Lymphocyte populations in human lymph nodes. Alterations in CD4⁺ CD25⁺ T regulatory cell phenotype and T-cell receptor Vβ repertoire

Abstract

Here we provide a description of lymphocyte populations in human lymph nodes (LN) with a special emphasis on the CD4⁺ lymphocyte population constitutively expressing CD25 at a high level and endowed with immunoregulatory properties [T regulatory (Treg) cells]. Lymph nodes were analysed by multicolour flow cytometry in parallel with correspondent peripheral blood (PB). Immunomagnetically purified Treg cells were tested for anergy and suppressive activity in a CD3/T-cell receptor (TCR)-driven proliferation assay. Compared to PB, there was a reduced T/B lymphocyte ratio in LN. Both LN and PB contained a similar proportion of CD4⁺ lymphocytes but, conversely, CD8⁺ lymphocytes were less represented in PB, with a consequent increase in the ratio of CD4⁺/CD8⁺ natural killer cells were < 2% (PB range 6–22%). No significant differences existed in the frequency of the other lymphocyte subpopulations examined (naïve-type CD4⁺ and CD8⁺ lymphocytes, activated B and CD4⁺ lymphocytes, and effector-type CD8⁺ lymphocytes). LN and PB contained similar percentages of CD4⁺ lymphocytes constitutively expressing intermediate or high levels of CD25. CD4⁺ CD25⁺⁺ cells constitutively coexpressed high levels of CD152 and were therefore identified as Treg cells. Treg cells in LN and PB differed in terms of CD45RB, HLA-DR, CD45RO, and CD62L expression. Also the TCRVβ repertoire diverged between Treg cells from LN and PB. Similar to Treg cells from PB, Treg cells from LN were anergic and efficiently inhibited other CD4⁺ and CD8⁺ lymphocyte proliferation. This study extends the information on the diversities in lymphocyte composition between human LN and PB, and reports for the first time a description of the phenotypic and functional characteristics of Treg cells in human LN, highlighting the importance of the LN microenvironment in shaping the surface phenotype of Treg cells.

Introduction

Lymphocytes are mobile cells, continuously recirculating between the blood and tissues, returning to the blood via the lymphatic system. It has been estimated that peripheral blood (PB) contains approximately 2% of all lymphocytes, the vast majority being found in lymphoid organs, namely the tonsils, spleen, young thymus and, especially, lymph nodes (LN).^1,2 Transit through LN interspersed among the lymphatics is considered a crucial event in the lymphocyte's life, as it is in these specialized sites that lymphocytes encounter antigens in association with antigen-presenting cells and become functionally competent. Consistent with the role the LN microenvironment has in governing lymphocyte maturation, during the last several years numerous reports have described various phenotypical and functional differences between lymphocyte populations from LN and PB, indicating that the two compartments are distinct in terms of lymphoid cell composition.^3–13 Unfortunately, most of the earlier studies used samples from patients affected by a plethora of different pathological situations and/or assessed lymphocyte populations in LN by using a relatively limited combination of relevant surface markers. Moreover, even when multicolour flow cytometry has been used, only a few of the relevant lymphocyte subsets involved in regulatory and effector functions have been investigated. In particular, there is only scanty information on the recently described CD4⁺ lymphocyte population constitutively expressing the interleukin-2 receptor α-chain CD25 and commonly referred to as T regulatory (Treg) cells, which are responsible for peripheral tolerance.^14–17 These lymphocytes express the typical αβ T-cell receptor (TCR) and develop during normal T-cell development as well as during the peripheral T-lymphocyte response to antigens under the influence of cytokines and costimulatory milieu. Animal studies have revealed that this lymphocyte subset is pivotal in controlling autoimmunity, tumour immunity and transplantation tolerance,^14–17 raising considerable interest in studying Treg cells in humans. In fact, the presence of this lymphocyte population has been described in several human tissues, namely, adult PB, cord blood, tonsils and thymus, and even in the tumour microenvironment and tumour draining LN of some cancer patients,^18–22 but not yet in normal LN.

With this background, the current study was undertaken to determine the distribution of diverse lymphocyte populations in normal human LN in comparison with correspondent PB, with a special emphasis placed on the assessment of the phenotypic profile and functional activity of Treg cells. The present findings extend early reports on the differences existing between the two compartments in terms of lymphocyte composition, and show for the first time that Treg cells can be found in normal human LN. Treg cells from the two compartments differ in terms of surface phenotype and TCRVβ usage. Functionally, the Treg cells from LN resemble their counterparts in PB as they are anergic and endowed with a potent suppressive activity on the response of other lymphocytes.

Materials and methods

Patients and surgery

The study was approved by the ethical committee of the Catholic University. Written informed consent was obtained from all patients. Because the study of human LN from healthy individuals is limited by ethical considerations, we used four LN from patients with benign diseases (aged 32–70 years) undergoing lymphadenectomy as a diagnostic procedure, and 23 LN from 18 women (aged 38–74 years) with early stage cervical or endometrial cancer. All cancer patients underwent lymphadenectomy as part of primary surgical treatment, were free of nodal and haematogenous metastases and did not receive any medications before surgery. Upper pelvic, i.e. common iliac, presacral and internal iliac, and para-aortic LN were used for the study as they were distant from the location of the primary tumour. We considered these LN as the closest approximation to normality that can be obtained for studies of this type. PB was collected 1–2 days before surgery.

Obtaining cells from the LN and PB

Para-aortic and common iliac LN were used for the present study as they were distant from the tumour site and were therefore considered to remain uninfluenced by tumour products through the afferent lymphatics. Mononuclear cell suspension was obtained immediately after surgery by mechanically disaggregating the sample using a scalpel and needle followed by syringing through a 22-gauge needle. All steps were performed in sterile phosphate-buffered saline (PBS) containing 0·1% bovine serum albumin (BSA). A representative part of each LN was sent for routine histology. The cell suspension was washed twice in cold PBS–BSA, resuspended at 10⁶ cells/ml and used for immunostaining. Lymphocytes from PB were obtained as mononuclear cells by standard density gradient centrifugation of heparinized blood, as described previously.²³

Immunostaining

Optimal concentrations of monoclonal antibodies (mAb) were determined for each mAb by titration. Samples were examined immediately after staining without fixation to take the best advantage of light scatter signals in excluding non-viable cells. We used mAb to: CD25, CD45RA, CD45RO, CD56, CD62L, CD134 and HLA-DR (Becton Dickinson Biosciences, San Jose, CA); CD3 and CD25 (Serotec Ltd, Oxford, UK); CD25 (clone M-A251), CD27, CD45RB, CD80, and CD152 (originally referred to as cytotoxic-T-lymphocyte-associated antigen 4, CTLA-4) (Pharmingen, San Diego, CA); CD4, CD8 and CD20 (BIO-D, Bari, Italy). The mAb were labelled with the fluorescent dyes fluorescein isothiocyanate (FITC), phycoerythrin (PE) and peridinin chlorophyll protein (PerCP) and appropriately combined to assess relevant lymphocyte subsets. The Vβ usage was assessed by the mAb included in the Human αβ TCR Screening Panel (Endogen, Woburn, MA) Vβ3.1, Vβ5.2, Vβ5.3, Vβ5.1, Vβ6.7, Vβ8, Vβ12, Vβ13.1 and Vβ13.3. All Vβ-specific mAb were FITC conjugated. Expression of CD152 had to be determined by intracytoplasmic staining.¹⁸ To this end, cells were stained with PerCP-CD4 and FITC-CD25 mAb, washed and then fixed and permeabilized using the FIX and PERM cell permeabilization kit (Caltag Laboratories, Burlingame, CA) according to the manufacturer's instructions prior to being reacted with PE-CD152. Cells were measured by a FACScan flow cytometer (Becton Dickinson) using forward- and side-scatter signals to establish the lymphocyte gate and exclude unwanted events (i.e. not viable cells, debris and cell clumps) from cell evaluation. Fluorescence signals were collected in log mode. A minimum of 5000 cells of interest were acquired for each sample.

Functional assay

Purification of Treg for functional tests was performed as follows. CD4⁺ lymphocytes were purified as untouched cells by negative selection using the CD4⁺ T cell Isolation Kit from Miltenyi (Miltenyi Biotec, GmbH, Bergisch Gladbach, Germany) that contains CD8, CD11b, CD16, CD56, CD19 and CD36 mAbs. Microbeads directly coated with anti-CD25 mAb (Miltenyi Biotec) were then used to separate CD25⁺ from CD25⁻ cells. All steps were performed following the manufacturer's instructions. Purity was assessed by flow cytometry and found consistently to exceed 90% and 80% in CD4⁺ CD25⁻ and Treg preparations, respectively. For the latter, FITC-CD25 clone M-A251 (Pharmingen) was used as it recognizes an epitope not masked by the anti-CD25 mAb used in the purification step. Because immunomagnetic sorting does not allow a separation based on the intensity of staining, the suppressive cell population used for functional studies necessarily contains both CD4⁺ CD25⁺ and CD4⁺ CD25⁺⁺ cells. Thus, in all functional assays the term ‘Treg cells’ indicates a mixture of CD4⁺ CD25⁺ and CD4⁺ CD25⁺⁺ cells (typically 7 : 1). To analyse proliferation in response to polyclonal activation, CD4⁺ CD25⁻ and Treg cells (10 000 cells/well) were seeded in flat-bottomed black microtitre plates (Packard, Meridien, MO) precoated overnight with 1 μg/ml anti-CD3 mAb (clone UCHT1, Serotec Ltd, Oxford, UK) or with a mixture of anti-CD3 and anti-CD28 (clone YTH913.12 1 μg/ml, Serotec), and incubated at 37° in a 5% CO₂ atmosphere for 5 days. Plates were then harvested and cell proliferation was estimated using the ATPlite kit (Packard) and the automated luminometer Topcount (Packard), following the manufacturer's suggestions. Experiments were also designed to assess the regulatory capacity of Treg cells on CD4⁺ and CD8⁺ lymphocytes. To this end, we used an experimental approach involving the whole cell population as responder cells and the intracellular covalent coupling dye carboxyfluorescein diacetate succinimidyl ester (CFSE) (also referred to as CFDA-SE; Molecular Probes, Eugene, OR). This approach is less cumbersome to perform than those methods requiring various purification steps of the responder cells as it allows the assessment of the inhibitory activity on CD4⁺ and CD8⁺ lymphocytes in the same culture by a simple multicolour flow cytometric analysis.²⁴ Thus, mononuclear cells were loaded aseptically with CFSE as detailed elsewhere²⁵ and incubated with decreasing numbers of Treg cells (responder to /suppressor ratio ∼1 : 1–10 : 1). At the end of culture time, cultures were costained with PE-CD8 and PerCP-CD4 mAbs to quantify the number of cell divisions in relationship to the expression of the membrane marker by modfit™/cell proliferation model™ software (Sigma St Louis, MO).

Statistics

Data on the lymphocyte populations are presented as mean values ± SD. Analysis of variance (anova) or repeated measures anova followed by Tukey's test for post hoc comparison was used to assess the significance of changes in the proportions of the various populations in LN as compared to PB. For testing the association between expression of distinct TCRVβ in the CD4⁺ subsets identified by the variable expression of CD25, i.e. CD25⁻, CD25⁺ and CD25⁺⁺, Pearson's linear regression analysis was performed using the graphpad prism™ software. The presence of a significant correlation was taken as an indicator that the TCRVβ repertoire was similar. Conversely, the lack of a significant correlation was taken as an indicator of a different TCRVβ usage.²⁶

Results

Lymphocyte populations in LN and PB

The distribution of lymphocyte populations evaluated as the frequency of each population in the lymphocyte gate in LN and correspondent PB is summarized in Table 1. Significant differences were observed in the distribution of the bulk of T- and B-lymphocyte populations, with a clear prevalence of the latter and a corresponding decrease in the former. Consequently, the T/B lymphocyte ratio was markedly lower in LN than PB (Table 1). The proportion of activated B lymphocytes, assayed by the expression of CD80, was significantly higher in LN than PB (Table 1). Interestingly, a similar observation has already been reported and interpreted as reflecting the capacity of the LN microenvironment to regulate the capacity of B lymphocytes to act as antigen-presenting cells.¹² However, there was no difference in the percentage of CD80⁺ cells within the B-lymphocyte population (30 ± 13·7 versus 29 ± 12·4, LN and PB, respectively, not shown). This indicates that the increased amount of CD80⁺ B lymphocytes merely reflects the increased presence of the bulk B-lymphocyte population in LN (Table 1) and argues against the view that the LN microenvironment is selecting/expanding B lymphocytes endowed with antigen-presenting activity.¹² The percentage of CD4⁺ lymphocytes in LN and PB was similar, while the percentage of CD8⁺ lymphocytes was markedly lower in LN (Table 1). Consequently, the CD4/CD8 ratio in LN was almost twice that of PB (Table 1). No significant differences between LN and PB were observed in the frequency of activated CD4 lymphocytes, defined as CD4⁺ CD134⁺ cells²⁷ (3 ± 1·7 versus 3 ± 0·53, LN and PB, respectively, not shown) and naïve-type CD4⁺ lymphocytes defined as CD4⁺ CD45RA⁺ CD62L⁺ cells²⁸ (9 ± 4·3 versus 11 ± 2·3 in LN and PB, respectively, not shown). The proportion of naïve-type CD8⁺ lymphocytes, defined as CD45RA⁺ CD62L⁺²⁸ tended to be lower in LN (5 ± 3·8 versus 8 ± 2·3 LN and PB, respectively, not shown), probably to reflect the lower frequency of CD8⁺ lymphocytes in this compartment. Effector-type CD8⁺ lymphocytes, defined as CD27⁻ cells²⁹ were significantly less represented in LN than PB (Table 1). However, this difference was the result of the higher frequency of CD8⁺ lymphocytes in PB, the relative proportions of CD27⁻ cells within the CD8⁺ population being identical (10 ± 6·7 versus 10 ± 5·7, LN and PB, respectively, not shown). Natural killer (NK; CD3⁻ CD56⁺) cells were always significantly less represented in LN (<2% in all samples) (Table 1). Both LN and PB contained consistent proportions of CD4⁺ lymphocytes coexpressing CD25 at different intensities (Fig. 1) and thereafter referred to as CD4⁺ CD25⁺ and CD4⁺ CD25⁺⁺ cells (Table 1). The current literature indicates that in humans a constitutively high level of expression of CD25 and CD152 distinguishes Treg cells from CD4⁺ lymphocytes carrying intermediate levels of CD25 and CD152 as a result of a recent activation.³⁰ As depicted in Fig. 2, all CD4⁺ CD25⁺⁺ cells stained for CD152, thereby allowing the identification of these cells as Treg cells.^18,30 However, while CD4⁺ CD25⁻ cells completely failed to stain for CD152, a small proportion of CD4⁺ CD25⁺ cells did express the antigen at low density in all samples examined (Fig. 2). The relative proportions of CD4⁺ CD25⁺ and CD4⁺ CD25⁺⁺ did not differ significantly between LN and PB, although in the former a somewhat reduced percentage of CD4⁺ CD25⁺ and enhanced percentage of CD4⁺ CD25⁺⁺ cells was noticed (Table 1). The frequencies of the various lymphocyte populations in the few LN from patients with benign disorders available for analysis (CD3⁺ 57 ± 18·2, CD19⁺ 34 ± 17·2, ratio CD3⁺/CD19⁺ 1·7 ± 0·92, CD19⁺ CD80⁺ 8 ± 6·11, CD4⁺ 46 ± 15·5, CD8⁺ 11 ± 5·7, ratio CD4⁺/CD8⁺ 4·6 ± 1·82, CD8⁺ CD27⁻ 2 ± 1·9, NK cells 2 ± 0·5, CD4⁺ CD25⁻ 21 ± 7·1, CD4⁺ CD25⁺ 21 ± 14·5, CD4⁺ CD25⁺⁺ 3 ± 0·3) matched well with those of cancer patients, confirming that lymphocyte composition in the present series of LN was not significantly influenced by the underlying malignant pathology.

Table 1.

Lymphocyte populations (means ± SD) in lymph node (LN) and correspondent peripheral blood (PB)

	LN	PB
CD3+	56 ± 13·3*	74 ± 9·7
CD19+	41 ± 15·1*	12 ± 5·5
CD3+/CD19+ ratio	1·6 ± 0·83*	6·1 ± 3·60
CD19+ CD80+	11 ± 7·5*	0·8 ± 0·05
CD4+	48 ± 12·8	51 ± 8·8
CD8+	10 ± 3·3*	27 ± 11·6
CD4+/CD8+ ratio	4·5 ± 1·38*	2·6 ± 1·71
CD8+ CD27−	2 ± 1·1*	14 ± 6·4
CD3− CD56+	1 ± 0·4*	11 ± 4·5
CD4+ CD25−	22 ± 6·5	17 ± 3·8
CD4+ CD25+	23 ± 7·9	32 ± 6·4
CD4+ CD25++	3 ± 1·4	2 ± 1·0

^*P < 0·05 vs. correspondent population in PB (one-way anova).

Figure 1.

Representative dual colour dot plot displaying the coexpression of CD4 and CD25 in a typical LN sample. The electronic regions used to measure the relative proportions of CD4⁺ cells with different levels of expression of CD25, i.e. CD25⁻, CD25⁺ and CD25⁺⁺, R1, R2 and R3, respectively, are shown. These regions were also used to refer the coexpression of the other phenotypic markers and of the various TCRVβ to CD4⁺ CD25⁻, CD4⁺ CD25⁺ and CD4⁺ CD25⁺⁺ cells.

Figure 2.

Expression of intracytoplasmic CD152 in the three CD4⁺ subsets defined by CD25 expression (Fig. 1) in (a) LN and (b) PB. CD152 is expressed at high level by CD25⁺⁺ cells and is not expressed by CD25⁻ cells. CD25⁺ cells shows a marginal reactivity. Continuous line, discontinuous line and shaded histogram indicate CD25⁻, CD25⁺ and CD25⁺⁺ cells, respectively. Isotype control mAb and CD152 preblocked controls gave results superimposable to CD25⁻ cells and are omitted from the histograms

Phenotypical characterization of Treg cells in LN and PB

Having determined that LN and PB contained essentially similar proportions of Treg cells, we sought to determine whether the phenotypic profile of Treg cells from the two compartments diverged. To this end, a series of surface markers that reportedly characterize this lymphocyte subset in humans, namely, CD45RB^low, CD45RO, CD62L and HLA-DR, ^14–18,20 were evaluated. These antigens were assessed in relationship to the brightness of CD25 staining, i.e. in CD4⁺ CD25⁻, CD4⁺ CD25⁺ and CD4⁺ CD25⁺⁺ subsets, using the gates exemplified in Fig. 1. There was no difference in the proportions of cells expressing CD45RB^low between CD4⁺ CD25⁻, CD4⁺ CD25⁺ and CD4⁺ CD25⁺⁺ subsets in LN (Fig. 3, left panel). This was at variance with PB, where cells expressing CD45RB^low prevailed in CD4⁺ CD25⁺ and CD4⁺ CD25⁺⁺ subsets as compared to the CD4⁺ CD25⁻ subset (Fig. 3, right panel), an observation in keeping with earlier reports.^14–18,29 The results of staining with HLA-DR, CD45RO and CD62L in a typical LN and PB sample are shown in Fig. 4. The expression of these molecules was quite variable in all CD4⁺ subsets, sometimes involving the almost complete absence of cells negative for CD45RO and CD62L. Thus, data are expressed as specific fluorescence index (SFI) computed as described²⁵ and summarized in Table 2. Treg and non-Treg CD4⁺ subsets from LN exhibited a significantly higher level of expression of HLA-DR and a lower level of expression of CD45RO and CD62L as compared to the correspondent subsets in PB (Table 2). Remarkably, there was no significant difference in the expression of these antigens between the three CD4⁺ subsets within LN. Similarly, CD45RO and CD62L expression did not differ significantly between the three CD4⁺ subsets within PB, an observation at variance with previous reports. ^18,20,21 However, in this compartment only CD4⁺ CD25⁺⁺ cells expressed HLA-DR (Table 2), in line with early findings.^14–18,20 It is possible that the technical complexity in comparing markers expressed over a large range of intensities using mAb panels that differ among laboratories in terms of fine specificity and fluorochrome labelling may be responsible for these discrepancies.

Figure 3.

Scatter plot illustrating the proportions of CD45RB^low cells in the three subsets of CD4⁺ cells defined by CD25 expression (Fig. 1). In LN, there is no significant difference in the proportion of cells expressing CD45RB^low between CD4⁺ CD25⁻, CD4⁺ CD25⁺ and CD4⁺ CD25⁺⁺ subsets, whereas in PB, the proportion of CD4⁺ CD25⁺ and of CD4⁺ CD25⁺⁺ cells coexpressing CD45RB^low is significantly higher than in CD4⁺ CD25⁻ (repeated measures anova and Tukey's test). Horizontal lines indicated mean values. There was no significant difference between LN and PB in terms of CD45RB^low expression, although a trend toward an increased presence of CD45RB^low cells in CD4⁺ CD25⁺⁺ cells in PB as compared to the correspondent subset in LN could be observed (one-way anova and Tukey's test).

Figure 4.

Expression of HLA-DR, CD45RO and CD62L in the three CD4⁺ subsets with different CD25 expression (Fig. 1). LN and PB data are depicted as thin line and filled histograms, respectively. Histograms are representative of at least four stainings per antigen and all the mAbs used here were PE-conjugated. The unimodal distribution of all the antigens made it difficult to set unambiguously a boundary between positive and negative cells and data were therefore collectively evaluated as SFI (Table 2).

Table 2.

Phenotypic profile of CD4⁺ lymphocytes with different CD25 expression levels in LN and correspondent PB

	LN			PB
	CD25−	CD25+	CD25++	CD25−	CD25+	CD25++
HLA-DR	21·2 ± 16·9*	15·9 ± 11·0†	11·4 ± 7·6‡	0·1 ± 0·1	0·1 ± 0·1	2·4 ± 0·5§
CD45RO	21·5 ± 13·3*	19 ± 9·3†	20·4 ± 8·4‡	40·8 ± 8·6	38·9 ± 10·1	47·1 ± 17
CD62L	40·5 ± 19·4*	44 ± 21·6	45·1 ± 18‡	65·3 ± 13·7	54·6 ± 18·1	59·9 ± 13·8

Data are expressed as mean ± SD specific fluorescence index (SFI; see text for details).

^*P < 0·05 versus PB CD25⁻.

^†P < 0·05 versus PB CD25⁺.

^‡P < 0·05 versus PB CD25⁺⁺.

^§P < 0·05 versus PB CD25⁻ and CD25⁺.

*†‡One-way anova; §repeated measures anova and Tukey's test.

We next simultaneously tested for the expression of eight distinct TCRVβ families in the three CD4⁺ subsets in LN and the corresponding subsets in PB from seven individuals and, in doing so, assigned β-chain usage for 35–65% of the CD4⁺ lymphocytes in each donor. As a measure of concordance of the TCRVβ repertoire between two given samples, the correlation coefficient (Pearson's regression analysis) was calculated (r² = 1 being a perfect correlation and r² = 0 being no correlation), as described earlier.²⁶ Comparing LN and PB, we found a good correlation between the TCRVβ repertoire of CD4⁺ CD25⁻ cells (r² = 0·3855) and CD4⁺ CD25⁺ cells (r² = 0·3497) (Fig. 5a and b, respectively). In contrast, as outlined by the poor regression (r² = 0·1211) there was no association between the TCRVβ repertoire of CD4⁺ CD25⁺⁺ cells in LN and PB (Fig. 5c), indicating that Treg cells in LN and PB do not express the same TCRVβ repertoire. We then compared the TCRVβ usage by the three CD4⁺ subsets within each compartment (Fig. 6). In both compartments, correlation analysis revealed that the highest concordance was observed between CD4⁺ CD25⁻ and CD4⁺ CD25⁺ (r² = 0·7362 and 0·5587, Fig. 6a and d, respectively), CD4⁺ CD25⁺ showed an intermediate correlation with CD4⁺ CD25⁺⁺ (r² = 0·3645 and r² = 0·4861, Figs 6b,e) and CD4⁺ CD25⁺⁺ showed the lowest correlation with CD4⁺ CD25⁻ (r² = 0·2860 and r² = 0·2556, Figs 6c,f).

Figure 5.

Correlation analysis of TCRVβ expression in the three CD4⁺ subsets defined by the different CD25 expression (Fig. 1) in LN and correspondent PB. The good correlation shown by CD4⁺ CD25⁻ cells and CD4⁺ CD25⁺ cells between LN and PB (a and b, respectively, two-sided P < 0·0001) indicates a concordant usage of TCRVβ repertoire in these CD4⁺ subsets in the two compartments. Conversely, the analogous comparison made for CD4⁺ CD25⁺⁺ cells, i.e. Treg cells, indicates lack of correlation and therefore a different Vβ usage by Treg cells in the two compartments (c, two-sided P > 0·05).

Figure 6.

Correlation analysis of TCRVβ expression in the three subsets of CD4⁺ cells with different CD25 expression (Fig. 1), illustrating the relative usage of Vβ regions by the CD4⁺ CD25^−/+/++ subsets within LN (a–c) and PB (d–f). The percentage of every tested TCRVβ was plotted between the indicated subsets. The good correlation between CD4⁺ CD25⁺, CD4⁺ CD25⁺ and CD4⁺ CD25⁺⁺ subsets seen in both compartments (two-sided P < 0·001) indicates that TCRVβ distribution is not significantly skewed. However, CD4⁺ CD25⁻ and CD4⁺ CD25⁺⁺ cells show the lowest concordance in TCRVβ usage in both compartments (c,f).

Functional analysis of Treg cells in LN and PB

Two defining features of Treg cells are their anergy and their ability to inhibit the proliferation of other T-lymphocyte populations.^14–20 Anergy can be overcome by addition of anti-CD28.^18,20 As shown in Fig. 7, TCR/CD3 engagement alone was sufficient to induce proliferation of CD4⁺ CD25⁻ cells from PB and LN, and a second signal via CD28 enhanced the response. Treg cells from LN and PB were consistently hyporesponsive to TCR/CD3 activation and even coengagement of CD28 failed to restore the responsiveness. We next tested the capacity of Treg cells to suppress the proliferative response of other lymphocyte populations. To this end, Treg cells from LN were cocultured with lymphocytes from autologous PB as responder cells. Figure 8 exemplifies a TCR/CD3-driven proliferative assay in which 10⁵ responder cells were loaded with CFSE and incubated alone or in the presence of 3 × 10³ Treg cells for 5 days. CFSE halving was then separately measured in CD4⁺ and CD8⁺ responder cells. Data clearly document the suppressive capacity of Treg cells and show that suppression is equally exerted on CD4⁺ and CD8⁺ lymphocytes. Decreasing the proportions of Treg cells produced a progressively lower inhibition of the proliferative response of both CD4⁺ and CD8⁺ lymphocytes (not shown).

Figure 7.

Treg cells from LN are anergic similarly to Treg cells from PB. Isolated CD4⁺ cells were fractionated into CD4⁺ CD25⁻ and CD4⁺ CD25^+/++ cells, by immunomagnetic sorting and tested for their ability to proliferate in response to plate-bound CD3 or to plate-bound CD3 and CD28. After 5 days of incubation, cultures were stopped and assayed by ATPlite kit (Packard). Numbers on the y-axis represent counts per second (c.p.s.) as measured by Topcount (Packard). Representative of one of four similar experiments.

Figure 8.

Treg cells from LN suppress proliferation to CD3 engagement on both CD4⁺ and CD8⁺ lymphocytes. Unfractionated mononuclear cells from autologous PB were tested for their ability to proliferate in response to plate-bound CD3 in the absence (a,b) or in the presence (c,d) of immunomagnetically purified Treg cells (ratio 1 : 1). At the end of culture, cells were stained with PerCp-CD4 and PE-CD8 mAb and CFSE fluorescence histograms obtained separately for CD4⁺ (a,c) and CD8⁺ (b,d) lymphocytes. The area under each CFSE fluorescence histogram was fitted by cellcensusplus™ software to calculate the proliferation index (PI) shown in each panel. Representative of one of two similar experiments.

Discussion

In the present study, our goal was to describe the lymphocyte composition in normal human LN by making a direct comparison with the correspondent populations in PB. Because of the limitations in using samples from healthy individuals, for the present work we had to resort to LN collected during lymphadenectomy performed as a diagnostic procedure in selected patients with gynaecological cancers. That this approach represents an acceptable approximation to normal is supported by the following reasons. First, only metastasis-free LN collected at distant locations from the primary tumour site (to avoid possible influence by tumour products) were used. Second, all patients had early, localized tumours and were not receiving medication. Third, the proportions of the various lymphocyte populations in the small number of LN from subjects with benign disease matched well with those observed in LN from cancer patients. We found that the only consistent difference between LN and PB was in the relative proportions of the five major lymphocyte populations that constituted them, namely B and T lymphocytes, CD4⁺ and CD8⁺ lymphocytes and NK cells. In fact, all the observed differences regarding the various subsets within these major populations, merely reflected the differences in the proportions of these major populations. Interestingly, divergences in lymphocyte composition between LN and PB similar to those found here have been reported earlier in other than pelvic normal LN and also in LN obtained from patients affected by various diseases.^3–13 Collectively, those earlier and present findings suggest that the lymphocyte composition in LN is kept relatively constant, irrespective of the underlying pathology.

The main novel finding of this study is that human LN contain sizeable amounts of Treg cells, i.e. CD4⁺ CD25⁺⁺ CD152⁺⁺ cells. ^14–18,20 It is currently thought that T-lymphocyte-mediated control of self-reactive T lymphocytes is an important mechanism by which peripheral tolerance toward self-antigens is maintained.^14–17 In fact, a number of T-lymphocyte subpopulations capable of inhibiting the response of other T lymphocytes to self-antigens through a variety of mechanisms have been described in in vitro and animal studies.¹⁴ A complete characterization of all these suppressive populations is still lacking. However, it is becoming increasingly clear that Treg cells represent the major T lymphocyte population involved in the maintenance of peripheral tolerance. Because only passing reference has been made to the presence of Treg cells in human LN,²² we focused on the phenotypic and functional characteristics of this population. We found that Treg cells in LN did not show the preferential expression of CD45RB^low, characterizing Treg cells in PB (refs ^14–18,20 and present data). Moreover, Treg cells in LN expressed levels of CD45RO and CD62L significantly lower than the correspondent PB. Together with recent data on the existence of peculiar phenotypic profiles in Treg populations from different compartments in humans²¹ the present observation may imply that it is the local microenvironment that shapes the surface phenotype of Treg cells. The importance of the microenvironment is supported by the observation that also non-Treg cells in LN differed from the correspondent populations in PB in terms of CD45RO and CD62L expression.

The TCRVβ repertoire is first determined by negative and positive selection in the thymus and reflects the genetically determined repertoire of an individual.³¹ On this background, we examined the distribution of eight TCRVβ regions in Treg cells and the other CD4⁺ populations from LN and correspondent PB. We found that Treg cells in LN and PB exhibited a different TCRVβ repertoire. This diversity was unique, as TCRVβ repertoire of CD4⁺ non-Treg cells in LN, i.e. CD4⁺ CD25⁻ and CD4⁺ CD25⁺, showed a high concordance with that of PB. A similar finding has not been reported before and suggests that the peculiar environmental milieu of LN exerts a selective pressure on CD4⁺ lymphocytes preferentially driving the selection/expansion of the regulatory sets of clones needed in that particular context. An additional novel finding is the overall concordance in TCRVβ usage among the Treg and non-Treg CD4⁺ cells within each compartment, an observation which may imply a non-random TCRVβ usage among CD4⁺ populations. However, the strength of correlation differed among the three CD4⁺ subsets, being quite high between the CD4⁺ CD25⁻ and CD4⁺ CD25⁺ subsets, and low between the CD4⁺ CD25⁻ and CD4⁺ CD25⁺⁺ subsets. These results are in good keeping with the animal studies³² indicating that CD4⁺ CD25⁻ and CD4⁺ CD25⁺ subsets are more similar to each other than to CD4⁺ CD25⁺⁺ Treg cells, a consequence of the fact that CD4⁺ CD25⁺ subset is not homogeneous as it includes ordinary activated CD4⁺ lymphocytes from CD4⁺ CD25⁻ subset and a small number of CD4⁺ CD25⁺ Treg cells, i.e. CD4⁺ CD25⁺⁺ cells with a down-regulated expression of CD25. Of note, the presence of this small amount of Treg cells with an intermediate level of expression of CD25 might account for the small proportion of CD152⁺ cells we constantly found in the CD4⁺ CD25⁺ subset.

The regulatory function of human Treg cells has been demonstrated in adult PB, as well as in thymus and tonsils.^{18,20,33–36} Thus, it was important to ascertain whether Treg cells in LN expressed the same functional capacities attributed to Treg cells in those compartments. The present results show for the first time that Treg cells from human LN are anergic and endowed with the capacity to suppress the response of other T lymphocytes, both CD4⁺ and CD8⁺, to CD3/TCR-mediated stimulation.

In summary, we report a comparison of lymphocytes populating LN and PB in humans and describe novel differences in lymphocyte population distribution. To our knowledge, this is the first set of data describing that Treg cells can be found in normal human LN, express a peculiar phenotypic profile and TCRVβ repertoire and act as naturally anergic cells capable of suppressing the activation/proliferation of other T lymphocytes.