CYTOKINE RECEPTOR EXPRESSION IN HUMAN LYMPHOID TISSUE : ANALYSIS BY FLUORESCENCE MICROSCOPY

A highl y-sensitile tlourescence method, capable of detecting cytokine receptors present at low concentrations (around I DO molecules per cell) by flow cytometry, was adapted for use on tissue sections. Thi s method was used to examine the expression of several cytokine receptors in lymphoid ti ss ues. lL-2 receptors were distributed broadly, with higher concentrations in T cell areas. lL-1 receptor Type I was detected in T cell areas and in the follicular mantle, and was strongly expressed on vasc ular endothelium. IL-6 receptor was found at very low concentration , both within and outside germinal centres. The gp 130 molecule, which is involved in the functional receptor complex for IL-6 and several other cytokines, was present at higher concentrations, particularl y in the germinal centre . Analysis of receptor expression in secondary lymphoid tissue provides evidence bearing on the physiological roles of cytokines, as these tissues contain cell s at variou s stages of physiological activation located in well-defined functional zones.


INTRODUCTION
Cytokines control many aspects of the immune and haematopoietic systems, including the production, proliferation, differentiation, activation and migration of lymphoid and myeloid cells .Cytokine-responsive cells bear specific cytokine receptors.These are often two-or three-chain structures, with one chain interacting specifically with a single cytokine whilst the second chain, which usually confers high affinity on the interaction and tran sduces signals to the cell , may be shared by several cytokines (Taga and Kishimoto , 1992;Kondo et at., 1993;Noguchi et at., 1993).
Analysis of cytokine receptor expression by lymphoid and myeloid cell populations is useful in understanding the details of cytokine action, and may shed light on pathological as well as physiological processes.There have been many studies reporting cytokine receptor expression on circulating lymphocytes, and other studies reporting soluble receptor levels in the circulation, in health and in a variety of pathological conditions (for a review and commentary, see Zola, 1992 and 1994).However, studies of cytokine receptor expression in secondary lymphoid tissue have been limited.A number of immunohistochemical studies of cytokine receptor antibodies were presented at the 5th International Leucocyte Typing Workshop (Autschbach et at., 1993;De Weger et at., 1993;Berti et at., 1993;Zola et at., 1993d).These studies produced valuable results, but a common finding was that reactivity was variable and generally weak, presumably because of the low levels of cytokine receptor expression (Autschbach et at., 1993).Tissue studies are needed to further elucidate cytokine function, as secondary lymphoid tissues are the site of important immunological processes, such as cognate T celllB cell interaction and germinal centre (GC) reactions.Furthermore, the studies need to be done on tissue sections and not just on cell suspensions, in order to preserve the spatial relationships between cells, which are essential for in vivo immune responses.
Cytokines can mediate their effects on cells which bear very low concentrations of receptor (ranging from 10 to a few hundred molecules per cell), while conventional staining techniques detect molecules present at concentrations of several thousand perl cell (Zola, 1994;Zola et al., 1990b).For example, IL-J acts on thymocytes expressing less than 100 molecules of receptor (Dower et aI., 1985), while IL-4 acts on B cells expressing around 400 molecules of receptor per cell (Lowenthal et aI., 1988).This has necessitated the development of fluoresence staining and flow cytometric methods capable of high sensitivity in order to examine cytokine receptor expression by lymphoid functional sub-populations (Zolaet a!., 1989(Zolaet a!., , 1990b;;Olweus eta!., 1993;Coventry et aI., 1994;Zola, 1995).
In the current study we adapted our high-sensitivity flow cytometric procedures to mjcroscopy, and examined the expression of cytokine receptors in solid tissue.
Tissue preparing and staining.Lymphoid tissue was obtained fresh and cut into blocks (5-lOmm 3 ) which were snap-frozen in OCT compound (Miles Laboratories, Elkhart, In.).Frozen tissue blocks were stored at 80'C.Cryostat sections (6Ilm) were placed on gelatin-coated slides and air-dried overnight, to ensure adherence to the slide.A wax pen was used to draw a circle round the sections, to contain the reagents on the slide.Sections were blocked with 3% BSA in tris-buffered saline (TBS) for 30 min.All incubations were performed at room temperature in a humidified chamber, and were followed by washing with stirring for 5 min in TBS containing 1 % BSA.After each wash, excess liquid was removed by gentle blotting with paper tissue, immediately before the addition ofthe next reagent (to avoid drying-out).
Monoclonal antibody (I 00/lL) was placed on the specimen, at a dilution (in TBSI I %BSA) shown by preliminary titration to give optimal staining with minimal background staining.This was usually a 10-20-fo1d lower concentration than used for flo\\ cytometric analysis.The slides were incubated as described and then washed t\\ice.Bound antibody was detected by three different methods, which were compared for sensitivity.Three-stage immunofluorescence method.Bound monoclonal antibody was detected with biotinylated horse anti-mouse Ig (Vector Laboratories, Burlingame, Ca., catalog # BA-2000) (100!1L).This reagent required absorption with tonsil cells to reduce nonspecific staining.Absorption was carried out with 10 8 fresh or cryopreserved tonsil cells per mL of antibody at III 00 dilution, for 1 hr on ice, with occasional mixing.Preliminary titration indicated optimal staining at 11250 dilution (i.e.112.5 of the absorbed material) .Slides were incubated for 1 hr and washed as described, and the bound biotin was then visualised using Cy3-streptavidin (Jackson ImmunoResearch, West Grove, Pa., catalog # 016-160-084), 100!1L of a 1/500 dilution.Slides were again incubated and washed as above, after which excess buffer was removed by gently blotting with tissue paper.Slides were mounted in 90% glycerol in phosphate-buffered saline, containing DABCO to reduce fading.
Immunoperoxidase.This method was carried out as described previously (Coventry et al., 1994).The technique is similar in principle to the three-stage method, but utilises enzyme instead of fluorochrome.The reagents used were biotinylated rabbit anti-mouse Ig (Dako, Glostrup, Denmark, catalog # E354) followed by steptavidin-horseradish peroxidase (Pierce, Rockford, II., catalog # 21124).
Fluorescence microscopy.A number of different fluorescence microscopes were used in the course of this study, and significant differences were found in sensitivity.The Cy3 was well visualised using filter blocks intended for use with rhodamine, such as the Leica N2.1 block (excitation filter 515-560 nm, dichroic mirror at 580 nm and barrier filter at 580 nm) or Leica M2 (excitation filter 546± 14 nm, otherwise as N2.1).The choice of lens was important, with immersion lenses of high numerical aperture giving the highest sensitivity.
Cell suspension studies.Cells in suspension were stained and analysed by flow cytometry as described elsewhere (Zola et aI., 1990b;Zola, 1995).For examination in the microscope, stained cells were deposited on slides using a cytocentrifuge.

Comparison of fluorochromes
Phycoerythrin.This fluorochrome gives the highest sensitivity by flow cytometry.However, when examined by fluorescence microscopy, PE fluorescence faded within seconds under microscope illumination, making image capture and examination impractical.A number of anti-fading agents were examined, including DABCO, which is effective in reducing fading of fluorescein, and reagents marketed specifically for use with PE (AF23, AF93, AFll3, AF133 ; Citifluor, Agar Scientific, Stansted, Essex, UK), but fading was still too rapid to allow image capture and examination.Rapid image capture, using a highly-sensitive video camera, was attempted, but it was difficult to focus and select a field for image capture without losing much of the fluorescence.In an attempt to stabilise fluorescence , sections were fixed directly after staining and before mounting.Sections fixed by brief immersion in acetone showed a significant reduction in fading , which was obvious up to 48 hr after staining, but fixing reduced staining intensity, and PE-labelled samples did not appear as bright as samples stained with Cy3 (see below).A number of variations in fixative and conditions of fixation did not improve the retention of staining.
Cy-3.Cy-3 gave intense red staining that appeared by eye to be as bright as PE staining.Since the extinction coefficient and quantum yield of Cy-3 (130,000 and 0.15) are lower than those ofPE (2,000,000 and 0.8) the latter should be capable of giving a much brighter signal, but fading negates the theoretical advantages of PE.

Comparison of staining protocols
Staining was carried out using two-stage and three-stage protocols as described.Figure 1 and Figure 2 show that staining was brighter using the three-stage protocol.

Sensitivity of the Cy-3 system
In order to quantify the sensitivity of the Cy-3 system in comparison with the extensively-characterised PE system, samples of cells were stained in suspension with monoclonal antibody followed by biotinylated horse anit-mouse Ig.They were then stained either with PE-streptavidin (Caltag, San Francisco, Ca., catalog # SA 1004-4) for flow cytometry or Cy-3 streptavidin for cytocentrifugation and microscopy.The cytocentrifuge preparations were examined using alternatively phase contrast (Normarski) and dark-field fluorescence epi-illumination to count the proportion of cells stained.Flow cytometric results are shown in Figure 3, and the proportion of cells counted as positive in the cytocentrifuge preparations are given in the Figure caption for comparison.These results indicate that Cy-3 fluorescence microscopy allows sensitivity approximately equal to that of flow cytometry with PE.X63 is a negative control, while OKT3 is included to check that the number of un-stained cells is not being undere stimat ed in phase-contrast microscopy.
In a separate experiment, tonsil sections were stained by Cy-3 for ill1ll1un otlu ores - The greater sensitivity of the three-stage method is seen with CD25, which gives a bimodal pattern only in the three-stage method, and with CD3, which shows a brighter peak of positive cells (MFI=206 for the three-stage and 58 for the two-stage method).The negative controls are not increased by the extra layer of staining.Note that Cy-3 is not measured effectively with 488 nm excitation (compare results to Figure 3).
both able to detect a scattering of cells in the T cell zone expressing relatively high amounts of CD25, but immunofluorescence detected in addition a larger number of weakly-stained cells distributed throughout T and B cell areas.

Expression of receptors on lymphoid tissue sections
The expression of a number of cytokine receptors was examined in human tonsil, spleen, ileum, gut-associated lymph node and appendix.Table 2 summarises the distribution of staining in tonsil, which was studied most extensi vely (3 different tonsils, 5 separate staining experiments), whilst Table 3 summarises the results obtained on a number of different tissues from the same individual.Some of the observations are illustrated in Figure 5 and Figure 4.For some antibodies, the microscope studies were extended using flow cytometric analysis of tonsil cell suspensions.
CD 121 a (IL-l receptor Type I) showed strong staining of vascular endothelium in tonsil (Figure Sa), but vascular endothelium in spleen was stained only weakly.In the lymphoid areas, there was moderate staining of cells in the T cell zone (Figure Sa), the follicular mantle appeared slightly brighter than the T cell zone, and GC were usually

31.8%
Figure 3. Flow cytometric analysis of normal blood lymphocytes (gated on forward and orthogonal light scatter, top left panel).Staining was with the three-stage method using PE.The % values represent the proportion of lymphocytes staining with each antibody, and may be compared with the data obtained by microscopy after staining the same samples using the three-stage method with Cy-3: X63: 1.5%; CD25: 38%: OKT3: 66%.negative, although weak staining was seen in one tonsil.Obvious staining of lymphoid areas, as well as bright staining of vessels, was also seen in the other lymphoid tissues.

Table 2. Cytokine receptor distribution in tonsil sections
In suspension, however, no positive cells were seen in the lymphocyte population.c-kit was also not detected on lymphoid cells by flow cytometry, although in sections scattered brightly stained cells were seen in lymphoid areas (Figure 5b), and more abundantly in connective tissue.IL-4R (CDwI24) was not seen in sections, although we have previously demonstrated weak staining of a proportion of tonsil B cells in suspension studies ( 16).The IL-6 receptor CDI26 was seen at very low intensity in the GC and in extrafollicular areas.CD 130, which together with CD 126 forms the functional IL-6 receptor, stained more strongly, with preferential localization in the GC (Figure 5c).By flow cytometry (Figure 6) both CD 126 and CD 130 were associated primarily with IgDnegative, CD23-negative, CD38 -positive B cells, which are principally GC cells.CD 130 stained vascular endothelium, though not as brightly as CD121 a. IL-2 receptor staining was rather uniform for all three chains of the receptor, although CD25 showed stronger staining in T cell areas, with scattered bright cells in the T cell zone (Figure 4) and in the GC (Figure 5d).Flow cytometry showed no clear correlation with sub-populations (Figure 7).

DISCUSSION
The results presented show that immunofluorescence staining of tissue sections can give sensitivity comparable to flow cytometry for the detection of molecules expressed at low levels, in particular cytokine receptors.Some cytokine receptors were expressed at levels which precluded clear interpretation, but the sensitivity was generally greater than that achieved using immunoenzymatic methods and the method therefore provides an advance on existing techniques.Using high-sensitivity methods, the distribution of a number of cytokine receptors in secondary lymphoid tissue can be mapped and provides some interesting indications as to their function .
IL-I receptor (Type I) was found associated with vascular endothelium and was expressed in the T cell areas of lymphoid tissues .Receptor expression on vascular endothelium has been described, based on studies of cells in suspension (Akeson et aI. , 1992;Colotta et at., 1993).Aortic endothelial cells were found to express 1000-3000 copies of the Type I receptor, and no Type II receptor (Akeson et aI. , 1992).IL-l receptor on endothelial cells is functional , since IL-l increases ICAM -I expression on endothelial cells, and this can be blocked with antibody against the receptor (Akeson et aI., 1992).
The fact that CD 121 a is detectable on tonsil sections but not in suspension studies is unlikely to be due to a difference in sensitivity, since other receptors were detected equally well by tissue and suspension studies (CD25) or with greater sensitivity in suspension (lL-4 receptor).We were not able to detect significant expression of IL-IR on circulating blood cells in an earlier study (Zola et aI. , 1993d).An alternative explanation is that the IL-I receptor is stored intracellularly (and is thus accessible in sections but not on the cell surface) or that the receptor is lost from the cell surface during the processing of cell suspension.There is published evidence for shedding of the Type I receptor (Arend et at., 1994) and the Type II receptor (Colottaet aI. , 1993).The presence of IL-I R on mantle zone cells is particularly interesting in the light of the expression of membrane IL-I a selectively in the follicular mantle and extrafollicular areas (Zola et at., 1993c).

III III
r::

FLI-H'CD23FITC ---)
FLI-H'HB70FITC ---) FLI-H'CD3FITC ---) IL-2 receptors showed weak uniform staining throughout the lymphoid areas, with the ex chain (CD25) showing somewhat stronger general staining in T cell areas, as well as a scattering of brightly-stained cells in T cell areas and Gc.We have previously shown by high-sensitivity staining that IL-2R ex and ~ chains are expressed on a much larger population of cells than can be detected by conventional methods (Zola et al., 1989(Zola et al., , 1990c)), and the y chain is also quite broadly expressed in circulating cells (Ishii et aI., 1994).The cells in tissue that express high levels of CD25, detectable as bright cells in immunofluorescence and also stained by immunoperoxidase, are likely to be activated cells, since CD25 has a rapid turnover (Sayar et al., 1990).Cells expressing lower levels of the 3 receptor chains are potentially responsive to IL-2, especially since IL-2 can upregulate its own receptor (Waldmann, 1989).
IL-4 receptor was not detected by microscopy, although we have previously demonstrated expression of this receptor on a subset of B cells by flow cytometry (Zola et al., 1993a).This difference may be technical, although, as discussed above, there is no evidence for a systematic difference in sensitivity between our tissue section method and flow cytometric method.
IL-6 receptor gp80 (CDI26) was seen at low level in the Gc. and the gp130 signal transcution molecule (CD 130) was also seen primarily in the Gc. at significantly higher level.Although the difference in staining intensi ty for the two chains of the IL-6 receptor may simply reflect differences between the antibodies, it does suggest that GC cells may also use gp 130 for one or more of the cytokines that share this signalling molecule, cilliary neurotrophic factor (CNTF), oncostatin M, leukaemia inhibitory factor (LIF) and IL-ll.Once reagents are available, it would be interesting to examine the expression of the specific receptors for these molecules in GC, and the functional role of the corresponding cytokines in the GC reaction.Autschbach et al. (1993) also found that CD130 was more broadly expressed than CD126; they attributed GC staining with CD 130 to follicular dendritic cells.The role ofIL-6 as a B cell differentiation factor active in late stages of differentiation (Kishimoto, 1989) would lead to the expectation of IL-6 receptor expression on GC B cells, but we have found CD 126 to be present at very low levels and often not detectable on tonsil GC cells, both in suspension and section studies.It may be that IL-6 acts later, after plasma cell precursors have left the Gc.IL-7 receptor and the two TNF receptors were detected at such low levels that interpretation of distribution would be unreliable.We have previously demonstrated strong expression of IL-7R on peripheral blood lymphocytes, but much lower expression on tonsil cells in suspension (Zola et al., 1993d).We also found TNF receptor Type I positive on circulating lymphocytes, particularly CD4 cells, but relatively poorly expressed on tonsil cells in suspension (Zola et al., 1993b).We have not been able to detect TNF receptor II on circulating or tissue lymphocytes (Zola et al., 1993d(Zola et al., , 1993b)).
Our understanding of cytokine action comes from a number of experimental avenues, most of them involving models well removed from physiological states.We know what cytokines can do, from in vitro activation experiments and from mice over-expressing the cytokines.We know something about the processes for which individual cytokines are essential, from knock-out mice and from "experiments of nature", but neither of these sources of information tells us reliably what cytokines do physiologically.Lymphoid tissues can provide important clues, because they contain resting cells, physiologically activated cells, cells in transit and cells localised in well-defined functionally interrelated areas.Studies of cytokine (Hoefakker et al., 1993;Andersson et al., 1994) or cytokine mRN A (Emilie et ai., 1990) distribution in tissue have provided clues as to the physiological reactions that depend on cytokines.For example, the predominantly extrafollicular localization of cells producing IL-2, IL-4, IL-6 and IL-lO (Andersson et ai., 1994;Emilie et ai., 1990) suggest that the primary roles of these cytokines is not in the GC, in spite of current models which suggest that class switching and B cell differentiation, processes that are thought of as being driven by these cytokines, occur in the Gc.Receptor expression studies can provide essential complementary information to cytokine production data, and it is interesting that IL-2R was detected in the GC, but IL-4 and IL-6R were negative or weak and variable respectively.The finding ofIL-l R in the mantle and T cell area complements the demonstration (Andersson et ai., 1994) of ILl a in these same areas.We previously demonstrated by flow cytometry that membrane IL-l a is specifically found on B cells with the phenotype of mantle zone cells (Zola et ai., 1993c).The follicular mantle is generally thought of as a quiescent area, containing the un-stimulated remnants of the primary follicle.The co-expression of IL-l and its receptor in the follicular mantle suggest a much more active role for mantle cells.

Figure I .
Figure I. Flow cytometric analysis of Cy-3 fluorescence, comparing two-stage (upper patterns) and threestage (lower patterns) staining with CD3 and CD25.Each histogram is shown superimposed on a negative control antibody.The greater sensitivity of the three-stage method is seen with CD25, which gives a bimodal pattern only in the three-stage method, and with CD3, which shows a brighter peak of positive cells (MFI=206 for the three-stage and 58 for the two-stage method).The negative controls are not increased by the extra layer of staining.Note that Cy-3 is not measured effectively with 488 nm excitation (compare results to Figure3).

Figure 2 .
Figure 2. Com pari son of two-stage and three -stage staining in tonsil sections.Tonsil was sta ined with anti-[gO by the two procedures; the three-stage procedure (B) sho wed much brighter staining than the two-stage procedure (Al. type I (CD 120a) TNF type 2 (CD120b) c-kit (CD I 17) Tissue distribution Vascular endothelium bright, T-cell zone moderate, genninal centre negative Widespread weak staining, brighter T cell zone; scattered bright cells Weak staining uniformly throughout tissue Weak staining uniformly throughout tissue Negative Very weak staining Germinal centre stained preferentially, some staining in T cell zone Very weak staining Very weak staining Very weak staining Scattered bright cells, mainly in connective tissue .] scattered brightly stained cells in connective tissue' or in germinal centre 2 ; a few moderately-stained cells in T cell area'-

Figure 4 .Figure 5 .
Figure 4. Comparison of immunoperoxidase and immunofluourescence.Figures 4a-c are immunoperoxidase staining, Figures 4d-e immunofluorescence.Panels a and d show a negative control antibody, panels band e show IgD staining of the follicular mantle, and panels c and f show CD25 staining principally in the

Figure 7 .
Figure 7. Three-colour analysis of expression of CD25.CD25 (vertical axis) is plotted against from left, top row: CD3 (T cell marker), CD 19 (B cell marker), CD38 after gating for CD 19-positive cells (B cells), bottom row from left: CD23 on gated B cells, IgD on gated B cells.The lower right figure shows a negative control antibody, isotype-matched for the CD25 antibody, against CD3.

Table 1 .
Monoclonal antibodies against cytokine receptors used in this study.

Table 3 .
Comparison of different lymphoid tissues.