The role of CD8+ T cells in the initiation of insulin-dependent diabetes mellitus

1762 B. Wang et al. Eur. J . Immunol. 1996.26: 1762-1769

Bo Wang, Antonio Gonzalez, Christophe Benoist and Diane Mathis

Institut de GCnCtique et de Biologie MolCculaire et Cellulaire (CNRS/ INSERMAJLP) Illkirch, C.U. de Strasbourg, France

The role of CD8+ Tcells in the initiation of insulin-dependent diabetes mellitus

While it is generally accepted that Tcells are critical for the development of diabetes in the non-obese diabetic (NOD) mouse, the precise functions of the CD4' and CD8' subsets remain ill-defined. Transfer experiments have provided evidence that CD4' cells are the disease initiators, provoking massive mononu- clear leukocyte infiltration into the pancreatic islets, while CD8' cells play an effector role, responsible for the final destruction of islet beta cells. It was sur- prising, then, to find that NOD mice carrying a null mutation at the p2- microglobulin (p2-p) locus, and thereby lacking major histocompatibility complex class I molecules and CD8' Tcells, developed neither insulitis nor diabetes. Here, we argue that the absence of insulitis in these animals results from their lack of CD8' cells because islet infiltration is also absent when NOD mice are treated with an anti-CD8 monoclonal antibody (mAb) at a young age. Interes- tingly, the anti-CD8 effect is only observed when the mAb is injected during a discrete age window - 2 to 5 weeks after birth. Transfer experiments indicate that the lack of CD8' cells during this period somehow alters the phenotype of CD4' cells, preventing them from expressing their insulitic potential. This is not because they are generally immuno-incompetent nor because they are generally more prone to differentiating into cells with Th2 characteristics. Given that neither the p2-p mutation nor anti-CD8 treatment affect insulitis in a T cell receptor transgenic (tg) mouse strain with a CD4' Tcell repertoire highly skewed for an anti-islet cell reactivity, the most straight-forward interpretation of these observations is that CDS' cells are required for effective priming and expansion of autoreactive CD4' cells.

1 Introduction

Insulin-dependent diabetes mellitus is an autoimmune disease, the result of a leukocyte attack on the beta cells of the islets of Langerhans of the pancreas (for reviews, see [l, 21). Although this disease is an ancient one, and one that is quite prevalent, we remain rather ignorant of its pathogenesis. We do not know what triggers it, have only a sketchy knowledge of the sequence of events unfolding during its progression, and have very little understanding of the genetic and environmental factors which influence its course.

One open question is the role played by particular T lymphocyte subsets. It is known that T cells are important for the development of diabetes, but precisely how CD4' and CD8' cells contribute to disease progression has remained a contentious issue. Several groups have shown that both subsets are required to provoke disease when splenocytes are transferred from diabetic NOD mice into neonatal,

[I 156041

Received April 1, 1996: in revised form May 9. 1996; accepted May 9, 1996.

Correspondence: Diane Mathis and Christophe Benoist, Institut de GCnCtique et de Biologie MolCculaire et Cellulaire, BP 163, F-67404 Illkirch Cedex. C.U. de Strasbourg, France Fax: + 33-88.65.32.46

Abbreviations: NOD: Non-obese diabetic p2-p: p2-Micro- globulin tg: Transgenic SPF Specific pathogen free

Key words: Diabetes / NOD mouse / Autoimmune disease / Mon- oclonal antibody therapy / CD8' Tcell

irradiated young, or genetically T cell-deficient NOD recipients [3-71. CD4' cells, alone, can invade the islets, but diabetes develops usually only when CD8' cells are also present; CD8' cells, alone, do not even infiltrate the islets [5-71. This led to the proposition that CD4' cells are the initiators of diabetes, launching the attack on the islets and recruiting CD8' cells, the disease effectors responsible for the final destruction of beta cells. This proposal was consistent with a body of other data: the observation that CD4' cells precede CD8' cells into the islets when whole splenocytes from diabetic donors are transferred into heal- thy recipients [8,9]; the finding that diabetes but not insulitis is reduced when adult NOD mice are injected with anti-MHC class I or anti-CD8 mAb [lo]; the isolation of CD8' lines and clones with in vitro and in vivo cytotoxic activity against beta islet cells [ll-141.

However, more recent results on NOD mice carrying a null mutation in the fi2-microglobulin (p2-p) gene did not sup- port this simple dichotomy [ 15-18]. These animals express almost no MHC class I molecules and have a paltry CD8' T cell compartment; class I1 molecule expression and CD4' cells appear normal. One would have predicted that they would exhibit insulitis but not diabetes; surprisingly, they showed neither. These results seemed to argue that CD8' cells play a critical role in disease initiation. Indeed, such a role has been suggested over the years by Lafferty and colleagues, on the basis of islet graft experiments [19, 201, and more recently by Janeway and collaborators. based on transfer experiments with CD8' Tcell clones [21]. An initiating function would also be more in line with observations on the human disease, since CD8' cells dom- inate the infiltrate in newly diagnosed patients [22-241 and are often highly activated in the blood [25].

0 VCH Verlagsgesellschaft mbH, D-69451 Weinheim. 1996

Eur. J . Immunol. 1996.26: 1762-1769 CD8' Tcells initiate diabetes 1763

Unfortunately, the interpretation of the results on p2-p- deficient mice is more complicated than first appreciated. Trivial explanations involving co-inheritance of linked genes have been largely, though not completely, ruled out [16-181. But the absence of insulitis could just as well be due to the lack of MHC class I molecules as to the dearth of CD8' cells. Loss of MHC class I molecules could potentially prevent selection of T cells other than the CD8' subset [26, 271, influence the pool of peptides avail- able for selection of CD4+ cells, alter B cell parameters, such as Ig turnover or isotype profiles [28, 291, or affect the display of other cell-surface receptors, e.g. the insulin receptor [30]. In this report, we employ an independent approach to establish that CD8+ Tcells are critical for the initiation of diabetes - injection of an anti-CD8 mAb into young NOD mice.

2 Materials and methods

2.1 Mice

NODLt mice were originally obtained from Dr. E. Leiter (Jackson Laboratories, Bar Harbor, ME); the BDC2.5 TcR transgenic (tg) mice have been described and are 17th generation backcross to NOD/Lt [31] ; P2-po/NOD mice have been reported [ 151 and are at the eighth generation backcross to NOD/Lt. Animals were bred in our specific pathogen-free (SPF) facility and were housed in our con- ventional facility for the duration of the experiments.

2.2 In vivo anti-CD8 treatment

The non-depleting anti-CD8 mAb, YTS105.18 [32] was produced as ascites in pristane-primed nude mice. The control mAb 7.B.10 was generated by fusion of LPS- activated rat splenocytes with the P3X63Ag8.653 line (ATCC CRL 1580), and screened for secretion of rat IgG (C. Ebel and P. Gerber, unpublished). The anti-CD8 mAb was titrated on adult NOD mice by one i.p. injection; 50 1.11 of ascites fluid was the minimum amount required for a complete blockade of CD8 molecules in peripheral lymphoid organs. The following protocol was routinely used: half of the littermates were injected i.p. with 50 p1 of anti- CD8 ascites twice (1 day apart) at 2 weeks of age; the other half received the same amount of control mAb ascites.

2.3 Histology

NOD mice were killed 10 weeks after the last injection of Ab, or at the age indicated in the figures; BDC2.5 TcR tg mice were examined 3 weeks after Ab injection. Pancreata were removed, fixed in Bouin's solution, embedded in paraffin, sectioned (85 pm), and stained with hematoxylin and eosin as previously described [33]. At least 50 islets per mouse were scored for insulitis as described [31]. Islets were found to be either free of infiltrates, or affected by peri-insulitis (if the infiltrates were limited to the connec- tive tissue around the islets), or by insulitis (if the islet itself was infiltrated).

2.4 T cell stimulation and neonatal transfers

Spleens were removed 10 weeks after mAb injection, single-cell suspensions prepared, and red blood cells lysed in 0.87% ammonium chloride. The cells were washed twice and resuspended at 2 X 10' celldm1 in RPMl 1640 medium supplemented with 10 YO heat-inactivated FCS, 1 mM sodium pyruvate, 1 mM glutamine, 50 pM 2-ME and 5 pg/ml ConA. After 48 h at 37"C, cells were washed and residual ConA was neutralized by 10 mg/ml of methyl- 2-~-mannopyranoside (Sigma). After two washings, they were resuspended at lo7 celldm1 in RPMl 1640 medium. Depletions was performed with mAb and low-Tox-M rab- bit complement: cells ( 107/ml) were incubated with super- natant of the anti-CD4 mAb RL172.4 [34], or of the anti- CD8 mAb 31.M [35], on ice for 30 min. After washing with RPMl 1640, cells were resuspended to the original volume in RPMl 1640 containing 1/10 volume of freshly reconsti- tuted complement and were incubated at 37 "C for 30 min. The cells were washed twice with RPMl 1640 and the pro- cedure was repeated twice. Depletion was assessed by flow cytometry using anti-CD4 (clone YTS 191.1, Caltag) and anti-CD8 (YTS169.4, Caltag) by two-color staining. Live cells were counted using trypan blue.

Neonatal NOD mice (within 24 h of birth) were injected i.p. with 3 x lo7 total spleen cells or 5 x lo7 spleen cells depleted of either CD4' or CD8' cells. For the co-transfer experiments, mice were injected i.p. with 5 x lo7 anti- CD4 or anti-CD8 depleted spleen cells from each of the donor mice. Mice were killed at 20 days of age and examined for insulitis.

2.5 Antibody responses

Mice were injected S.C. in the hind footpads with 50 pg of KLH emulsified in CFA, and were boosted 3 weeks later with 25 pg of KLH i.p. in standard saline. Sera were taken 12 days after the initial immunization to assess primary responses and 8 days following antigen re-challenge for secondary responses. Antigen-specific Ab titers were determined as described [36].

2.6 Cytokine production

Splenocytes were depleted of CD8' cells as described above. Cells were stimulated as previously described [27] in flat-bottom, 96-well plates coated with purified anti- CD3 mAb (KT3) in RPMI medium supplemented with 10% FCS, 50 pM 2-ME, 1 mM sodium pyruvate, 1 mM glutamine at 37°C. Supernatants were taken at 24 h to test for IL-2 production and at 40 h for IL-4 and IFN-y production. IL-2 was measured using the IL-Zdependent cell line CTLL-2. IL-4 and IFN-y were quantitated using ELISA, with paired mAb AN-18.17 and biotin-conjugated R4-6A2 for IFN-y (ATCC HB-170) and paired mAb BVD4-ID11 and biotin-conjugated BVD6-24G2 for IL-4 (Pharmingen, used according to the manufacturer's protocol).

1764 B. Wanget al. Eur. J . Immunol. 1996.26: 1762-1769

3 Results

3.1 Anti-CDB mAb treatment of young NOD mice inhibits the development of insulitis

As an alternative approach to assess the role of CD8+ Tcells in the early stages of diabetes, we evaluated the effect of treating young NOD mice with an anti-CD8 mAb. Littermates were injected with YTSlOS or a control mAb at 2 weeks of age, and their pancreata were removed and scored for insulitis at 12 weeks. As indicated in Fig. lA, treatment with antLCD8 severely inhibits the development of insulitis: most animals show no detectable inva- sion; a few exhibit light infiltration.

To interpret these results better, it is important to know the status of the CD8' Tcell compartment in treated animals. Thus, we injected cohorts of mice with anti-CD8 or the control mAb, removed their lymphoid organs at various times, and stained cells with an antLCD8 reagent which is not blocked by YTSlOS because it recognizes an independent epitope. Fig. 1B shows results on lymph node cells; similar data were obtained on cells from the blood and spleen (not shown). Anti-CD8 treatment of young mice leads to a rapid, but reversible, loss of CD8' cells: most disappear during the first week after injection, essentially all are gone by 2 weeks, they begin to reappear at 7 weeks, and reach high, though not quite normal, levels by 14 weeks. Staining with anti-rat Ig or with antLCD4 and anti- TcRaP confirmed that the anti-CD8 treatment does not just block or modulate surface CD8 molecules but actually results in the disappearance of CD8' cells from the lymphoid organs (not shown).

The reappearance of CD8+ Tcells beginning at 7 weeks after anti-CD8 treatment provoked us to examine insulitis at later times after injection. As indicated in Fig. lC, the anti-CD8 effect is long-lasting: most 20-week-old mice, i. e. taken 18 weeks after injection, show little or no insulitis. A few have heavily invaded islets and they probably corre- spond to those with light infiltration at 12 weeks (panel A).

Thus, anti-CD8 treatment of young NOD mice leads to long-term protection from insulitis, long after CD8' T cells have reappeared.

3.2 Treatment with anti-CDB must be during a discrete age window for effective inhibition of insulitis

Next we studied how the anti-CD8 effect is influenced by the age at which animals are treated. NOD mice aged 2, 3, 4, S or 7 weeks were injected with YTSlOS or the control mAb and their pancreata removed and scored for insulitis at 12 weeks of age. Interestingly, anti-CD8 inhibits insulitis when administrated at S weeks or earlier but has little or no effect when given at 7 weeks (Fig. 2A). This is true when the different-aged animals are all given the same dose of anti-CD8 or when adjustments in the dose are made for relative body weight (not shown).

Since mice in o u r NOD colony first develop visible insulitis at about 5 to 6 weeks, the question arises as to how our regimen of anti-CD8 treatment influences the progression

of already established insulitis. In the experiment depicted in Fig. 2B, 7-week-old littermates were divided into three groups: one was immediately killed and evaluated for insulitis; the second and third were treated with control mAb

, Ctl mAb 0 2 weeks anti-CDtY mAb 0 2 weeks

lnsulitis Q 12 weeks

0 5 10 15 20 Time after antibody injection (weeks)

Ctl mAb anti-CD8 mAb B 2 weeks 0 2 weeks

lnsulitis Q 20 weeks Figure I. Anti-CD8 treatment protects from insulitis. A: Early treatment with anti-CD8 mAb protects NOD mice from insulitis. Two-week-old NOD littermates were divided into two groups. One group was injected twice (at 1-day interval) with anti-CD8 mAb (YTSlOS) and the other with a control rat mAb. Ten weeks following the last injection, pancreatic islet infiltration was evaluated on hematoxylideosin-stained sections. Insulitis is displayed as the percentage of infiltrated islets per total islets examined for each mouse, scored as peri-insulitis (hatched bars) and insulitis (filled bars). B: Lymph node CD8' cells disappear temporarily after treatment. Lymph node cells from mice treated with anti-CD8 (filled circles) or control mAb (open circles) treated mice were stained at various time points after antibody injection with anti- CD4 and anti-CD8. c: Anti-CD8 treatment of young NOD mice gives long-term protection from insulitis. NOD mice were treated at 2 weeks of age and the presence of insulitis was evaluated at 20 weeks. Peri-insulitis (hatched bars) and insulitis (filled bars).

Eur. J. Immunol. 1996.26: 1762-1769

@ 100 1

CD8+ Tcells initiate diabetes 1765

e9 '0°1

I I U anti-CD8 @ week Ctl

lnsulitis Q 12 weeks

@ 100 -I

Ctl anti-CD8 lnsulitis lnsulitis

0 7 weeks Q 11 weeks

Figure 2 . Anti-CD8 treatment protects only young NOD mice from insulitis and has no effect on established insulitis. A: NOD mice were treated with anti-CD8 or control mAb at 2, 3, 4 , 5 or 7 weeks of age, and insulitis evaluated at 12 weeks; display as for Fig. 1. B: Seven-week-old NOD mice were treated with anti-CDS or control mAb. Insulitis was scored at the time of treatment, or at 11 weeks.

or YTSlOS and were killed and scored at 11 weeks of age. Insulitis is visible in most 7-week-old animals, and during the next 4 weeks it increases in average severity, whether the control or anti-CD8 reagent is administered.

Together, the data from these two experiments indicate that antLCD8 treatment is only effective when animals are injected during a discrete age window, corresponding to the period before insulitis first begins in our colony. Once insulitis is established, our protocol of anti-CD8 treatment neither clears the infiltrate nor substantially blocks its fur- ther progression.

3.3 Anti-CDS treatment alters the insulitic potential of CD4' T cells

Previous experiments showed that spleen cells from diabetic NOD mice are capable of rapidly inducing diabetes when transferred into neonatal, irradiated young, or genetically T cell-deficient NOD recipients [3-71. CD4' Tcells can home to and infiltrate the islets in the absence of CD8' cells, but the inverse is not true [5-71. Fig. 3A demonstrates that Con A-activated splenocytes from pre-

0

spleen

Ctl Treated Treated CD4s

0 '""1

p2 M+fO p2 MOIO

Figure 3. Lack of autoreactive CD4 cells in anti-CD8-treated mice. A: CD4' cells from normal non-diabetic NOD mice transfer insulitis after activation. Spleen cells from non-diabetic 10-week-old NOD mice were stimulated with Con A for 48 h, and depleted of CD8+ or CD4' cells (populations referred to as CD4s and CD8s, respectively, for simplicity; the transferred populations do contain residual B cells, however). Cells (with or without depletion, as indicated) were transferred i.p. to neonatal NOD mice, in which insulitis was scored 20 days after transfer. B: Two-week-old NOD mice were treated with antLCD8 (Treated) or with control mAb (Ctl), their splenocytes harvested ten weeks later and stimulated with Con A as above. These cultures were injected into neonatal NOD mice: 3 X lo7 total spleen cells from control Ab-treated mice (Ctl), 3 x lo7 total spleen cells from anti-CD8 treated mice (Treated), or 5 X lo7 spleen cells depleted of CD8' (CD4s) from treated mice. Insulitis was scored as above, 20 days after transfer. C: p2-p deficiency also leads to inactive CD4' cells. Ten-week-old NOD mice carrying a heterozygous (P2M+/o) or homozygous mutation in the p2-p locus (P2Mo/o) were used as donors of Con A-activated sDleen cells for transfer into neonatal NOD mice (3 x lo7 cells

diabetic animals from our NOD colony behave in an ana- i.'p.). Insulitis was scored as above, 20 days after transfer.

1766 B. Wang et al.

loo 1 Eur. J. Immunol. 1996.26: 1762-1769

CD4' cells from anti-CD8-treated mice remain totally incapable of initiating insulitis.

Thus, removal of the CD8' Tcells residing in 2-5-week- old NOD mice somehow influences the phenotype of CD4' cells, preventing them from acquiring insulitic potential. This change lasts for months and can not be reversed by CD8' cells developing later or by the CD8' population derived from regular NOD mice.

" u U U Ctl CD4s Treated CD4s Treated CD4s

+ Ctl CD8s

Figure 4. Normal CD8' cells cannot complement inactive CD4+ cells. Two-week-old NOD mice were treasted with antLCD8 (Treated) or with control mAb (Ctl), their splenocytes harvested 10 weeks later and stimulated with Con A as above. Neonatal recipients received 5 x lo7 spleen cells depleted of CD8+ from mice treated with control Ab (Ctl CD4s) or antLCD8 (Treated CD4s) or a mix of 5 X lo7 treated CD4s supplemented with 5 X 10' CD4-depleted spleen cells from control mice (Treated CD4sf Ctl CD8s). Insulitis was scored 20 days after transfer as above.

logous fashion: premature insulitis is observed in neonatal NOD recipients of total splenocytes or splenocytes depleted of CD8' cells (for the sake of simplicity, referred to in the figure and hereafter as the CD4' population) but not in recipients given splenocytes depleted of CD4' cells (referred to as the CD8' population). This provides a use- ful system for exploring the mechanism by which antLCD8 treatment of young NOD mice inhibits the development of insulitis. Does the block reflect a direct effect on Tcells or a more indirect effect, e .g . on routes of lymphocyte circulation or on islet beta or antigen-presenting cells? Even though they can not infiltrate the islets in situ, can Tcells from treated animals transfer insulitis after artificial activation in vitro?

In an initial experiment (Fig. 3B), we compared the behavior of spleen cells from control mAb- or YTS10.5-treated animals. In this and all other experiments discussed below, cells were taken 8-10 weeks after mAb injection in order to permit repopulation of the CD8 compartment. Strik- ingly, neither total nor CD4' Con A-activated splenocytes from anti-CD8-treated animals were capable of transferr- ing insulitis.

We then tested whether the spleen cells from NOD mice carrying the p2-p null mutation behave in an analogous fashion (Fig. 3C). Whole splenocytes from these animals (containing CD4' but essentially no CD8' Tcells) are also unable to transfer insulitis.

These results indicate that the CD4' T cells, themselves, are somehow inept, whether alone or accompanied by CD8' cells from treated animals. To see whether this incompetence can be overcome by association with CD8' cells from regular NOD mice, we performed the experiment depicted in Fig. 4. When CD8' Tcells from control Ab-treated animals are added just prior to transfer, the

3.4 The CD4' (or CDS') T cell compartment of anti- CDS-treated mice does not exhibit general functional defects

The incompetence of the CD4' Tcells from anti-CDX- treated mice in the transfer assay prompted us to look for general defects in the functioning of this compartment. First, we tested whether the CD4' cells can be stimulated in vivo or in vitro by assaying activation markers on cells taken directly from animals or subsequently cultured in the presence of Con A. No differences are observed in the staining profiles of gated CD4' cells from mice treated with control mAb or YTSlO5. This is true for early activation markers such as CD25 (Fig. 5A, upper) and CD69 (not shown) or late markers like CD44 (Fig. 5A, lower) and CD62L (not shown). Second, we compared the ability of control and anti-CD8-treated mice to make an antibody response against the Tcell-dependent antigen KLH. After a secondary KLH challenge, the two sets of animals make indistinguishable IgM and IgG anti-KLH responses (Fig. 5B).

We also examined the immuno-competence of the CD8' T cell compartment which appears during the weeks after anti-CD8 treatment (data not shown). NOD mice injected with antLCD8 at 2 weeks and left to recover until 10 weeks of age are able to mount a normal CTL response against influenza virus, assessed on either Kd- or Dh-expressing targets after a secondary challenge of cultured splenocytes. They are also capable of rejecting with normal kinet- ics grafts of tail skin from C57BL/6 donors.

3.5 In vitro activation of CD4' T cells from anti-CDS treated or B2-p-deficient mice does not lead to a diminished Thl and augmented Th2 response

Several lines of evidence argue that Thl cells promote diabetes in NOD mice; there have been some suggestions that Th2 cells inhibit diabetes, but this is more controversial (discussed in [37-391). Although the ThlRh2 duality is usually considered to come into play during the progression of insulitis to diabetes, it still seemed possible that the defect exhibited by CD4' cells from anti-CD8-treated and P2-p-deficient animals reflects an altered Th phenotype.

To test for a general shift in the Thl/Th2 profile, we activated in vitro CD4' splenocytes from anti-CD8-treated and P2-p-deficient NOD mice and the appropriate control animals, and measured the production of IFN-y and IL-4. As illustrated in Fig. 6 , the CD4' cells from mice with manipulated CD8' compartments actually produce more IFN-y than control animals. Anti-CD8-treated mice also

Eur. J. Immunol. 1996.26: 1762-1769 CD8' Tcells initiate diabetes 1767

Ctl Treated

t

v CD25

___)

CD25

w CD44

0 0

IgG IgM 0

- CD44

0

0

0 8

I 8

0 '

'5immiz& Ctl

Figure 5 . CD4+ cells in treated mice are not globally deficient. A: Two-week-old NOD mice were treated with anti-CD8 (Treated) or with control mAb (Ctl), their splenocytes harvested ten weeks later and stimulated with Con A for 24 h. Induction of CD25 (top panels) and of CD44 (bottom panels) was evaluated by flow cytometry on gated CD4' cells. Thin lines: unstimulated controls; thick lines: Con A stimulated. B: Antibody response to KLH. Ten weeks after treatment, mice treated with anti-CD8 or control mAb were immunized with KLH. Secondary antibody responses to KLH were determined by ELLSA with IgG- and IgM-specific reagents. Each point represents an individual mouse.

make more IL-4 than controls, but NOD mice carrying the p2-p null mutation make less than wild-type NOD animals.

Thus, the inability of CD4' T cells to initiate insulitis can not be explained by a simple, general shift in theirThlRh2 profile. It remains possible that more complex changes in cytokine expression are involved or that cells displaying particular specificities have an altered ThlRh2 proclivity.

0

O h 7 7 7 r -

Cell number (x 10-5)

s'$i&, 0

0 1 2 3 4 Cell number (x 105 )

Figure 6. Early anti-CD8 treatment does not induce a dominant Th2 phenotype. Splenocytes from 10-week-old NOD mice treated with antLCD8 (Treated) or with control mAb (Ctl) at 2 weeks of age, or 10 week-old NOD mice carrying a heterozygous (/32M+lo) or homozygous mutation in the p2-p locus (P2Mo/o), were stimulated with anti-CD3 after depletion of CD8' cells, and production of INF-y and IL-4 was determined by ELISA.

3.6 Anti-CDS treatment of young BDC2.5 TcR tg mice does not influence insulitis

As mentioned above, introduction of the (32-p null mutation completely blocked the appearance of insulitis in NOD mice [15-181. However, very different results were obtained with a strain of mice carrying the rearranged TcR genes from a diabetogenic Tcell clone. BDC2.5 TcR tg mice have a Tcell repertoire highly skewed for the transgene-encoded specificity, routinely develop a severe insulitis beginning at 3 weeks of age, and succumb to diabetes some months later [31]. Somewhat surprisingly, BDC2.5 TcR tg animals carrying the p2-p null mutation did show the usual pancreatic infiltrate at 5 weeks (J. Katz, C.B., D.M., unpublished data). We wondered whether NOD andTcR tg mice also respond differently to anti-CD8 treatment. As illustrated in Fig. 7, the very same condi- tions which drastically inhibit insulitis in the NOD system have no detectable effect in the TcR tg model. It may be

-- Ctl anti-CD8

@ 2 weeks @ 2 weeks

lnsulitis 8 5 weeks Figure 7. Anti-CD8 treatment does not affect insulitis in TcR transgenic mice. BDC 2.5 TcR transgenic mice were injected with anti-CD8 or control mAb at 2 weeks of age; pancreas sections were evaluated at 5 weeks and insulitis was scored. Peri-insulitis (hatched bars) and insulitis (filled bars) are shown as above.

1768 B. Wang et al.

worth pointing out that the transgenic mice were injected precisely at 14 days of age, an age when insulitis has never been observed.

Eur. J . Immunol. 1996.26: 1762-1769

4 Discussion

Our studies have demonstrated that injection of anti-CD8 mAb into young NOD mice at any time between 2 and 5 weeks of age inhibits the development of insulitis. Previ- ous experiments examining the effect of early anti-CD8 treatment failed to observe such an inhibition [40]. We can only surmise that the conflicting results reflect differences in the treatment protocols, e.g. the particular mAb employed, the dose administered, the number of injec- tions.

That early anti-CD8 treatment, and the resultant loss of CD8' Tcells, drastically inhibits insulitis in NOD mice confirms and extends prior reports that introduction of a p2-p null mutation onto the NOD genetic background pre- vents infiltration of the islets [15-181. The present results confirm that the phenotype is not due to linked genes brought in during the crosses. And they extend our understanding of the mechanism, supporting the notion that the phenotype results from the absence of CD8' T cells rather than to another consequence of the lack of MHC class I molecules.

How does the absence of CD8' cells during the early weeks of life inhibit insulitis on a long-term basis? Many studies have shown that CD4' cells from diabetic NOD mice can transfer insulitis (though not diabetes) in the absence of CD8' cells [5-71, and we confirm here that the same is true of in vitro activated CD4' cells from predia- betic NOD animals. Thus, one would have predicted that the CD4' cells residing in anti-CD8-treated mice would be able to invade the islets in situ and, after in vitro activation, would be able to transfer insulitis to young NOD recipients. Since neither of these occurred, we must con- clude that the absence of CD8' cells during this age window has somehow altered the properties of the CD4' T cell compartment, Yet, this compartment is not generally immuno-incompetent because CD4' cells appear normally activatable in vivo and in vitro and are fully capable of "helping" an anti-KLH IgM and IgG antibody response.

One might consider two possible scenarios for how CD8' Tcells in young NOD mice influence the phenotype of potentially insulitic CD4' cells. The first involves a specific response: some time during the 2-5-week window a neo- antigen is expressed, inducing CD8' cells to make an anti- islet cell response; the resulting disruption and inflamma- tion triggers priming of an anti-islet response by CD4' cells, leading to massive recruitment of leukocytes, eventually visible as insulitis. The novel antigen could be expressed on or in islet cells, or alternatively, could reside elsewhere but have sufficient homology to an islet cell antigen to induce a cross-reactive response. In theory, the neo-antigen could be synthesized by, or induced by, an infectious agent and this would fit well with many reports hinting at an infectious etiology for human diabetes [2]. However, this is not supported by the observation that the cleaner a NOD colony is the higher the incidence and the earlier the onset of diabetes, nor by the fact that germ-free

NOD mice develop diabetes ([14]; our unpublished results). The antigen is more likely, then, to be a newly synthesized product of an endogenous gene, perhaps turned on by the many physiological changes which take place at the time of weaning. An intriguing candidate would be an endogenous retrovirus gene product, given their acknowledged strain variability and reports of pecu- liarities of expression in the NOD strain [42, 431. In addi- tion, retroviral transcripts are often hormone-inducible so one could well imagine new products arising in the islet cells or elsewhere coincident with weaning.

The second possible scenario invokes a more general influence of CD8' Tcells on the phenotype of CD4' cells. It could just be that the presence of CD8' cells, as a population, influences the differentiation and/or activity of CD4' cells. An obvious candidate would be an effect on the Thl/ Th2 profile, given that CD8' cells are capable of secreting IFN-y and IL-4 and can become polarized into analogous Tcl/Tc2 effector populations [44,45]. However, other phe- notypic perturbations are possible as well, such as the modulation of homing behavior or adhesion properties under the influence of molecules secreted by CD8' cells. For example, the recently described chemokine, lympho- tactin, is a lymphocyte attractant and is secreted by CD8', but not CD4', Tcells [46]. There is some precedence for this notion of a general influence because certain abnor- malities in the behavior of the CD4' compartment in p2-p- deficient mice have been observed [47-491. Nonetheless, this explanation is somewhat unsatisfying because it fails to explain why CD8' cells of 2-5-week-old animals are required, i .e. why the CD8' cells which appear during the weeks after treatment are unable to reverse the deficiency in the CD4' compartment. It is easy to envision how older animals might differ in some specific response (not the same repertoire, not the same state of tolerance); it is more difficult to imagine how they could have different properties as a population (though perhaps the difference might be attributed to the fewer dividing cells or relatively recent thymic emigrants in older animals).

Any explanation for the anti-CD8 effect needs to account for results on the BDC2.5 TCR tg strain. Even transgenics incapable of rearranging endogenous TCRa genes and completely devoid of CD8' a : p T cells develop insulitis and diabetes [38]. Moreover, neither the 82-p null mutation (J. Katz, C.B., D.M.; unpublished results) nor anti- CD8 treatment (this report) inhibits insulitis in this strain. It may be that the requirement for CD8' cells is sur- mounted by the extremely high numbers of the potentially pathogenic specificity in the transgenics. During normal circulation, these cells could be activated by cognate or cross-reactive antigen frequently enough to provoke an effective response. This might also explain why CD4' cells are invasive in the absence of CD8' cells in transfer experiments - the CD4+ cells might have already been activated and expanded to a relative high precursor frequency.

Finally, it is important to recognize that our demonstration of a critical role for CD8' cells in initiating diabetes does not argue against an additional effector function. Transfer experiments have shown that CD8' cells are required for progression of insulitis to diabetes in the NOD system [5-71, although experiments employing large numbers of CD4' clones did not always show such a dependence [19,

Eur. J. Immunol. 1996.26: 1762-1769 CD8' Tcells initiate diabetes 1769

50,511. This issue has not been resolved to the satisfaction of all, but it remains entirely possible that disease progression involves three stages, dependent on CD8+, then CD4', and finally CD8' cells, again. While the experiments described in this paper do not address the existence of the second intervention by CD8+ cells, they argue strongly for the first.

We would like to thank Drs. H . Waldmann and R. Zinkernagel for the gift of antibodies, I? Michel and E Fischer for help with the mice, T Ding for the many sections, C . Ebel and F1 Gerber forpro- ducing the antibodies, C . Waltzinger for help with the flow cytometry. This work was supported by institute funds from the Institut National de la Sante' et de la Recherche Me'dicale, the Centre National de la Recherche Scientifique, the Centre Hospitalier Uni- versitaire Rkgional and by grants to DM and CB from the Juvenile Diabetes Foundation International and the Human Frontiers Sci- ence Program. B. W! was supported by a fellowship from the Uni- versite' Louis Pasteur de Strasbourg and A . G . by successive fellow- ships from the European Economic Community, the Centre National de la Recherche Scientifique and el Ministerio de Educa- cion y Ciencia.

5 References

1 Atkinson, M. A. and Maclaren, N. K., N. Engl. J . Med. 1994.

2 Bach, J. F., Endocrine Rev. 1994. 15: 516. 3 Bendelac, A., Carnaud, C., Boitard, C. and Bach, J . F., J .

Exp. Med. 1987. 166: 823. 4 Miller, B. J., Appel, M. C., O'Neil, J. and Wicker, L. S. , J.

Irnmunol. 1988. 140: 52. 5 Thivolet, C., Bendelac, A., Bedossa, P., Bach, J. F. and Car-

naud, C., J. lmmunol. 1991. 146: 85. 6 Yagi, H., Matsumoto, M., Kunimoto, K., Kawaguchi, J.,

Makino, S. and Harada, M., Eur. J. Immunol. 1992.22: 2387. 7 Christianson, S. W., Shultz, L. D. and Leiter, E. H., Diabetes

1993. 42: 44. 8 O'Reilly, L. A., Hutchings, P. R., Crocker, P. R. , Simpson, E.,

Lund, T., Kioussis, D. , Takei, E, Baird, J. and Cooke, A., Eur. J. Irnmunol. 1991. 21: 1171.

9 Bedossa, F?, Bendelac, A., Bach, J . F. and Carnaud, C., Eur. J. Immunol. 1989. 19: 1947.

10 Taki, T., Nagata, M., Ogawa, W., Hatamori, N., Hayakawa, M., Hari, J., Shii, K., Baba, S. and Yokono, K., Diabetes 1991. 40: 1203.

11 Nagata, M., Yokono, K., Hayakawa, M., Kakwase, Y., Hata- mori, N., Ogawa, W., Yonezawa, K., Shii, K. and Baba, S., J. Immunol. 1989. 143: 1155.

12 Hayakawa, M., Yokono, K., Nagata, M., Hatamori, N., Ogawa, W., Miki, A., Mizoguti, H. and Baba, S., Diabetes 1991. 40: 1210.

331: 1428.

13 Nagata, M. and Yoon, J.-W., Diabetes 1992. 41: 998. 14 Nagata, M., Santamaria, P., Kawamura, T., Utsugi, T. and

Yoon, J.-W., J. Immunol. 1994. 152: 2042. 15 Katz, J., Benoist, C. and Mathis, D., Eur. J . Imrnunol. 1993.

23: 3358. 16 Wicker, L. S., Leiter, E. H., Todd, J. A., Renjilian, R. J.,

Peterson, E., Fischer, P. A , , Podolin, P. L., Zijlstra, M., Jae- nisch, R. and Peterson, L. B., Diabetes 1994. 43: 500.

17 Serreze, D. V., Leiter, E. H. , Christianson, G. J., Greiner, D. and Roopenian, D. C., Diabetes 1994. 43: 505.

18 Sumida, T., Furukawa, M., Sakamoto, A., Namekawa, T., Maeda, T., Zijlstra, M., Iwamoto, I., Koike, T., Yoshida, S., Tomioka, H. andTaniguchi, M., Int. Immunol. 1994. 6: 1445.

19 Bradley, B. J . , Haskins, K., La Rosa, F. G. and Lafferty, K. J., Diabetes 1992. 41: 1603.

20 Wang, Y., Pontesilli, O., Gill, R. G., La Rosa, E G. and Laf- ferty, K. J., Proc. Natl. Acad. Sci. USA 1991. 88: 527.

21 Wong, E S., Visintin, I., Wen, L., Flavell, R. A. and Janeway, C. A. , J. Exp. Med. 1996. 183: 67.

22 Bottazzo, G. F., Dean, B. M., McNally, J . M., Mackay, E. H., Swift, P. G. E and Gamble, D. R., N . Engl. J . Med. 1985.313: 353.

23 Itoh, N., Hanafusa, T., Miyazaki, A., Miyagawa, J. I . , Yama- gata, K., Yamamoto, K., Waguri, M., Imagawa, A., Tamura, S. , Inada, M., Kawata, S., Tarui, S. , Kono, N. and Matsu- zawa, Y., J. Clin. Invest. 1993. 92: 2313.

24 Somoza, N., Vargas, F., Roura-Mir, C., Vives-Pi, M., Fernandez-Figueras, M. T., Ariza, A., Gomis, R., Bragado, R., Marti, M., Jaraquemada, D. and Pujol-Borrel, R., J. Imrnunol. 1994.153: 1360.

25 Hehrnke, B., Michaelis, D., Gens, E., Laube, F. and Kohnert, K.-D., Diabetes 1995. 44: 1414.

26 Bendelac, A., Killeen, N., Littman, D. R. and Schwartz, R. H., Science 1994. 263: 1774.

27 Couez, D., Malissen, M., Buferne, M., Schmitt-Verhulst, A.- M. and Malissen, B., Int. Immunol. 1991. 3: 719.

28 Ghetie, V., Hubbard, J. G. , Kim, J.-K., Tsen, M.-F., Lee, Y. and Ward, E . S., Eur. J . Immunol. 1996. 26: 690.

29 Spriggs, M. K., Koller, B. H. , Sato, T., Morrissey, P. J . , Fans- low, W. C., Smithies, O., Voice, R. F., Widmer, M. B. and Maliszewski, C. R., Proc. Natl. Acad. Sci. USA 1992.89: 6070.

30 Edidin, M., Immunol. Today 1988. 9: 218. 31 Katz, J. D. , Wang, B., Haskins, K., Benoist, C. and Mathis,

32 Cobbold, S. P., Martin, G. and Waldmann, H., Eur. J. Immu-

33 Bohme, J. , Schuhbaur, B., Kanagawa, O., Benoist, C. and

34 Ceredig, R., Lowenthal, J. W., Nabholz, M. and MacDonald,

35 Sarmiento, M., Glasebrook, A. L. and Fitch, F. W., J . Immu-

36 Lemeur, M., Gerlinger, P., Benoist, C. and Mathis, D., Nature

37 Liblau, R. S. , Singer, S. M. and McDevitt, H. O., Immunol.

38 Katz, J. D. , Benoist, C. and Mathis, D. , Science 1995. 268:

39 Charlton, B. and Lafferty, K. J., Current Biol. 1995. 7: 793. 40 Hayward, A. R., Cobbold, S. P., Waldmann, H., Cooke, A.

41 Pozzilli, P., Signore, A., Williams, A. J. K. and Beales, P. E.,

42 Leiter, E. H. , FASEB J . 1989. 3: 2231. 43 Suenaga, K. and Yoon, J.-W., Diabetes 1988. 37: 1722. 44 Seder, R. A. and Le Gros, G. G., J . Exp. Med. 1995. 181: 5. 45 Sad, S., Marcotte, R. and Mosmann, T. R., Immunity 1995.2:

271. 46 Kelner, G. S., Kennedy, J., Bacon, K. B., Kleyensteuber, S.,

Largaespada, D. A., Jenkins, N. A., Copeland, N. G., Bazan, J. F., Moore, K. W., Schall, T. J. and Zlotnik, A. , Science 1994. 266: 1395.

47 Marusic-Galesic, S., Udaka, K. and Walden, P., Eur. J. lmmunol. 1993. 23: 3115.

48 Muller, D., Koller, B. H. , Whitton, L., Lapan, K. E., Brig- man, K. K. and Frelinger, J . A., Science 1992.255: 1576.

49 Quinn, D. G. , Zajac, A. J., Frelinger, J. A. and Muller, D., Int. Immunol. 1993. 5: 1193.

SO Peterson, J. D. and Haskins, K., Diabetes 1996. 45: 328. 51 Daniel, D., Gill, R . G., Schlott, N. and Wegmann, D., Eur. J .

D., Cell 1993. 74: 1089.

nol. 1990. 20: 2747.

Mathis, D., Science 1990. 249: 293.

H. R., Nature 1985. 314: 98.

nol. 1980. 125: 2665.

1985. 316: 38.

Today 1995. 16: 34.

1185.

and Simpson, E. , J . Autoimmunity 1988. 1: 91.

Immunol. Today 1993.14: 193.

Immunoll995.25: 1056.

Documents

The role of CD8+ T cells in the initiation of insulin-dependent diabetes mellitus