Role of stromal-cell derived factor-1 in the development of autoimmune diseases in non-obese diabetic mice

Abstract

The chemokine stromal-cell derived factor-1 (SDF-1) controls maturation, trafficking, and homing of certain subsets, lymphoid cells including immunogenic B and T cells, as a ligand of the CXCR4 chemokine receptor. Insulin-dependent diabetes mellitus (IDDM) and Sjögren's syndrome (SS), both highly regulated autoimmune diseases, develop spontaneously in non-obese diabetic (NOD) mice. To investigate the role of SDF-1 in the development of autoimmune diseases, we injected groups of NOD female mice with antibodies to SDF-1 (anti-SDF-1), which resulted in a 30% reduction of diabetes up to 30 weeks of age, delayed average diabetes onset by 10 weeks, and suppressed insulitis. Autoimmune sialoadenitis was evident in anti-SDF-1-injected mice (SDF-1-Ig group) at the same level as in all groups of mice, whether injected with non-specific antibodies or not. In addition, in the SDF-1-Ig group, a greater number of immunoglobulin M (IgM)⁻ IgD⁻ B220^low CD38⁺ CD43⁺ CD23⁻ progenitor B cells and IgM⁺ IgD⁺ B220^high CD43⁻ CD38⁺ CD24⁺ CD23⁺ mature B cells remained in the bone marrow, whereas infiltration of mature IgM⁺ B cells was less extensive in peripheral tissues. Our results suggested that anti-SDF-1 antibodies injection was effective in inhibiting diabetes and insulitis without affecting autoimmune sialoadenitis or SS in NOD mice. SDF-1 may be an essential chemokine for trafficking and migration of autoreactive B cells in the development of diabetes.

Introduction

Insulin-dependent diabetes mellitus (IDDM) and Sjögren's syndrome (SS) are highly regulated autoimmune diseases, certainly in mice and probably in humans,^1–4 and both spontaneously develop in non-obese diabetic (NOD) mice at 15–25 weeks of age.³ The NOD mouse is an established model of human IDDM with many of the genetic and immunological features of the human form of the diseases.⁵ The development of IDDM is characterized by the generation of pancreatic islet β-cell protein-reactive T-lymphocytes, and the infiltration of these cells, dendritic cells, and monocytes into the islets, as well as the terminal destruction of β-cells.^3,6–8 SS, on the other hand, is a systemic autoimmune disease characterized by oral and ocular dryness, accompanied by clinical observations of a progressive loss of salivary and lacrimal function, that is related to the presence of a perivascular and periductal leucocyte infiltrate^9,10 and systemic production of autoantibodies to ribonucleoprotein.¹¹ In recent reports it has been made clear that both the CD4 and CD8 subsets of T-cells play a crucial role in the development of IDDM in NOD mice,^12,13 who also develop lymphocytic inflammation in their submandibular salivary (sialoadenitis) and lacrimal (dacryoadenitis) glands.^14,15 These findings have led to the notion that recruitment of a threshold frequency of autoreactive T-cells into the pancreatic islets and salivary glands may be required for progression to β cell and salivary gland tissue destruction.

It has recently been proposed that B lymphocytes may play a more critical role in the induction of immunological activation as an antigen-presenting cell (APC) population, and that they are essential for the initial development and/or activation of β cell autoreactive T cells in NOD mice.^16–18 Further, B lymphocytes have a greater capacity to induce various immunotolerogenic functions than other APC populations.^19–21 It has also been reported that IDDM susceptibility was restored in NOD,Igµ^null mice (IDDM resistance) reconstituted with syngenic NOD,Igµ^null mice bone marrow plus purified NOD B lymphocytes, but not with syngenic bone marrow.¹⁸ Thus, the maturation of B cells in bone marrow may be essential for the development of IDDM in NOD mice.

For the immigration of lymphocytes into lymphoid organs, cell adhesion molecules play a functional role.^22,23 Moreover, chemokines serve as selective triggers of adhesion molecule regulation during lymphocyte homing, and are also involved in the recruitment and proper positioning of leucocytes within specialized lymphoid tissues, including lymphoid tissues, Peyer's patches, the thymus, and the spleen.^24–26

Stromal cell derived factor-1 (SDF-1) was initially cloned from mouse bone marrow stromal cells and a CXC chemokine originally described as pre-B-cell growth-stimulating factor (PBSF).^27,28 It is expressed constitutively in several tissues, including the bone marrow, thymus, spleen, and liver,^27,29 rather than being up-regulated during inflammation or immune reactions. SDF-1 contains a highly efficacious lymphocyte chemoattractant³⁰ that controls maturation, trafficking, and homing of certain lymphocyte subsets.^31–34 Autoreactive B cells infiltrate organs without inducing tolerance from bone marrow for the initiation of autoimmune diseases in NOD mice. This indicates that SDF-1 controls B-cell development and trafficking, and may have a crucial role in the development of autoimmune diseases. The aim of the present study was to examine the role of SDF-1 chemokines in the development of IDDM and SS in NOD mice by injection of anti-SDF antibodies.

Materials and Methods

Animals and diabetes

NOD/LtJ and NOD/LtSz-prkdc^scid/prkdc^scid (NOD-severe combined immunodeficient (SCID) mice were originally purchased from The Jackson Laboratory (Bar Harbor, ME). More than 225 females housed in our animal laboratory were examined for diabetes weekly, by quantitating urine glucose levels with reagent strips (Uritix, Bayer Medical Ltd, Newbury, UK) and confirming positives by blood glucose measurements. From those, 7-week-old-mice were randomly selected for injection and divided into three groups: the SDF-1-Ig group (n = 13), which were injected with goat anti-mouse SDF-1 polyclonal antibodies (C-19); the Control-Ig group (n = 12), which were injected with goat IgG polyclonal antibodies (non-specific); and the Untreated group (n = 200), which were left untreated. All antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA), and diluted (1 µg in 500 µl) in sterile phosphate-buffered saline (PBS) for injection. Intraperitoneal (i.p.) injection doses in the SDF-1-Ig and Control-Ig groups were 1 µg every other day from 7 to 30 weeks of age. B10.D2 mice and C57BL/6 mice were obtained from Japan SLC, Inc. (Shizuoka, Japan). ‘F1’ (infant NOD/LtJ & B10.D2 mice) and NOD-SCID mice were used as negative control for SS monitoring.

Antibodies

Fluoroscein isothiocyanate (FITC)-conjugated rat anti-CD4 (H129, 19), anti-IgM (R6-60.2), anti-CD90.2 (Thy-1.2, 53-2.1), T-cell receptor (TCR) β (H57-597), Sca-1 (D7) monoclonal antibodies (mAb), NK-1.1 (PK136), CD38 (90), CD24 (M1/69), BP-1 (6C3), CD23 (B3B4), CD19 (1D3) and anti-CD43 (1B11) mAb, phycoerythrin (PE)-conjugated rat anti-CD8 (53-6,7), CD2 (RM2-5) and B220 (RA3-6B2), and cychrome-conjugated rat anti-CD44 (IM7) mAb were purchased from PharMingen (San Diego, CA).

Collection of saliva

At 30 weeks old, all experimental mice were injected (after anaesthesia) with a cocktail of isoproterenol (0·20 mg/100 g body wt) and pilocarpine (0·05 mg/100 g body wt) (Sigma Chemical, St. Louis, MO) in PBS as secretagogue. Subsequently, saliva was collected from the mouth by a micropipette for 15 min and stored in 1·5 ml microfuge tubes. Each saliva sample was examined for its ability to hydrolyse starch using an amylase kit (Sigma). In short, 20 µl aliquots of 1000-fold dilutions of saliva in phosphate buffered saline (PBS) were added to 1 ml of amylase reagent, containing 4,6-ethlidene (g7)-p-nitrophenyl (G1)-α and d-maltoheptaside (ET-G7PNP), mixed immediately by inversion, and incubated at 37° for 2 min. Absorbance was recorded at 405 nm.

Histological examination

All mice were killed at 30 weeks of age after saliva and blood serum samples had been collected. The pancreas and salivary glands were snap-frozen in OCT-Compound, cut into 6 µm sections (Microm, Zeiss, Germany), and stained with haematoxylin-eosin for histological observation. Histological observations and photomicrography were performed using an Olympus BX50WI microscope (Olympus Inc., Tokyo, Japan).

Immunohistochemistry

Immunostaining was performed using a Vecta Stain Elite ABC Kit (Vector Laboratories, Berlingame, CA). Pancreases from the 4 month-old female SDF-1-Ig, Control-Ig, and Untreated groups were snap-frozen in OCT-Compound and cut into 6-µm-thick sections at −20°. Non-specific protein-binding sites were blocked by incubating sections with 10% normal goat serum for 10 min. Endogenous peroxidase activity was inhibited by incubation with methanol containing 0·3% H₂O₂. Slides were stained with antimouse CD4, CD8, and macrophage (F4/80) monoclonal antibodies in a humidity chamber for 1 hr at 37°. The slides were washed twice in PBS, stained with biotinylated goat anti-rat IgG antibody (CALTAG, San Francisco, CA) for 30 min, and then incubated with peroxidase–avidin complex substrates for 30 min. The slides were washed in PBS and a colour reaction was performed by incubation in aminoethyl carbazole. After washing in PBS, the slides were counterstained with Mayer's hematoxilin.

Flow cytometry

Surface markers were identified using mAbs in conjunction with two- or three-colour immunofluorescence analysis with a FACScan (Becton-Dickinson, Mountain View, CA). Double-labelled surface phenotypes used were CD4/CD8, IgM/B220, IgD/B220, CD38/B220, CD24/B220, BP-1/B220, CD43/B220, and CD23/B220. The triple-labelled surface phenotype was CD44/CD2/Sca-1. Single-cell suspensions from bone marrow, spleen, and thymus specimens were prepared using a homogenizer in Hanks' balanced salt solution (HBSS) medium. The cells were incubated with the above antibodies for 1 hr, washed three times in HBSS, and fixed in a 3% formaline/PBS solution before being analysed with flowcytometer software (Cell Quest, Becton Dickinson, San Jose, CA).

Statistical analysis

A Kaplan–Meier cumulative survival test was used to compare the incidence of diabetes. Comparative analyses were performed by anova. A P-value <0·05 was considered statistically significant for two-tailed comparisons. All statistical analyses were performed using StatView for the Macintosh operating system.

Results

Incidence of diabetes and insulitis

The incidence of diabetes was approximately 40% in the SDF-1 group, which was 30% less than that in the Untreated and Control-Ig groups at 30 weeks of age (Fig. 1). We also observed that the average onset of diabetes was delayed by approximately 10 weeks in the SDF-1-Ig group as compared to the Untreated group. Furthermore, histological photomicrographs revealed no insulitis in a non-diabetic F1 mouse (Fig. 2a) or peripheral insulitis in anti-SDF-1 injected mice, in which the vast majority of islets remained undestroyed at 30 weeks of age (Fig. 2b), whereas the islets of control mice were almost completely destroyed by autoreactive lymphocytes (Fig. 2c). The histological examinations at 8 months also revealed severe insulitis in the Control-Ig group (Grade 0: 12·6±5·7%, Grade 1: 26·9±11·5%, Grade 2: 12·7±4·3%, Grade 3: 12·3±6·6%, and Grade 4: 35·3±7·8%) (Fig. 3), while the SDF-1-Ig group tended to develop mild insulitis (Grade 0: 22·8±5·3%, Grade 1: 32·8±5·2%, Grade 2: 23·8±2·8%, Grade 3: 14·8±3·3% and Grade 4: 5·7±2·2%). In particular, there was a larger difference in Grade 4 insulitis between the SDF-1-Ig and Control-Ig groups.

Figure 1.

Administration of anti-SDF-1 antibody to NOD mice from 7 to 30 weeks of age reduced the incidence of diabetes by 30% at 225 days and delayed the average onset of diabetes by approximately 160 days. The cumulative incidence of diabetes was monitored in three groups of female NOD mice: 13 were treated with anti-SDF-1 antibody (thick line), 12 were treated with control IgG antibody (gray line), and 200 received no treatment (thin line). Injection was started at 7 weeks of age with 1 µg given i.p. every other day until 30 weeks of age.

Figure 2.

Photomicrographs of pancreas specimens show islets at 30 weeks of age, detected as diabetic-negative (H-E,×200). Arrowheads=mononuclear lymphocytes. (a) N=normal islet from an F1 mouse. (b) Typical histological appearance in a SDF-1-Ig group mouse, in which the islet had been peripherally infiltrated by lymphocytes, through the vast majority remained normal. (c) Control-Ig group mouse islets in which the β-cells have been completely destroyed by lymphocytic infiltrates.

Figure 3.

Analysis of insulitis in SDF-1-Ig and Control-Ig group mice at 8 months of age. The degree of mononuclear cell infiltration was graded as follows: Grade 0; no infiltrating cells in the islets, Grade 1; infiltrating cells adjacent to the islets but not in the islets (peri-insulitis), Grade 2; infiltrating cells occupying less than 25% to 50% of the islets, Grade 3; infiltrating cells occupying 25% to 50% of the islet area, Grade 4; infiltrating cells occupying more than 50% of the islets. The assessment of insulitis was performed using haematoxylin and eosin stained pancreas sections. Results are expressed as the means±SD of the insulitis counts for six independent mice selected randomly. The asterisk denotes significantly different insulitis counts between SDF-1-Ig and Control-Ig (P < 0·01).

Autoimmune sialoadenitis

All NOD mice spontaneously developed autoimmune sialoadenitis, as histological examinations revealed large numbers of inflammatory cells infiltrating the salivary glands with some acinar degeneration in all SDF-1-Ig and Control-Ig mice at 30 weeks of age (Fig. 4). Table 1 summarizes the relative intensity of autoimmune sialoadenitis in NOD mice during the course of the experiment, based on physiological and biochemical changes. Total volume and volume per 100 g of body weight of secreted saliva were significantly smaller in all injected mice than in the Untreated group, particularly when the mice were diabetes-positive. The SDF-1-Ig group was similar to the Control-Ig group in total volume, volume per 100-g body weight, and amylase activity. Moreover, the auto-DNA antibody in the SDF-1-Ig and Control-Ig groups was positive (data not shown).

Figure 4.

Photomicrographs of submandibular salivary glands at 8 months of age, detected as diabetic-negative (H-E,×200). Arrowheads=mononuclear lymphocytes, arrows=acinar degeneration. (a) Salivary gland from a F1 mouse with normal seromucous acinar cells and tubular cells. (b) and (c) salivary glands of SDF-1-Ig and Control-Ig group mice, respectively, showing mononuclear lymphocyte infiltration and acinar degeneration.

Table 1.

Comparison of physiological and biological changes between SDF-1-Ig and Control-Ig NOD mice

Mice		Body weight (g)	Total secreted saliva (µl in 15 min)	Saliva volume volume/BWt. (µl/100 g)	Saliva protein concentration (µg/µl)	Saliva amylase activity (mg starch/min/ml)
NOD-SCID	(n = 5)	24·8±2·6	179·2±32·1	728·6±149·0	7·2±1·9	482·7±216·1
	Non-diabetic (n = 5)	27·6±1·3	122·4±31·5**	442·0±110·0*	6·1±1·5	339·0±227·2
Untreated	Diabetic (n = 5)	21·8±3·3	41·7±21·2*	200·0±104·0*	7·8±1·5	1032·3±275·1*
	Non-diabetic (n = 7)	26·5±0·6	85·0±31·1*	320·0±116·0*	6·0±1·0	522·7±167·7
SDF-1-Ig	Diabetic	23·7±5·1	24·9±11·4*	105·0±42·0*	6·7±0·4	992·0±318·2**
Control-Ig	Non-diabetic (n = 3)	26·7±0·3	120·4±31·5	452·0±121·0**	6·1±0·2	543·7±82·5
	Diabetic (n = 9)	21·2±2·9	42·9±30·5*	195·0±122·0*	7·2±0·9	970·3±181·0*

All values are expressed as the mean±SD.

^*versus NOD-SCID mice, P < 0.01.

^**versus NOD-SCID mice, P < 0.05.

Changes in B-cell progenitor cells and mature cells

In an attempt to understand changes that occurred in haematopoietic cells, we attempted to characterize those from bone marrow using several haematopoietic cell phenotype markers, as mentioned above. Samples gated for forward light scatter (FSC) and sidelight scatter (SSC) by fluorescence-activated cell sorting (FACS) were used to identify viable lymphocytes. The proportions of major subsets were determined by quadrant analysis. There were higher proportions of CD44 and CD2 DP cells found in bone marrow from the SDF-1-Ig group than from the Control-Ig group (Fig. 5a, left). Furthermore, region 1 (R1) in the CD44⁺ CD2^high cells showed a higher prevalence of Sca-1⁺ cells in the SDF-1-Ig group as compared to the Control-Ig group (Fig. 5a,b). However, CD44⁺ CD3^low cells (R2) and CD44⁺ CD2⁻ cells (R3) did not show any differences between the SDF-1-Ig and Control-Ig groups in the proportion of Sca-1⁺ cells. These results indicated that a greater number of CD44⁺ CD2^high Sca-1⁺ progenitor cells remained in the bone marrow at 30 weeks of age in the SDF-1-Ig group than in the Control-Ig group. To characterize B-cell lineage in bone marrow from the SDF-1-Ig and Control-Ig groups, various antibodies to the lineage marker were used for FACS analysis. A strong expression of CD38 in haematopoietic cells has been implicated in the homing and stabilization of progenitor cells.³⁵ We performed double staining using anti-B220 and -CD38 antibodies to bone marrow cells from the SDF-1-Ig and Control-Ig groups. As a result, the proportion of B220^low CD38⁺ cells (B-cell progenitor) in the SDF-1-Ig group was approximately 1·6-fold greater than in the Control-Ig group (Fig. 6a,b).

Figure 5.

Developmental changes in the expression of surface antigens on bone marrow cells from SDF-1-Ig and Control-Ig group mice (non-diabetic) at 4 months of age. Samples gated on the forward light scatter (FSC) and side light scatter (SSC) were used to identify viable lymphocytes. Bone marrow cells falling into the lymphocyte gate were double-stained with FITC-anti CD2 and cychrome-anti-CD44 antibodies. CD2 positive cells were divided into three groups [CD2^high CD44⁺ cells (R1), CD2^low CD44⁺ cells (R2), and CD2⁻ CD44⁺ cells (R3)]. Log fluorescence intensities were plotted in these experiments. Numbers in quadrants indicate the percentage values of cells within the lymphocyte gate. Typical data were representative of three independent experiments (a). Bar graphs showed the mean±SD for the percentage of lymphocytes expressing the indicated cell surface marker selected as relevant data (b). The asterisk denotes significantly different insulitis counts between SDF-1-Ig and Control-Ig (P < 0·05).

Figure 6.

Developmental changes in the expression of surface antigens on bone marrow cells from SDF-1-Ig and Control-Ig group mice (non-diabetic) at 4 months of age. Samples gated on the forward light scatter (FSC) and side light scatter (SSC) were used to identify viable lymphocytes. Bone marrow cells falling into the lymphocyte gate (R1 in the upper panels) were double stained with either FITC-anti-IgM, FITC-anti-IgD, FITC-anti-CD43, FITC-anti-CD38, FITC-anti-CD24, FITC-anti-BP-1, or FITC-anti-CD23 together with PE-anti-B220. Log fluorescence intensities were plotted in these experiments. Numbers in circles and squares indicate percentage values of cells within the lymphocyte gate. Typical data were representative of three independent experiments (a). Bar graphs showed the mean±SD for the percentage of lymphocytes expressing the indicated cell surface marker selected as relevant data (b). The asterisk denotes significantly different insulitis counts between SDF-1-Ig and Control-Ig (*P < 0·01, **P < 0·05).

Early B-cell progenitors in the bone marrow of mice can be divided into subpopulations on the basis of cell surface markers. B-cell progenitors (prepro-B, pro-B, and early pre-B cells) are surface B220- and CD43-positive, and can be subdivided on the basis of CD24 and BP-1 surface markers.³⁶ Further, B-cell bearing immunoglobulin antigen receptors are produced in bone marrow by differentiation from pro-B (B220⁺ CD43⁺ IgM⁻) and pre-B (B220⁺ CD43⁻ IgM⁻) cells through an ordered sequence of Ig heavy chain and light chain genes rearrangements. The proportions of B220^low IgD⁻ and B220^low IgM⁻ cells (pro- or pre-B cells) in the SDF-1-Ig group were higher than in the Control-Ig group, and the proportion of B220^high CD43⁻ cells (mature B cells) was approximately 4·7-fold greater than in the Control-Ig group (Fig. 6b). The distributions of other positive marker cells within the SDF-1-Ig and Control-Ig groups were similar (Fig. 6a). Taken together, greater numbers of IgM⁻ IgD⁻ B220^low CD38⁺ CD43⁺ CD23⁻ progenitor B cells and IgM⁺ IgD⁺ B220^high CD38⁺ CD43⁻ CD24⁺ CD23⁺ mature B cells remained in the bone marrow of the SDF-1-Ig group. Moreover, the proportions of B220^high IgM^high cells and B220^low IgM^low cells in the spleen of the SDF-1-Ig group were significantly lower than in that of the Control-Ig group (Fig. 7a,b).

Figure 7.

Characterization of mature B and T-lymphocytes in the spleens of SDF-1-Ig and Control-Ig group mice (nondiabetic) at 4 months of age. All data are presented as percentages. (left column) Differences in IgM and B220 positive cells were observed in the thymus of SDF-1-Ig and Control-Ig mice. (right column) No large differences between CD4⁻ and CD8⁻ positive cells were observed in the spleens. Typical data were representative of three independent mice experiments (a). Bar graphs showed the mean±SD for the percentage of lymphocytes expressing the indicated cell surface marker selected as relevant data (b). The asterisk denotes significantly different insulitis counts between SDF-1-Ig and Control-Ig (P < 0·05).

T-cell and macrophage characterization

To define T-cell and macrophage trafficking in the lymphocytes, pancreas, and salivary glands from the SDF-1-Ig, Control-Ig, und Untreated groups of NOD mice, we performed FACS and immunohistochemistry analyses using antibodies against CD4 and CD8. The proportions of CD4⁺ and CD8⁺ cells were not largely different between the SDF-1-Ig and Control-Ig groups (Fig. 7). Moreover, in the thymus, the proportions of CD4⁺ and CD8⁺ cells were not different between the two groups (data not shown). Immunohistostaining of lymphocytes that had infiltrated at the same stage of insulitis grade 2 in the SDF-1-Ig and Control-Ig groups showed no difference in the proportion of CD4, CD8, and macrophage cells between the two groups (data not shown).

Discussion

We examined the effect of anti-SDF antibodies on the incidence of IDDM and SS in NOD mice, as well as the mechanisms of development and progression of those diseases. Mice injected with the antibodies exhibited delayed development of diabetes and insulitis during the experimental period (Figs 1 and 2). One mechanism contributing to the reduced incidence of insulitis and diabetes may have been a down-modulation of the activation and migration of autoreactive T and B cells from bone marrow to lymphocyte organs. Therefore, we examined hematopoietic cells in bone marrow using FACS analysis. Our data showed that significantly more CD44⁺ CD2^high Sca-1⁺ progenitor T and B-cells remained in the bone marrow of the SDF-1 group at 30 weeks of age than in that of the control group (Fig. 4). B220^high CD38^high progenitor B-cells also remained in the bone marrow of the SDF-1-Ig group (Fig. 5). SDF-1 was differentially expressed in progenitor cells, having been produced by the committed cells (CD34⁺/CD38⁺), and not the more primitive progenitor cells (CD34⁺/CD38⁻), whereas the CXCR4 receptor was equally expressed in both cell subsets.³⁷ Therefore, these results suggested that the anti SDF-1 antibodies might block progenitor cells from sensing each other and clustering together, as well as have an influence on their respective migratory and trafficking activities within the bone marrow microenvironment. B cells in which CD38, a marker of the progenitor B cell, was expressed were expressed more strongly in the bone marrow B cells (B220^low cells) of SDF-1-Ig group mice. This might have correlated with the population of IgM⁻ or IgD⁻ cells under B220^low cells; however, there were no changes in distribution among B220^low CD24^high, B220^high CD24^low, and B220^low BP-1⁺ cells between the SDF-1-Ig and Control-Ig groups. Therefore, we considered that the anti-SDF-1 antibodies did not have effect on B-cell maturation from late pro-B cells to pre-B cells.

In the SDF-1-Ig group, the proportion of progenitor B cells in bone marrow cells was higher than in the Control-Ig mice. However, in the mature B cells, which did not express CD43, the difference in population was not so obvious between the groups (Fig. 5). For the populations of both mature B cells in the late stage that expressed CD23 mature B cells that expressed the cell adhesion molecule CD38, the populations were similar in the SDF-1-Ig and Control-Ig groups. Accordingly, it was suggested that in NOD type mice like those in the Control-Ig group, maturation might advance rapidly from pre B cells or immature B cells to mature B cells or late mature B cells that express CD23.^38,39 It is also possible that mature B cells migrate to organs easily through the blood flow. It seems that the trafficking was restricted by SDF-1, because NOD mice injected with the anti-SDF-1 antibody maintained the same proportion of IgM⁺ IgD⁺ B220^high CD43⁻ CD38⁺ CD24⁺ CD23⁺ mature B cells. In the SDF-1 group mice, it was considered that progenitor B cells susceptible to autoantigen tolerance were not induced, because they had not yet matured rapidly to mature B cells, in contrast to the Control-Ig mice. Developmental arrest is an early outcome of antigen binding in immature B cells, as it blocks the acquisition of adhesion molecules and receptors important for B-cell migration and activation, through it is rapidly reversible by the removal of antigen.³⁸

Chemokines, by their chemoattraction properties, help lymphocytes to migrate from their site of origin to the periphery,^34,40 and possibly to sites of autoinflammation as well. In NOD mice, SDF-1 may be a chemokine that directly or indirectly affects the migration or trafficking of IgM⁺ IgD⁺ CD43⁻ B220^high CD24⁺ CD23⁺ matured B cells to the periphery. Down-modulation of SDF-1 by injection of anti-SDF-1 antibodies might have restricted the trafficking of matured B-cell subsets from the bone marrow to lymphoid organs, such as the spleen. Thus, we speculated that the total amount of autoreactive B cells that migrated to the islets of the pancreas was lower in the SDF-1 group, because of the reduced circulation of matured B cells from the bone marrow. As a result, sufficient mediator B cells might have been unavailable to activate T cells for infiltration to the islets and destroy beta-cells in the SDF-1-Ig group, as compared to the Control-Ig group. However, anti-SDF-1 antibodies could not bring any change to the trafficking of CD4, CD8, or macrophages in the spleen and thymus, or to lymphocyte infiltration of pancreatic islets at the stage of insulitis grade 2. This indicated that the effects of the anti-SDF-1 antibody in the development of diabetes in NOD mice could be weak or invisible to total T cells, and strong to autoreactive T cells at the periphery.

It was also considered that the B-cell differentiation observed in the SDF-1-Ig group mice might have been closer to normal. Previous investigators have reported unique roles for IgM and IgD in determining whether B-cell tolerance or activation occurred, because antigen binding to IgM alone or immature IgD B cells induced tolerance, whereas the binding of antigen by mature B cells generally provided stimulatory signals.⁴¹In vivo, the functions of IgD expression attributed to the engagement of sIgD vary from the acquisition of resistance to tolerance induction,^42–44 and the initiation of the B-cell response⁴⁵ to a role in B-cell memory.^46–48 Therefore, it is speculated that in the Control-Ig mice, with genes consisting mostly of the NOD type, the B-cell lineage from progenitor B cell to immature B cells, which are susceptible to tolerance, might have matured before clonal deletion and anergy. They may then develop into mature B cells, which are tolerance resistant, and rapidly migrate, dependent on SDF-1, to each organ, as in the case of B-cells with characteristics reactive to the autoantigen.

In the present study, we found that the anti-SDF antibody could not prevent lymphocytic infiltration of the salivary gland or dysproduction of saliva (Table 1). In recent studies, separate genes have been identified that contribute to the activation of effector T-cell clones for the development of IDDM and SS in model mouse of autoimmune diseases.^11,49,50 These diabetogenic genes identified are major histocompatibility complex (MHC) class II in NOD mice.^51,52 However, Robinson et al. indicated that the unique NOD MHC class II I-A^g7 is not essential for exocrine tissue autoimmunity in these mice.⁵³ The pathogenesis remains unclear, however, Ro/ss-A52 000 MW in humans and α-fodrin in mice were recently identified as organ-specific autoantigen molecules involved in SS.^11,54 In addition, it has been demonstrated that restricted T-cell clones (CD4⁺ Vβ8⁺ and CD4⁺ Vβ6⁺) are actively involved in autoimmune response against a particular antigen in the salivary glands of NOD mice.⁵⁴ Moreover, non-MHC loci appear also to contribute to diabetes and SS-like diseases in NOD mice.^52–54 In the present study, it was found that SS occurs independently of autoimmune diabetes in NOD mice. These findings indicate that the effects of chemokines may not be associated with effector T and B cells involved in the development of SS in this animal model; however, the chemokine or receptor may not have a corelation with the development of SS.

In summary, it was found that the chemokine SDF-1 may be an essential chemoattractor in trafficking of mature B cells from bone marrow to the periphery or inflammatory sites in the development of diabetes, and that it has an important role in the trafficking and homing of progenitor B cells. Furthermore, we showed that the anti-SDF antibody is effective in inhibiting diabetes, but not SS. We therefore suggest that SDF-1 is one of the diabetogenic elements in the development of diabetes.