples collected at the spleen entry and exit by use of real-time PCR. It appeared that although the absolute levels were different in two BLV-infected sheep (4535 and 2672), the proviral loads (normalized with the amplification of 18S ribosomal DNA and represented as numbers of copies per 100 PBMCs) were not significantly different (unpaired Student t test) in the splenic artery and vein (Table 1). A second assay was designed to evaluate the capacities of infected cells isolated from the two blood vessels to spontaneously express virus ex vivo. After short-term culturing in RPMI medium, the proportions of cells expressing the viral p24 capsid antigen, as determined by flow cytometry, were similar in the splenic artery and vein (Table 1).

We thus conclude that the phenotype of cells isolated from the peripheral blood upstream and downstream of the spleen is not subject to any significant change, at least in those sheep in a stable state of infection. These preliminary conclusions were somewhat expected, considering the high splenic flow rate and the overall daily turnover rates of peripheral B cells (see Discussion).

Splenectomy of BLV-infected sheep restores a normal B-lymphocyte kinetics.

Although no major instant differences were observed at the splenic entry and exit, it remained possible that the cell turnover in BLV-infected sheep is affected over a longer time period by splenectomy. Therefore, we surgically splenectomized BLV-infected or control sheep, and after a delay of 4 to 6 weeks for recovery, the operated animals were injected intravenously with the CFSE fluorescent dye in order to determine their B-lymphocyte kinetic profile. Blood was collected before and at regular time intervals after CFSE injection, PBMCs were purified, and B lymphocytes were analyzed by flow cytometry.

Representative dot plots of IgM and CFSE dual labeling performed with PBMCs from an infected sheep (no. 4535) and a control sheep (no. 4534) are illustrated in Fig. 1. Before CFSE injection, only barely detectable background levels (<0.1%) were measured in the B-cell population. Before splenectomy, the percentages of B lymphocytes labeled with CFSE dropped from 60.7% and 74.3% after 15 min to 10.1% and 3.1% at 5 weeks in control and BLV-infected sheep, respectively (Fig. 1). These consecutive flow cytometry analyses thus illustrate the accelerated disappearance rate of CFSE-positive cells in infected sheep.

FIG. 1.

A BLV-infected sheep (4535) and a control (4534) were injected intravenously with 25 mg of CFSE before and after splenectomy. Blood was collected by jugular venipuncture before and 15 min, 2 days, and 5 weeks after injection. After PBMC isolation, B lymphocytes were labeled with anti-IgM antibodies, and 20,000 events were collected by use of flow cytometry. The x axis and the y axis correspond to CFSE and B-lymphocyte labeling, respectively. The percentages of CFSE⁻ B⁺ and CFSE⁺ B⁺ cells in the total B-lymphocyte population are indicated in the upper quadrants of the dot plots.

After splenectomy, the CFSE kinetics were drastically modified, with higher levels of CFSE-positive B cells being detected at 2 days (56.7 and 69.4% compared to 20.8 and 31.4% in control and BLV-infected sheep, respectively). More importantly, the percentages of B⁺ CFSE⁺ cells measured at 5 weeks were similar in both splenectomized animals (10.2% in the control and 9.8% in the BLV-infected sheep) (Fig. 1), while they were different in the unsplenectomized animals (10.1 and 3.1% in the control and in BLV-infected sheep, respectively).

To further extend the significance of this observation, we determined the kinetics of CFSE-positive B lymphocytes in a series of eight splenectomized sheep and in eight nonoperated animals (Fig. 2). In each category, three sheep were infected with a wild-type BLV provirus, whereas three others were used as negative controls. Two additional sheep included in the study were persistently infected with a BLV strain harboring a mutation in the G4 accessory gene (51, 52). This type of mutant is considered to be attenuated because it replicates poorly and is not pathogenic in sheep (21). All eight nonoperated sheep yielded similar B-cell kinetic profiles shortly after CFSE injection (at 5 days) (Fig. 2A). In the long term (Fig. 2B), a clear difference in dynamics appeared between the controls and the sheep infected with the wild-type provirus, confirming and extending our previous results pertaining to the increased cell turnover associated with BLV infection (10). Interestingly, the CFSE kinetics of B cells in sheep infected with the attenuated G4 mutant were similar to those seen for the controls (Fig. 2B).

FIG. 2.

CFSE kinetics of B cells in BLV-infected and control sheep. The left and right graphs correspond to the percentages of B cells labeled with CFSE within the B-lymphocyte population measured in short-term (5 days) and long-term (83 days) time intervals, respectively. Nonsplenectomized (A and B) and splenectomized (C and D) sheep were analyzed. CFSE kinetics were performed on sheep infected with wild-type BLV (no. 2091 [▪], 3002 [▬], 4270 [⧫], 4535 [▴], and 4536 [•] solid lines) or with an attenuated G4 mutant (no. 1071 [▬], 1077 [□], 4537 [⧫], and 4538 [+] dashed lines). Uninfected control animals (dotted lines) were 2147 (□), 3004 (×), 4164 (⋄), 4533 (▵), and 4534 (○). Thick arrows label day 27, when the percentages of CFSE-positive B cells of nonsplenectomized BLV-infected sheep reach levels of <5%. Data from unsplenectomized sheep were previously published in reference 10 and are replotted in this work.

Splenectomy had a dramatic effect on the CFSE kinetics; the percentages of labeled B cells remained high (above 60%) the first day after CFSE injection (Fig. 2, compare panels C and A). Therefore, removal of the spleen restricts the initial drop of CFSE-labeled B cells, most likely due to dilution by unlabeled splenic B lymphocytes (see Discussion). Over a longer time period, the percentages of labeled B cells progressively decreased in all sheep in a manner independent of their BLV infection status (Fig. 2D). Indeed, the CFSE profile in sheep infected with the wild-type virus was completely modified after splenectomy and overlapped that of the controls. The difference in the CFSE kinetics of B lymphocytes between BLV-infected and control animals is thus abrogated after splenectomy (Fig. 2B and D). We conclude that spleen removal profoundly affects the CFSE kinetics of B lymphocytes in BLV-infected sheep and leads to a profile characteristic of splenectomized control animals.

The CD11b-positive B-lymphocyte subpopulation accounts for the difference in CFSE kinetics of B cells in BLV-infected sheep.

BLV preferentially replicates in CD11b-positive B lymphocytes, although cells negative for this integrin receptor are also less efficient targets for the virus (7). Furthermore, the peculiar CFSE kinetics observed in BLV-infected sheep pertains to the CD11b⁺ phenotype (10). To test whether splenectomy was associated with a modification of this particular cell phenotype, we performed flow cytometry analyses to dissociate CD11b⁺ and CD11b⁻ CFSE-labeled B-lymphocyte subsets. The kinetics of CD11b-negative B lymphocytes were similar between BLV and control samples (Fig. 3A). In contrast, dissociation of the B-lymphocyte populations based on CD11b expression clearly showed that B⁺ CD11b⁺ cells are responsible for the difference in CFSE kinetics of B lymphocytes observed for BLV-infected sheep, particularly at day 27 (Fig. 3B), confirming and extending our previous report (10). After splenectomy, the CFSE profiles mostly overlapped in all sheep, independently of viral infection (Fig. 3C and D).

FIG. 3.

Differential CFSE kinetics pertains to the CD11b⁺ population. CFSE kinetics were performed on sheep infected with wild-type BLV (no. 2091 [▪], 3002 [▬], 4270 [⧫], 4535 [▴], and 4536 [•] solid lines) or with an attenuated G4 mutant (1071 [▬], 1077 [□], 4537 [⧫], and 4538 [+] dashed lines). Uninfected control animals (dotted lines) were 2147 (□), 3004 (×), 4164 (⋄), 4533 (▵), and 4534 (○). After intravenous CFSE injection in nonsplenectomized (A and B) and splenectomized (C and D) sheep, PBMCs were collected at three selected days and labeled with IgM and CD11b antibodies. Labeled cells were then analyzed by flow cytometry on the basis of 50,000 events. The percentages of cells labeled with CFSE within the B⁺ CD11b⁻ (panels A and C) and B⁺ CD11b⁺ (panels B and D) subpopulations are indicated. Thick arrows label day 27, when the percentages of CFSE-positive B cells of nonsplenectomized BLV-infected sheep reach levels of <5%. Data from unsplenectomized sheep were previously published in reference 10 and are replotted in this work.

CFSE kinetics were also evaluated in other B-cell subpopulations expressing L-selectin, CD21, or CD5. These markers were selected primarily because the migration competence of B lymphocytes correlates with the surface expression of L-selectin and CD21, mediating subsequent integrin-dependent adherence, or because CD5 is present on BLV-infected cells, at least at defined stages of BLV infection (16, 33, 55). For these three markers, however, the CFSE profiles were similar between the positive and negative subpopulations, and no difference between splenectomized and infected sheep and control sheep was observed (Fig. 4). We conclude that, unlike the expression of L-selectin, CD21, or CD5, the expression of the CD11b integrin accounts for the particular CFSE kinetics of B cells in infected sheep and that splenectomy abrogates this alternate dynamic profile.

FIG. 4.

CFSE kinetics of B-cell subpopulations in splenectomized sheep. CFSE kinetics analyses were performed on splenectomized sheep infected with wild-type BLV (no. 3002 [▬], 4270 [⧫], and 4535 [▴] solid lines) or with an attenuated G4 mutant (1077 [□] and 4538 [+] dashed lines). Uninfected control animals (dotted lines) were 3004 (×), 4164 (⋄), and 4534 (○). After intravenous CFSE injection in sheep, PBMCs were collected at three selected days and labeled with IgM and L-selectin, CD21- or CD5-specific antibodies. Labeled cells were then analyzed by flow cytometry on basis of 50,000 events. The percentages of cells labeled with CFSE within different subpopulations of B lymphocytes are indicated. (A) B⁺ L-selectin⁻; (B) B⁺ L-selectin⁺; (C) B⁺ CD21⁻; (D) B⁺ CD21⁺; (E) B⁺ CD5⁻; (F) B⁺ CD5⁻.

Splenectomy does not significantly alter the proviral loads.

As a major lymphoid tissue, the spleen is involved in the immune control of viral infections (42). Although the numbers of viral copies measured in the splenic artery and those measured in the splenic vein are similar (Table 1), it remained possible that splenectomy affected proviral loads over a longer time period. Therefore, real-time PCRs were performed in splenectomized and nonoperated sheep at days 0, 14, and 55 of the CFSE kinetics assay. It appeared that sheep infected with the G4 mutant displayed lower proviral loads than those infected with the wild-type virus, confirming the attenuated phenotype of the G4 mutant virus (the P value was 0.02 according to the unpaired Student t test) (Fig. 5). Importantly, the proviral loads, expressed as viral copy numbers per 100 cells, were not modified by splenectomy in animals infected either with a G4 mutant or with a wild-type virus (P values were 0.07 and 0.53, respectively, according to the unpaired Student t test). It is important to note that sheep 4535 is the only animal shown in Fig. 5 for which proviral loads were measured both before and after splenectomy. To determine whether proviral loads within a given animal were affected by splenectomy, the number of viral copies per cell was determined for three sheep 2 months before and 2 months after the operation. It appeared that splenectomy did not modify significantly the proviral loads (P values were 0.75, 0.72, and 0.64 according to the paired Student t test).

FIG. 5.

Proviral loads (represented as numbers of copies per 100 cells) were measured by real-time PCR and normalized with the amplification of 18S ribosomal DNA. Analyses were performed at days 0, 14, and 55 after CFSE injection. The standard deviations of individual values reflect the variability of the PCR quantification (three repetitions). Proviral loads were quantified in sheep infected with wild-type BLV (no. 2091, 3002, 4270, 4535, and 4536) or with the G4 mutant strain (no. 1071, 1077, 4537, and 4538). For each animal, the three successive values correspond to the proviral loads determined at days 0 (•), 14 (▴), and 55 (▪).

The difference between cell death rates for infected and control sheep is abrogated after splenectomy.

To quantify the death and proliferation parameters (respectively, the rates at which cells disappear or proliferate) after splenectomy, we used a mathematical model that we described previously (2) and that we briefly explain below. A reduction in the proportion of CFSE-positive B lymphocytes in peripheral blood can be due to four different factors or parameters: cell death (parameter #1), label dilution below the threshold of detection due to proliferation (parameter #2), de novo production of lymphocytes (bone marrow, Peyer's patches) (parameter #3), and recirculation of unlabeled lymphocytes from the lymph system into the bloodstream (parameter #4). Taking into account that equilibrium between blood and lymphoid organs is achieved by 2 days after CFSE injection (i.e., the exit of unlabeled cells from lymph to peripheral blood compartment no longer significantly contributes to the drop of CFSE-labeled cells after this period) (2), analyses of data after this time allowed us to disregard parameter #4. Moreover, considering that the size of the peripheral blood B-lymphocyte population is constant over the 3-month duration of the experiment (as observed by cell enumeration [data not shown]), parameter #3 is constrained to balance the difference between death (parameter #1) and proliferation (parameter #2) rates. Our mathematical model requires two types of flow cytometry experimental data: I (the ratio of the mean fluorescence intensities of B⁺ CFSE⁺ and B⁺ CFSE⁻ populations, assumed to halve upon cell division) and P (the proportion of B lymphocytes labeled with CFSE). Obtaining these data allows p (the proliferation rate) and d (the death rate) to be estimated (Fig. 6). In nonsplenectomized sheep (Fig. 6A), the death rates were significantly higher in infected sheep than in the controls (the P value was 0.005 according to the two-tailed unpaired Student t test), whereas the mean proliferation rates were unaffected. In contrast, animals infected with the G4 attenuated virus displayed proliferation and death rates similar to those observed for control sheep (P values were 0.19 and 0.42, respectively, according to the two-tailed unpaired Student t test). We then compared the death and proliferation rates for all eight splenectomized and nonoperated sheep (Fig. 6A and B). Remarkably, the only statistically significant difference was a decrease of the death rate for splenectomized sheep infected with wild-type BLV (the P value was 0.04 according to the two-tailed unpaired Student t test). As expected, all kinetic parameters were, after splenectomy, in the range of the controls (not statistically significant).

FIG. 6.

Summary of the cell kinetic parameters determined for control sheep (noninfected) and for animals infected with a wild-type BLV provirus or an attenuated strain (G4 mutant) as indicated. Minimal proliferation and death rates were estimated by fitting a theoretical model with two types of flow cytometry data: the percentages of B lymphocytes labeled with CFSE and their relative fluorescence intensities. The proliferation and death rates correspond to the percentages of cells produced by proliferation and those disappearing per day, respectively. Statistical relevance (NS, not statistically significant; **, highly statistically significant) among the death rate parameters was calculated according to the two-tailed unpaired Student t test.

Together, our data demonstrate that increased B-cell death observed in BLV-infected sheep is abrogated after spleen removal.

DISCUSSION

Combining functions relating to innate immunity and adaptive immunity, the spleen also contributes to a variety of roles as diverse as the removal of old erythrocytes from the circulation and the destruction of blood-borne microorganisms or cellular debris (31). The spleen contains two structurally distinct compartments: the blood-containing red pulp, in which pathogens and cellular debris, as well as ageing erythrocytes, are efficiently removed from the blood by macrophages; and the white pulp, a highly organized lymphoid region where adaptive immune responses can be initiated (31, 56). Considering these crucial activities, the spleen is surprisingly dispensable; splenectomy has little impact on survival. However, splenectomized individuals (especially young children) are particularly vulnerable to sepsis caused by encapsulated bacteria (3, 43, 56), essentially because of an inability to respond to T-independent antigens (4, 23, 41). Interestingly, the role of splenectomy in cancer development and therapy has been a matter of intensive debate, since it improves survival in some cases (13, 15, 22) and induces detrimental effects in others (11, 19, 39).

The spleen also plays a major role in the immune surveillance of viral infections, but the net outcome of this process remains unclear. The entry of activated dendritic cells or marginal-zone B cells into the white pulp can initiate an antiviral immune response through the activation of T cells. For instance, spleen cells from rats experimentally infected with HTLV-1 exhibit a strong Th response specific for HTLV-1 (25). Several viruses can directly interact with the c-type lectin SIGNR1, expressed by marginal-zone macrophages (28, 31). Depending on the system, the result of this complex interplay can thus be beneficial (20, 24, 45), neutral (14, 26, 36), or prejudicial (8, 17, 48) to viral persistence and/or pathogenesis. In this context, the effect of splenectomy in HTLV-infected patients, although not clearly defined, has been used to palliate hypersplenism (40) and immune thrombocytopenia (30).

In this report, we aimed at understanding the involvement of the spleen during BLV infection of sheep. In this experimental model of leukemia, the peripheral blood B-cell population is prone to higher death rates (9, 10). We demonstrate here that this process in BLV-infected sheep is strictly dependent on the integrity of the spleen. We found only scarce information on this topic in the literature, although leukemic cells have long been known to infiltrate the spleen, potentially provoking splenomegaly in BLV-infected cattle (49). A single paper reports that splenectomy in calves affects neither the establishment of BLV infection nor the development of a serological response to viral antigens over a 5-week period (46). We now show that the cell dynamics is however dramatically modified in splenectomized BLV-infected sheep.

Analysis of the kinetic profiles reveals that splenectomy restricts the initial drop of CFSE-labeled B cells. Therefore, it is likely that the majority of early labeled cell loss in nonsplenectomized animals is due to the exit of unlabeled cells from the spleen. To further strengthen this assumption, we serially collected blood from the splenic and jugular veins after CFSE labeling of a BLV-infected sheep. It appeared that over a period of 2 h after intravenous CFSE injection, the percentages of B lymphocytes labeled with CFSE increased from 20 to 45% in the splenic vein, while these levels dropped from 99% to 60% in the jugular vein (data not shown). This kinetic profile is consistent with a fast equilibration between unlabeled B lymphocytes massively exiting the spleen and fluorescent B lymphocytes from the peripheral blood (55), leading to a global drop in the proportions of CFSE-labeled cells. We previously estimated the contribution of splenic B lymphocytes to this early labeled dilution at about 76% (2).

Despite the massive entry of spleen-derived B lymphocytes into the peripheral blood, the phenotypes of cells in the splenic artery and vein were indistinguishable in terms of CD11b expression, proviral load, and ability to synthesize the p24 antigen ex vivo (Table 1). At first sight, this observation would suggest that the spleen does not significantly contribute to the elimination of virus-infected cells. However, since about 5% of the total blood volume transits through the spleen every minute (28) and since the global death rate of the peripheral B-cell population in BLV-infected sheep is estimated at 11.9% per day (10), it retrospectively would have been surprising to have revealed a massive destruction of BLV-infected cells over such a very short time period.

In contrast, the long-term CFSE kinetics of B cells in splenectomized sheep uncovers an essential role exerted by the spleen (Fig. 2). Indeed, our data demonstrate that increased B-cell death occurring in BLV-infected sheep is abrogated after splenectomy (Fig. 6). The most likely interpretation of these results is that the spleen efficiently controls viral infection by triggering the elimination of infected cells. However, the drawback of this surgical approach is that it might perturb several physiologic processes, such as an increase in the absolute cell numbers in the peripheral blood (28), a modification of the efficiency of lymphocyte recirculation (32, 37), and the induction of immunosuppression (53). With our splenectomized sheep, we did observe an augmentation of the absolute cell counts (typically 1.5- to 2-fold), but this process equally affected all lymphocyte subpopulations (CD4⁺, CD8⁺, Tγδ⁺, and B cells, as well as the proportion of infected cells [Fig. 5 and data not shown]). Similar observations have been reported in reference 41. These observations thus indicate that the spleen is required to control cell homeostasis in the peripheral blood in a manner independent of viral infection. Besides, splenectomy could affect lymphocyte recirculation and modify blood residency time. Although the recirculation of peripheral blood lymphocytes in sheep is not affected by splenectomy (41), a definitive answer to this point would require cannulation of the efferent lymphatic ducts of splenectomized BLV-infected animals. Finally, surgery by itself could potentially lead to immunosuppression and impair efficient cell clearance in infected animals (53). We think that this process is unlikely to occur, because other types of surgical operations, such as the cannulation of mesenteric lymphatic ducts, led neither to a perturbation in CFSE kinetics nor to detectable immunosuppression. It thus appears that the absence of splenic control is compensated by other lymphoid tissues, such as lymph nodes, ileal Peyer's patches, and perhaps accessory spleens (3, 28, 47).

Alteration of the cell dynamics in BLV-infected splenectomized sheep associated with a failure of immune surveillance could potentially accelerate leukemogenesis. Indeed, a decrease in the cell death rates without a compensatory reduction of proliferation would lead to cell accumulation. We did not observe the onset of leukemia during the weeks after splenectomy, suggesting the existence of regulatory mechanisms such as decreased cell production/differentiation or accumulation in other organs. However, a splenectomized sheep developed leukemia as early as 132 days after infection (90 days after the surgery), a very short latency period never observed before (1, 21, 29). Indeed, the earliest onset of neoplastic transformation among 45 nonsplenectomized sheep infected by the same BLV proviral clone (strain 344) over a period of 15 years was 448 days. Further long-term follow-up of significant numbers of animals infected with wild-type and mutant viruses is needed to confirm this preliminary but intriguing observation.

Together, our results thus clearly reveal the central function of the spleen in the regulation of the particular cell dynamics pertaining to BLV-infected sheep. Although the precise mechanisms remain to be characterized, our working hypothesis is based on a process of immune surveillance relying on the integrity of the spleen. Indeed, it is generally agreed that an activated cytotoxic and humoral immune response is directed towards the virus and that this surveillance counteracts the accumulation of infected cells (reviewed in reference 12). In this recent paper, we proposed a unifying model of viral persistence and spread in which the virus would play an active role by continuously expressing viral proteins, such as Tax, promoting cell proliferation. Tax-expressing cells would therefore harbor a selective growth advantage but would concomitantly become detectable by the host immune response. Our results indicate that the cell disappearance rates in infected sheep decrease after splenectomy, most probably reflecting a reduction in the efficiency of the antiviral response.