Abstract In type 1 diabetes mellitus (T1DM), the

processes which control the recruitment of immune

cells into pancreatic islets are poorly defined. Complex

interactions involving adhesion molecules, chemokines

and chemokine receptors may facilitate this process.

The chemokine, monocyte chemoattractant protein-1

(MCP-1), previously shown to be important in leuko-

cyte trafficking in other disease systems, may be a key

participant in the early influx of blood-borne immune

cells into islets during T1DM. In the non-obese dia-

betic (NOD) mouse, the expression of MCP-1 protein

has not been demonstrated. We employed dual-label

immunohistochemistry to examine the intra-islet

expression, distribution and cellular source of MCP-1

in the NOD mouse following cyclophosphamide

administration. NOD mice were treated with cyclo-

phosphamide at day 72–73 and MCP-1 expression

studied at days 0, 4, 7, 11 and 14 after treatment and

comparisons were made between age-matched NOD

mice treated with diluent and non-diabetes-prone

CD-1 mice. Pancreatic expression of MCP-1 was also

examined in NOD mice at various stages of sponta-

neous diabetes. In the cyclophosphamide group at day

0, MCP-1 immunolabelling was present in selective

peri-islet macrophages but declined at day 4. It

increased slightly at day 7 but was more marked from

day 11, irrespective of diabetes development. The

pattern of MCP-1 expression in macrophages was dif-

ferent over time in both the cyclophosphamide and

control groups. In the cyclophosphamide group, there

was a change over time with an increase at day 11. In

the control group, there was little evidence of change

over time. There was no significant difference in the

mean percentage of MCP-1 positive macrophages

between the cyclophosphamide-treated diabetic and

non-diabetic mice. During spontaneous diabetes in the

NOD mouse, only a few peri-islet MCP-1 cells

appeared at day 45. These became more numerous

from day 65 but were absent at diabetes onset. We

speculate that a proportion of early islet-infiltrating

macrophages which express MCP-1 may attract addi-

tional lymphocytes and macrophages into the early

inflamed islets and intensify the process of insulitis.

Keywords MCP-1 Æ NOD mouse ÆImmunohistochemistry Æ Diabetes

Introduction

The onset of human type 1 diabetes mellitus (T1DM) is

preceded by a prolonged asymptomatic pre-diabetic

period during which various beta cell-directed immune

processes are evoked. These processes include pro-

gressive infiltration of pancreatic islets by immune cells

Y. Bai Æ R. Chai Æ S. Reddy (&)School of Biological Sciences, University of Auckland,Private Bag, 92019 Auckland, New Zealande-mail: s.reddy@auckland.ac.nz

E. RobinsonDepartment of Epidemiology and Biostatistics, Universityof Auckland, Private Bag, 92019 Auckland, New Zealand

J. M. RossDepartment of Radiology with Anatomy, University ofAuckland, Private Bag, 92019 Auckland, New Zealand

S. ReddyDepartment of Paediatrics, University of Auckland, PrivateBag, 92019 Auckland, New Zealand

J Mol Hist (2006) 37:101–113

DOI 10.1007/s10735-006-9045-6

123

ORIGINAL PAPER

Immunohistochemical study of monocyte chemoattractantprotein-1 in the pancreas of NOD mice followingcyclophosphamide administration and during spontaneousdiabetes

Yan Bai Æ Elizabeth Robinson Æ Ryan Chai ÆJacqueline M. Ross Æ Shiva Reddy

Received: 11 May 2006 / Accepted: 28 June 2006 / Published online: 29 July 2006� Springer Science+Business Media B.V. 2006

and intra-islet release of beta cell toxic molecules,

culminating in beta cell destruction and insulin insuf-

ficiency (Eisenbarth et al. 1987; Foulis et al. 1986;

Mathis et al. 2001). Immunohistochemistry of pancre-

atic tissues at diagnosis of diabetes demonstrated

infiltration of islets by CD8 cells, macrophages and B

cells and with IgG deposits (Bottazzo et al. 1985). This

immune cell infiltrate, referred to as insulitis, is an

important pathological hallmark of type 1 diabetes

(Bottazzo et al. 1985; Foulis et al. 1986). However, the

mechanisms by which blood-borne immune cells are

directed towards pancreatic islets during early T1DM

and their subsequent intra-islet expansion are poorly

understood. Various adhesion molecules expressed by

endothelial cells and by blood trafficking lymphocytes

may act in concert with a special class of molecules

known as chemokines, their cognate receptors and

certain beta cell signals and thereby facilitate leukocyte

migration into the islet (Boring et al. 1999; Frigerio

et al. 2002; Bendall 2005).

Chemokines belong to the superfamily of cytokines,

with molecular weights between 8 kD and 10 kD. They

elicit important pleotropic effects such as leukocyte

chemotaxis, inflammation, infection, immunity and

angiogenesis (Boring et al. 1999). Chemokines are

subdivided into four subfamilies (C, CC, CXC and

CX3C) based on the location of the first conserved

cysteine residues in the amino terminus (Bendall 2005).

The CXC family has two cysteine residues separated

by a single amino acid. The largest group is the CC

family where two cysteine residues are immediately

adjacent to each other. Chemokine receptors are

members of a class of seven-transmembrane G-protein

coupled receptors and form part of a larger super-

family which includes receptors for hormones, neuro-

transmitters, paracrine substances and inflammatory

mediators (Bendall 2005).

Monocyte chemoattractant protein-1 (MCP-1), also

known as CCL2, is a member of the CC chemokine

family and is produced by a wide variety of cells

including lymphocytes, monocytes, endothelial cells,

mesangial cells and fibroblasts in response to proin-

flammatory stimuli (Oppenheim et al. 1991). It is a

potent chemoattractant for monocytes and T cells and

activates leukocytes in several inflammatory diseases,

in certain cancers and during renal injury (Fuentes

et al. 1995; Guazzone et al. 2003; Hilgers et al. 2000;

Tesch et al. 1999; Yamamoto et al. 2003).

Non-obese diabetic (NOD) mice develop T1DM

spontaneously and are a close model for the human

disease (Kikutani and Makino 1992). In this

model, pancreatic islets are progressively invaded by

macrophages, T cells and B cells from 35 days to

40 days of age, with subsequent onset of diabetes be-

tween 90 days and 250 days of age. In the NOD mouse,

chemokines have been implicated in immune cell traf-

ficking into islets (Atkinson and Wilson 2002; Cameron

et al. 2000; Frigerio et al. 2002; Kim et al. 2002; Mea-

gher et al. 2003). Isolated islets from NOD mice, ex-

press MCP-1 mRNA as early as 2 weeks of age and

peak by 8 weeks (Chen et al. 2001). Exposure of rat

beta cells to interleukin-1b (IL-1b) resulted in induc-

tion of MCP-1 mRNA and MCP-1 protein release

(Chen et al. 2001). Transgenic expression of MCP-1 in

mouse beta cells with a (C57Bl/6 · C3H)F2 back-

ground resulted in mostly monocyte accumulation

within the islets but diabetes failed to develop (Grewal

et al. 1997).

Although MCP-1 mRNA has been shown to be

present in isolated islets of the NOD mouse, direct

evidence for its expression at the protein level and its

cellular sources are lacking (Chen et al. 2001). In this

study, the expression of MCP-1 protein was investi-

gated in pancreatic sections of the NOD mouse at

various time-points following acceleration of diabetes

with cyclophosphamide. Dual- and triple-label immu-

nohistochemical techniques were used to establish and

quantify the cellular sources of the protein at these

various time-points. Expression of the chemokine was

also studied at various ages during spontaneous dia-

betes in the NOD mouse.

Materials and methods

Animals

A colony of NOD mice was established at the Animal

Resources Unit of the School of Biological Sciences,

The University of Auckland. This colony originated

from a nucleus of 3 breeding pairs obtained from Dr A

Merriman at the University of Otago, Dunedin, New

Zealand. The University of Otago colony was estab-

lished from breeding pairs obtained originally from

Jackson Laboratories, Bar Harbor, Maine, USA. The

present Auckland colony is maintained on a standard

autoclaved diet (Harlan Teklad Global 18% Protein

Rodent Diet, code 2018S, United Kingdom) and sterile

water ad libitum and has a current rate of spontaneous

diabetes of approximately 80% among females be-

tween the ages of 90 days and 250 days. Diabetes in

NOD mice is defined as the presence of a hypergly-

caemic value exceeding 12 mM on 3 consecutive days

in the non-fasting state. In non-diabetic NOD mice, the

non-fasting blood glucose levels range from 4 mM to

6 mM.

102 J Mol Hist (2006) 37:101–113

123

Newly weaned (day 21) non-diabetes-prone CD-1

mice were obtained from the Animal Resources Unit

of the Faculty of Medical and Health Sciences, The

University of Auckland and maintained on the same

diet as NOD mice.

Study groups, administration of cyclophosphamide,

tissue collection and cryosectioning

Female NOD mice were obtained from several

breeding pairs, weaned at day 21 and allocated at

random into two groups as follows:

Group 1 Standard mouse chow plus drinking water

(control group; n = 24)3Group 2 Standard mouse chow plus drinking water

(cyclophosphamide group; n = 54)

At days 73–74, all but 6 mice from Group 2 were

injected with cyclophosphamide (Sigma, St Louis, Mo.,

USA; 300 mg/kg body weight) by the intra-peritoneal

route while mice from Group 1 were injected with

diluent (sterile water) as described previously (Reddy

et al. 2001a, 2003). From the cyclophosphamide group,

6 mice were killed at days 0 (without cyclophospha-

mide administration), 4 and 7. At day 11 (cyclophos-

phamide group), 14/36 remaining mice which had

developed diabetes and 6 mice which were diabetes-

free, were killed. At day 14, all remaining mice were

killed (11 with diabetes and 5 without diabetes).

From the control group, 6 mice were killed at each

time point (days 4, 7, 11 and 14). Following sacrifice,

almost the entire pancreas, along with a small piece of

the adjoining spleen, was harvested (see below). Each

pancreas was then divided immediately into halves,

one of which was snap-frozen (splenic half) and stored

at – 80�C while the other fixed in Bouin’s solution

(duodenal portion). Prior to tissue removal, a small

sample of blood (approximately 10 ll) was taken from

the tip of the tail for blood glucose measurement with a

blood glucose meter (Accu-Check Blood Glucose

Meter, Roche Diagnostics, USA).

Spontaneous NOD mouse group

Pancreata from female NOD mice were collected as

described above at days 21, 30, 34, 40, 45, 65 and at

onset of spontaneous diabetes (diabetes onset between

100 days and 120 days; 3 NOD mice per age).

CD-1 mice

Pancreata were collected from non-diabetes-prone

female CD-1 mice at days 73, 76, 79, 83 and 86

(corresponding to days 0, 4, 7, 11 and 14 from the NOD

cyclophosphamide group; 3 mice/time-point) and at

days 35, 45 and 65 (3 mice/age group) and processed as

detailed above.

Frozen pancreatic tissues were serially cryosec-

tioned (6 lm) from different levels, adhered to super-

frost slides, air-dried briefly, fixed in cold acetone and

stored at – 20 �C until required for immunohisto-

chemistry. Adjacent sections were routinely stained by

haematoxylin and eosin (H&E) to verify the presence

of islets and their histopathological status.

Primary antibodies, non-immune IgG, and normal

sera

Affinity-purified rabbit polyclonal antibodies to highly

purified recombinant mouse MCP-1 (98% pure) were

obtained from PeproTech, Rehovat, Israel. By Wes-

tern blotting, the suppliers have shown monospecificity

of the antibody for recombinant mouse MCP-1 under

reducing and non-reducing conditions. ELISA testing

of anti-murine MCP-1 and the biotinylated anti-murine

MCP-1 against a variety of chemokines including

murine CTACK, C10, eotaxin, MCP-5, MIP-1a, MIP-

1b, MIP-1c, MIP-3b and RANTES at 50 ng/ml showed

no cross-reaction. This antibody was previously em-

ployed to immunostain MCP-1 in cryosections of

mouse pancreas (Frigerio et al., 2002). Other studies

have shown that rabbit polyclonal antibodies to rat

MCP-1 supplied by PeproTech also show immunohis-

tochemical specificity for the homologous chemokine

(Sung et al. 2002).

Rat monoclonal antibodies to mouse CD3 T cells

were obtained from Dr H Georgiou, Walter and Eliza

Hall Institute of Medical Research, Melbourne, Aus-

tralia. Rat monoclonal antibodies against mouse mac-

rophages (recognize CD11b or Mac-1 marker on

macrophages, clone 170) were purchased from Serotec,

Oxford, UK. Guinea pig anti-bovine insulin serum was

available in this laboratory (Reddy et al. 2003).

For immunohistochemistry, all primary antisera

were titrated to give maximal immunohistochemical

reactivity. Normal sera from the goat, sheep, donkey,

rabbit, guinea pig, rat and mouse were available in this

laboratory.

Immunohistochemical localization of MCP-1

An indirect immunofluorescence procedure was em-

ployed for the immunohistochemical localization of

MCP-1 and was adapted from a protocol for the

localization of cleaved caspase-3 (Reddy et al. 2003).

J Mol Hist (2006) 37:101–113 103

123

Sections were routinely washed in excess phosphate-

buffered saline (PBS), pH 7.5, containing 0.3% sapo-

nin (Sigma) at the end of each incubation step. Sapo-

nin-PBS buffer was also employed as the diluent for all

immunohistochemical reagents for MCP-1 immuno-

histochemistry, unless stated otherwise.

Prior to immunolabelling, cryosections were re-fixed

in cold acetone, equilibrated in PBS-saponin and

incubated with 5% normal donkey serum for 1 h at

37�C. After washing, sections were incubated with

rabbit anti-MCP-1 (1:100 diluted in 5% normal donkey

serum in saponin-PBS) for 6 h at 37�C and then for

18 h at 4�C, followed by further washing. Donkey anti-

rabbit IgG-biotin (1:100; Jackson ImmunoResearch

Laboratories, West Grove, Pa., USA) was then added

and incubated for 1 h at 37�C. Sections were reacted

with streptavidin-Alexa 568 (1:200; Molecular Probes,

Eugene, Ore., USA) for 1 h at 37�C. Following wash-

ing, sections were mounted for light and/or confocal

microscopy (see below).

Dual- and triple-immunolabelling

Serial sections were first stained for MCP-1 as de-

scribed above and then equilibrated in PBS. Each

section was then incubated with rat antibodies to either

Mac-1 (1:10 in PBS), CD3 (undiluted) or guinea pig

anti-insulin serum (1:1,000 diluted in 5% normal don-

key serum in PBS) overnight at 4�C. Sections were

washed in PBS and reacted with goat anti-rat IgG-

cyanin-2 (1:50, Rocklands Laboratories, Gilbertsville,

Pa., USA) or donkey anti-guinea pig IgG-FITC (1:50,

Jackson ImmunoResearch Laboratories). They were

washed and mounted for light and confocal micro-

scopy. Triple immunostaining was carried out by

ximmunolabelling for MCP-1 first as above (streptavi-

din-Alexa 568 fluorochrome), then for insulin (FITC

fluorochrome) and finally for macrophages or CD3 T

cells (donkey anti-mouse IgG-cyanin-5, 1:100, Jackson

Immunoresearch Laboratories). Sections were moun-

ted for confocal microscopy.

Determination of the mean percentage of MCP-1

positive macrophages per islet in the

cyclophosphamide group

For the cyclophosphamide study, the total number of

macrophages within peri- and intra-islet regions was

enumerated by either direct microscopic counting or

following photography in at least 10 islets from dif-

ferent levels of the pancreas. Macrophages which

showed positive immunolabelling for MCP-1 were

also counted. Results were expressed as the mean

percentage of MCP-1 positive macrophages per islet

± SEM for each time-point.

Immunolabelling of MCP-1 and interleukin-1b (IL-

1b)

Two adjacent sections from day 11 cyclophosphamide-

treated mice (diabetic) were mounted on separate

slides and each section was immunolabelled in parallel

for either MCP-1 as described in this study or for IL-1bas described previously (Reddy et al. 2001b). For IL-1bimmunolabelling, 0.3% saponin-PBS was employed as

both diluent and wash buffer. Sections were equili-

brated in 0.3% saponin-PBS, blocked with 5% normal

donkey serum and then incubated with goat anti-

mouse IL-1b (1:75) for 60 h at 4�C. This was followed

by incubation with donkey anti-goat IgG-biotin and

finally with streptavidin-Alexa 568 (1:200). The stain-

ing patterns of the two proteins were compared within

the same islets in the adjacent sections.

Microscopy and imaging

Immunohistochemically stained sections and those

stained by H&E were examined with a Zeiss fluores-

cence microscope equipped with a digital camera.

Single- and dual-labelled images were recorded, and

then processed with Adobe Photoshop 8.0 (Adobe

Systems, Mountain View, Calif., USA).

Selected immunofluorescent sections were also

examined by confocal laser scanning microscopy (TCS

4D Leica, Heidelberg, Germany), as reported previ-

ously (Reddy et al. 2003). Separate as well as merged

images were saved in TIFF file format and then

assembled in Adobe Photoshop 8.0.

Controls for immunohistochemistry

In the immunohistochemical procedure, all primary

antibodies were substituted with the corresponding

non-immune IgG or serum at equivalent dilutions.

Primary (for two-step staining) and secondary steps (for

three-step staining) were omitted. In the double- and

triple-immunolabelling protocols, species-incompatible

secondary antibodies (FITC-, cyanin-2 or biotin-linked)

were also employed to verify immunohistochemical

specificity. Rabbit anti-MCP-1 IgG (5 ng/ll) was pre-

absorbed with highly purified recombinant MCP-1

(5 ng/ll and 15 ng/ll; PeproTech) in 0.3% saponin-PBS

buffer for 48 h at 4�C. Rabbit anti-MCP-1 (5 ng/ll)

without MCP-1 antigen was incubated in parallel. The

antigen-antibody mixture was centrifuged before use in

the immunohistochemical protocol for MCP-1. In this

104 J Mol Hist (2006) 37:101–113

123

protocol, the resulting immunolabelling was compared

in adjacent sections, following exposure to MCP-1 pri-

mary antibody with and without preabsorption.

Statistical analyses

Islets were classified as containing MCP-1 positive

macrophages or not. A logistic regression model with

the animal as the random effect was used to investigate

changes over the 14 day period and whether they dif-

fered between cyclophosphamide-injected and control

NOD mice. A second model was used to investigate

differences in the presence of macrophages expressing

MCP-1 between cyclophosphamide-treated mice which

developed diabetes and those which did not.

Results

Development of cyclophosphamide-induced

diabetes and islet pathology

In the cyclophosphamide group (Group 2), diabetes

developed in 14/36 NOD mice at day 11 and in 11/16

NOD mice at day 14. Diabetes did not develop in

diluent-treated mice during the same period (Group 1).

The severity of insulitis and its trend at various time-

points after cyclophosphamide administration were

similar to our previous reports (Reddy et al. 1999,

2001a; present data not shown). In the cyclophospha-

mide-treated animals, intra-islet immune cell pheno-

types within the insulitis region were predominantly

CD4 and CD8 cells and macrophages. In this group,

macrophage and T cell numbers increased markedly in

intra-islet areas at days 11 and 14 and were also ob-

served in exocrine regions as reported previously

(Reddy et al. 1999; present data not shown).

Immunohistochemical controls

The immunohistochemical monospecificity of anti-

MCP-1 employed in this study was shown by a variety

of control studies. Immunostaining for MCP-1 was

absent when the primary antibody was replaced with

diluent or normal sera (Fig. 1a–c). Suppression of

immunostaining for MCP-1 was observed when the

primary antibody was preabsorbed with MCP-1 antigen

(Fig. 1d–e). The use of species-incompatible secondary

antibodies also resulted in an absence of immunola-

belling (results not shown). In the dual-staining proto-

col for MCP-1 with macrophages, T cells and insulin,

omission of one of the primary antibodies did not result

in non-specific labelling (results not shown).

Fig. 1 Specificity of MCP-1 immunohistochemical protocol.Photomicrographs showing the same islet in 3 adjacent sectionssubjected to (a): Standard MCP-1 immunohistochemical proto-col; (b): Substitution of MCP-1 primary antibody with diluent(5% normal donkey serum in 0.3% saponin-PBS) and (c):Substitution of donkey anti-rabbit IgG-biotin with diluent. (d):

Islet from day 11 cyclophosphamide-treated mouse with strongimmunolabelling for MCP-1 while (e) shows marked suppressionof MCP-1 immunolabelling in the same islet from an adjacentsection, following preabsorption of primary antibody with excessantigen (MCP-1). Scale bars: 50 microns

J Mol Hist (2006) 37:101–113 105

123

MCP-1 expression, co-localization and its intra-islet

spatial relationship in the cyclophosphamide group

In pancreatic sections from female CD-1 mice age-

matched to the NOD cyclophosphamide group (days

73, 76, 79, 83 and 86) and spontaneous diabetes group

(days 35, 45 and 65), MCP-1 immunolabelling was ei-

ther absent or very weakly expressed in a few cells

located in the islet periphery which were likely to be

resident macrophages (Fig. 2a, b). However, parallel

Fig. 2 Photomicrographs of islets from CD-1 and cyclophospha-mide administered NOD mice immunolabelled for MCP-1. (a,b): Day 83 normal CD-1 mouse showing absence of MCP-1immunolabelling in an islet (a); (b) is the same islet as (a)subsequently stained by H&E. (c): An islet from an NOD mouse11 days after cyclophosphamide administration (83 days old)stained by H&E and shows advanced infiltration of the islet by

immune cells (arrowheads) and with a highly reduced number ofendocrine cells (arrows). (d–f): Day 11 cyclophosphamide isletdual-labelled for MCP-1 (d) and macrophages (e). Noteimmunolabelling of MCP-1 in macrophages is located mostly inthe periphery of the islet. (f): Same islet as in (d) and (e) butfrom an adjacent section stained by H&E. Note that the islet isalmost totally occupied by immune cells. Scale bars: 50 microns

Fig. 3 Photomicrographs of islets from day 0 and day 4cyclophosphamide-treated NOD mice showing merged imagesfollowing dual-labelling for MCP-1 (red) and macrophages (a, d,

green), insulin (b, e, green) or CD3 T cells (c, f, green). Arrowsin (a) point to macrophages which express MCP-1. Cy:cyclophosphamide; MF: macrophage. Scale bars: 50 microns

106 J Mol Hist (2006) 37:101–113

123

studies in NOD mice from day 11 cyclophosphamide

group (diabetic) showed intense staining of MCP-1

positive cells within the islet periphery and sometimes

in the intra-islet regions (Fig. 2d, e). Islets at this stage

usually showed severe insulitis (Fig. 2c, f).

In NOD mice at day 0 (day 72–73), a few peri-islet

cells corresponding to macrophages were positive for

MCP-1 (Fig. 3a). The expression of MCP-1 was not

observed in beta cells or T cells (Fig. 3b, c). At day 4

after cyclophosphamide, the number of MCP-1 cells

within the islets showed a marked decline, being

present in only a few macrophages in the periphery of

some islets (Fig. 3d–f). This pattern was similar in day

4 mice treated with diluent (results not shown). At

day 7 after cyclophosphamide, there was a notable

increase in MCP-1 positive cells located in peri-islet

Fig. 4 Photomicrographs ofislets from day 7 and day 11cyclophosphamide-treatedNOD mice dual-and triple-labelled for MCP-1 (red),macrophages (green or cyan),CD3 T cells (cyan) and insulin(green). (a): day 7, MCP-1alone, confocal microscopy;note positive MCP-1 cells inthe peri-islet region. (b):Same islet as (a), confocalimage, MCP-1 + macrophages + insulin;arrows point to MCP-1positive macrophages; (c):Same islet as (a, b) from anadjacent section, triple-labelled for MCP-1 + insulin+ CD3 T cells, confocalimage; note absence of MCP-1 in CD3 T cells and insulincells. (d): Day 7 islet showingtriple-labelling for MCP-1 +insulin + macrophages,confocal image, arrows pointto MCP-1 positivemacrophages. (e): Day 11cyclophosphamide-treatedand diabetic, showing an isletimmunolabelled for MCP-1.(f): Numerous MCP-1positive macrophages in theextra-islet area of the samesection as (e); (arrows). (g, h):Higher magnification ofMCP-1 positive cells withinthe rectangular enclosure in(e) showing cytoplasmicimmunolabelling of thechemokine (g) in a proportionof macrophages (h). D:Diabetic, Cy:cyclophosphamide. Scalebars: 20 microns except (f) 50microns

J Mol Hist (2006) 37:101–113 107

123

and occasionally in intra-islet regions. Dual- and triple-

labelling and confocal microscopy confirmed that these

cells were macrophages (Fig. 4a–d). At days 11 and 14,

after cyclophosphamide administration, there was a

marked increase in the number of MCP-1 immunola-

belled cells, both in the peri-and intra-islet regions and

in exocrine areas including the peri-vascular and sinu-

soidal space (Fig. 4e-h and Fig. 5a–l). Dual-labelling

identified a proportion of macrophages in the three

regions which were positive for MCP-1 (Fig. 4e, f; Fig.

5a–l). Higher magnification images revealed that MCP-

1 immunolabelling was cytoplasmic and often punctu-

ate (Fig. 4e, g, h). In the diluent-treated group, MCP-1

immunolabelling at days 7, 11 and 14 was similar to day

0 (data not shown).

In the cyclophosphamide group, qualitative analysis

showed that there was a marked variation in the num-

ber of islets which expressed MCP-1. At days 0, 4 and 7,

approximately 30–50% of the islets contained at least

one MCP-1 immunolabelled cell per islet in both the

cyclophosphamide and diluent-treated groups. At days

11 and 14 following cyclophosphamide, there was a

marked increase in the mean percentage of islets con-

taining MCP-1 positive cells (approximately 70–90% of

Fig. 5 Photomicrographs of islets from day 11 and day 14cyclophosphamide-treated NOD mice showing merged imagesafter dual-labelling for MCP-1 (red) and either macrophages (a,d, g, j, green), insulin (b, e, h, k, green) or CD3 T cells (c, f, i, l,

green). Arrows in (a, d, g, i) denote macrophages which expressMCP-1. D: Diabetic, ND: non-diabetic, Cy: cyclophosphamide;MF: macrophage. Scale bars: 50 microns

108 J Mol Hist (2006) 37:101–113

123

islets contained at least 1 MCP-1 positive cell). The

number of MCP-1 positive cells per islet in day 14 dil-

uent-treated mice was similar to that seen in the earlier

stages of diluent-treated NOD mice (data not shown).

Statistical analyses of MCP-1 expression during

cyclophosphamide administration

The number of macrophages expressing MCP-1 chan-

ged over time in the cyclophosphamide group but not

in the control group (P = 0.002; Fig. 6). In the cyclo-

phosphamide group, there was a change over time

(P = 0.007) and in particular there was an increase at

day 11. In the control group, there was little change

over time (P = 0.07). No significant difference was

found in the mean percentage of MCP-1 positive

macrophages between cyclophosphamide-treated mice

which developed diabetes and those which did not

(P = 0.6). However, there was a difference between

the time-points, with the chemokine more likely to be

present at day 11 than at day 14 (P = 0.046).

Immunolabelling for MCP-1 and IL-1b

Immunolabelling of adjacent sections at day 11 from

the cyclophosphamide group showed that MCP-1 and

IL-1b positive cells were differentially distributed

within the islets, suggesting the two proteins were

present in different cell types or areas (Fig. 7).

MCP-1 expression at various stages of spontaneous

diabetes in the NOD mouse

In young NOD mice prior to early insulitis (day 30–35)

and during minimum insulitis (day 40), MCP-1 positive

cells were absent. At day 45, some islets showed a small

number of MCP-1 positive cells in the islet periphery

(Fig. 8a–d). The number of MCP-1 positive cells within

the islets showed an increase at day 65 (Fig. 9a–e). At

diabetes onset, despite the presence of advanced

insulitis and the presence of numerous macrophages in

the peri-islet region, MCP-1 expression was absent

(Fig. 9f–j).

Discussion

Chemokines are chemotactic cytokines which bind to

their receptors and direct leukocyte migration towards

inflammatory sites. Monocyte chemoattractant pro-

tein-1 was one of the first chemokines attributed with

this function during several disease states (Boring et al.

1999). We speculated that limited injury to the beta cell

during initial stages of type 1 diabetes may result in

increased MCP-1 expression within the islets, followed

by recruitment of blood-borne leukocytes into the is-

lets. To gain further insights on the role of chemokines

during diabetes, we applied an immunohistochemical

technique to identify the cellular source/s of MCP-1 in

the islets of the NOD mouse in situ.

Our present studies in the cyclophosphamide model

show that MCP-1 was expressed almost exclusively in a

proportion of macrophages and was absent in the islets

Fig. 6 Mean percentage ± SEM MCP-1 positive macrophagesper islet at various stages of cyclophosphamide-treated and age-matched diluent-treated NOD mice. Open bars: diluent-treatedgroup; solid bars: cyclophosphamide group. D: diabetic, cyclo-phosphamide-treated; ND: non-diabetic, cyclophosphamide-treated.

Fig. 7 Photomicrographs ofan islet from a day 11cyclophosphamide-treatedNOD mouse immunolabelledfor MCP-1 (a) and IL-1b (b)in adjacent sections andshowing the same islet. Notethat the distribution of thetwo proteins within the islet isnot co-incident. Scale bars: 50microns

J Mol Hist (2006) 37:101–113 109

123

of age-matched non-diabetes-prone CD-1 mice. In the

cyclophosphamide group at day 0, only a few macro-

phages showed immunolabelling for MCP-1 and these

were usually scattered in the islet periphery. This de-

clined at day 4 and is consistent with our previous

observations of a decline in insulitis at this stage

(Reddy et al. 2003). However, MCP-1 positive mac-

rophages became more numerous at day 7, often

exhibiting stronger immunolabelling than at earlier

time-points. At subsequent time points, MCP-1

expression within the islets became more extensive and

concomitant with advancing insulitis. The absence of a

substantial difference in the overall pattern of MCP-1

expression in the cyclophosphamide group between the

diabetic and non-diabetic mice at days 11 and 14 sug-

gests that expression may be dependent on insulitis

rather than diabetes. This is in accord with a previous

study where transgenic expression of MCP-1 in beta

cells of mice with a (C57Bl/6 · C3H)F2 background

resulted in a monocyte-rich insulitis but not diabetes

(Grewal et al. 1997). However, in that transgenic

model, the insulitis region consisted predominantly of

F4/80+ monocytes with only minor populations of CD4,

CD8 and B220 cells. This pattern of immune cell

phenotype is markedly different from the islet immune

cell subpopulations previously reported in NOD mice

(Reddy et al. 1995).

The cytoplasmic and often punctuate immunola-

belling of MCP-1 in macrophages strongly suggests

that these cells are an important source of the

chemokine and that the macrophage immunolabelling

was not a consequence of chemokine uptake from

neighbouring islet or immune cells. However, other

techniques such as in situ hybridization would be nec-

essary to support this conclusion.

The present findings indicate that during spontane-

ous and cyclophosphamide-induced diabetes, the

number of islets which express MCP-1 increases with

advancing age, except at the onset of spontaneous

diabetes. This appears to parallel the increase in the

severity of insulitis. However, the absence of MCP-1 in

insulitis-rich islets at the onset of spontaneous diabetes

suggests that a minimum number of beta cells may be

necessary for MCP-1 expression. The intra-islet loca-

tion of MCP-1 positive cells was more extensive in the

cyclophosphamide model from day 11 when beta cell

loss appears to intensify. The expression of MCP-1 was

seen in only a proportion of macrophages. This may be

due to the known inherent heterogeneity of macro-

phage subpopulations within the intra-islet and exo-

crine areas of the NOD mouse (Rosmalen et al. 2000).

The absence of MCP-1 immunolabelling in beta

cells and T cells shown in this study is noteworthy.

Previous studies showed that mRNA for MCP-1 was

detectable in isolated islets of NOD mice from as early

as 2 weeks of age (Chen et al. 2001). In addition,

MCP-1 mRNA increased in rat and human beta cells

following exposure to IL-1b (Chen et al. 2001). The

reasons for the difference between our findings and

those of Chen et al. in the NOD mouse are unclear but

Fig. 8 Photomicrographs ofislets from a day 45 NODmouse dual-labelled forMCP-1 (a) and insulin (b). (c)and (d): Same islet fromadjacent sectionsimmunolabelled formacrophages (c) and stainedby H&E (d). Note only a fewcells express MCP-1 (arrows).MF: macrophages; Ins:insulin. Scale bars: 50 microns

110 J Mol Hist (2006) 37:101–113

123

may be due to the increased sensitivity of real time

PCR compared with immunohistochemistry (employed

in this study). However, it is known that the mRNA

and protein levels are not always concordant. The

MCP-1 gene expression studies of Chen et al. utilized

freshly isolated islets from NOD mice. There is a

possibility that residual macrophages attached to

freshly isolated islets during the isolation process may

have contributed to islet MCP-1 mRNA. In a separate

study, in situ hybridization showed MCP-1 signals in

cells corresponding to infiltrating mononuclear cells

following transplantation of allogeneic mouse islets

under the kidney capsule of recipient mice rendered

diabetic with streptozotocin (Schroppel et al. 2004).

Other studies showed that exposure of mouse islets in

culture or the NIT-1 beta cell line to IL-1b, interferon-

c (IFN-c) and tumour necrosis factor-a (TNF-a), re-

sulted in increased levels of mRNA for MCP-1,

RANTES (regulated upon activation, normal T cell

expressed and secreted) and IP-10 (Frigerio et al.

2002). Our present immunohistochemical findings

showing expression of MCP-1 in macrophages concur

with the findings of Frigerio et al. (2002). These

investigators employed a mouse model of insulitis and

utilized the same MCP-1 antibodies as those used in

our present study (from PeproTech Inc). They dem-

onstrated that MCP-1 immunolabelling was present in

cells corresponding to peri-islet macrophages but not

beta cells (Frigerio et al. 2002).

Immunohistochemistry and in situ hybridization

showed MCP-1 to be constitutively expressed in hu-

man pancreatic beta cells and in isolated islets. The

levels increased following exposure to proinflammato-

ry cytokine and lipopolysaccharide (Piemonti et al.

2002). However islets from Balb/c mice show ex-

tremely low levels of constitutive MCP-1 mRNA

expression (Chen et al. 2001). These studies suggest

that major species differences may exist between the

human and mouse regarding the cellular sources and

level of expression of MCP-1 in islets in vivo.

Our present immunohistochemical findings showing

macrophages as the predominant source of MCP-1 in

infiltrated islets of NOD mice support previous

findings in other relevant immune-mediated disease

models. In atherosclerotic lesions of humans, immu-

nohistochemistry showed that a small subset of

macrophages in macrophage-rich areas expressed

MCP-1 (Yla-Herttuala et al. 1991). Other related

chemokines, such as macrophage inflammatory pro-

teins (MIP-1a and MIP-1b), were first identified as

products of stimulated macrophages (Sherry et al.

Fig. 9 Photomicrographs of islets showing immunolabelling forMCP-1 in the NOD mouse. (a–e): Day 65 NOD mouse; (f–j):Newly diabetic NOD mouse (a): Islet with MCP-1 labelling(arrows); (b): Same islet as (a) dual-labelled for insulin (green).Note absence of MCP-1 in beta cells. (c, d): Adjacent sectionsshowing the same islet as in (a, b) dual-labelled for MCP-1 + macrophages (c) and MCP-1 + CD3 T cells (d). (e): Sameislet from an adjacent section stained by H&E. (f, g): An isletdual-labelled for MCP-1 (f) and insulin (g). Note absence ofMCP-1 and insulin cells. (h, i): Same islet as in (f, g) fromadjacent sections, immunolabelled for macrophages (h) and CD3T cells (i) while (j) is the same islet from an adjacent sectionstained by H&E. MF: macrophages; Ins: insulin. Scale bars: 50microns

J Mol Hist (2006) 37:101–113 111

123

1992). In addition, RANTES and MCP-1 were first

reported as products of activated T cells (Schall et al.

1988). A number of microbial agents are known to

stimulate beta chemokine production by macrophages,

including lipopolysaccharide and certain viruses such

as HIV-1 (Sherry et al. 1992; Verani et al. 1997;

Kornbluth et al. 1998). In the MLR-Faslpr mouse

model of autoimmunity, immunohistochemical exami-

nation of kidney tissue showed predominant MCP-1

expression in epithelial podocytes and lesser amounts

in epithelial cells and macrophages of crescents (Tesch

et al. 1999). In a rat model of hypertension and during

bleomycin-induced scleroderma, MCP-1 is expressed

in a small proportion of macrophages as well as in

other cell-types (Hilgers et al. 2000; Yamamoto et al.

2003).

In conclusion, the current study shows that a pro-

portion of islet macrophages are the predominant

source of MCP-1 in the NOD mouse following cyclo-

phosphamide-treatment and during spontaneous

diabetes. We speculate that some islet-infiltrating

macrophages may express MCP-1 at an early stage in

response to specific signals. Such MCP-1 positive

macrophages may then attract additional lymphocytes

and monocytes/macrophages into the inflamed islets,

enhance immune cell ‘‘build-up’’ and augment the rate

of beta cell destruction during T1DM. Further studies

are required to understand more fully the complex

signalling network which underpins the development

of insulitis during diabetes. Such studies may provide

new experimental paradigms for developing targeted

strategies aimed at inhibiting the process of insulitis

during early human T1DM.

Acknowledgements Financial assistance from the AucklandMedical Research Foundation, the Child Health ResearchFoundation of New Zealand and the Maurice and Phyllis PaykelTrust is gratefully acknowledged. We thank Ms Beryl Davy andMs Lorraine Rolston for histological assistance and Mr VernonTintinger for maintaining the NOD mouse colony.

References

Atkinson MA, Wilson SB (2002) Fatal attraction: chemokinesand type 1 diabetes. J Clin Invest 110:1611–1613

Bendall L (2005) Chemokines and their receptors in disease.Histol Histopathol 20:907–926

Boring L, Charo IF, Rollins BJ (1999) MCP-1 in human disease:insights gained from animal models. In: Hebert CA (ed).Chemokines in disease, biology and clinical research. Hu-mana Press, Totowa New Jersey, pp 53–65

Bottazzo GF, Dean BM, McNally JM, MacKay EH, Swift PG,Gamble DR (1985) In situ characterization of autoimmunephenomena and expression of HLA molecules in the pan-creas in diabetic insulitis. N Engl J Med 313:353–360

Cameron MJ, Arreaza GA, Grattan M, Meagher C, Sharif S,Burdick MD, Strieter RM, Cook DN, Delovitch TL (2000)Differential expression of CC chemokines and the CCR5receptor in the pancreas is associated with progression totype 1 diabetes. J Immunol 165:1102–1110

Chen M-C, Proost P, Gysemans C, Mathieu C, Eizirik DL(2001) Monocyte chemoattractant protein-1 is expressed inpancreatic islets from prediabetic NOD mice and in inter-leukin-1b-exposed human and rat pancreatic islet cells.Diabetologia 44:325–332

Eisenbarth GS, Connelly J, Soeldner JS (1987) The ‘‘natural’’history of type 1 diabetes. Diabetes Metab Rev 3:873–891

Foulis AK, Liddle CN, Farquharson MA, Richmond JA, WeirRS (1986) The histopathology of the pancreas in Type 1(insulin-dependent) diabetes mellitus: a 25-year review ofdeaths in patients under 20 years of age in the UnitedKingdom. Diabetologia 29:267–274

Frigerio S, Junt T, Lu B, Gerard C, Zumsteg U, Hollander GA,Piali L (2002) b cells are responsible for CXCR3-mediatedT-cell infiltration in insulitis. Nat Med 8:1414–1420

Fuentes ME, Durham SK, Swerdel MR, Lewin AC, Barton DS,Megill JR, Bravo R, Lira SA (1995) Controlled recruitmentof monocytes and macrophages to specific organs throughtransgenic expression of monocyte chemoattractant protein-1. J Immunol 155:5769–5776

Grewal IS, Rutledge BJ, Fiorillo JA, Gu L, Gladue RP, FlavellRA, Rollins BJ (1997) Transgenic monocyte chemoattrac-tant protein-1 (MCP-1) in pancreatic islets producesmonocyte-rich insulitis without diabetes. J Immunol159:401–408

Guazzone VA, Rival C, Denduchis B, Lustig L (2003) Monocytechemoattractant protein-1 (MCP-1/CCL2) in experimentalautoimmune orchitis. J Reprod Immunol 60:143–157

Hilgers KF, Hartner A, Porst M, Mai M, Wittmann M, Hugo C,Ganten D, Geiger H, Veelken R, Mann JF (2000) Monocytechemoattractant protein-1 and macrophage infiltration inhypertensive kidney injury. Kidney Int 58:2408–2419

Kikutani H, Makino S (1992) The murine autoimmune dia-betes model: NOD and related strains. Adv Immunol51:285–322

Kim SH, Cleary MM, Fox HS, Chantry D, Sarvetnick N (2002)CCR4-bearing T cells participate in autoimmune diabetes. JClin Invest 110:1675–1686

Kornbluth RS, Kee K, Richman DD (1998) CD40 ligand(CD154) stimulation of macrophages to produce HIV-1-suppressive b-chemokines. Proc Natl Acad Sci USA95:5205–5210

Mathis D, Vence L, Benoist C (2001) b-Cell death during pro-gression to diabetes. Nature 414:792–798

Meagher C, Sharif S, Hussain S, Cameron MJ, Arreaza GA,Delovitch TL (2003) Cytokines and Chemokines in thepathogenesis of murine type 1 diabetes. In: Santamaria P(ed) Cytokines and chemokines in autoimmune disease.Kluwer Academic/Plenum Publishers, pp 133–158

Oppenheim JJ, Zachariae CO, Mukaida N, Matsusima K (1991)Properties of the novel proinflammatory supergene ‘inter-crine’ cytokine family. Ann Rev Immunol 9:617–648

Piemonti L, Leone BE, Nano R, Saccani A, Monti P, Maffi P,Bianchi G, Sica A, Peri G, Melzi R, Aldrighetti L, Secchi A,Di Carlo VD, Allavena P, Bertuzzi F (2002) Human pan-creatic islets produce and secrete MCP-1/CCL2: relevancein human islet transplantation. Diabetes 51:55–65

Reddy S, Wu D, Swinney C, Elliott RB (1995) Immunohisto-chemical analyses of pancreatic macrophages and CD4 andCD8 T cell subsets prior to and following diabetes in theNOD mouse. Pancreas 11:16–25

112 J Mol Hist (2006) 37:101–113

123

Reddy S, Yip S, Karanam M, Poole CA, Ross JM (1999) Animmunohistochemical study of macrophage influx and the co-localization of inducible nitric oxide synthase in the pancreasof non-obese diabetic (NOD) mice during disease accelera-tion with cyclophosphamide. Histochem J 31:303–314

Reddy S, Karanam M, Robinson E (2001a) Prevention ofcyclophosphamide-induced accelerated diabetes in theNOD mouse by nicotinamide or a soy protein-based infantformula. Int J Exp Diabetes Res 1:299–213

Reddy S, Young M, Ginn S (2001b) Immunoexpression ofinterleukin-1b in pancreatic islets of NOD mice duringcyclophosphamide-accelerated diabetes: co-localization inmacrophages and endocrine cells and its attenuation withoral nicotinamide. Histochem J 33:317–327

Reddy S, Bradley J, Ginn S, Pathipathi P, Ross JM (2003)Immunohistochemical study of caspase-3-expressing cellswithin the pancreas of non-obese diabetic mice duringcyclophosphamide-accelerated diabetes. Histochem CellBiol 119:451–461

Rosmalen JG, Martin T, Dobbs C, Voerman JS, Drexhage HA,Haskins K, Leenen PJ (2000) Subsets of macrophages anddendritic cells in nonobese diabetic mouse pancreaticinflammatory infiltrates: correlation with the developmentof diabetes. Lab Invest 80:23–30

Schall TJ, Jongstra J, Dyer BJ, Jorgensen J, Clayberger C, DavisMM, Krensky AM (1988) A human T cell-specific moleculeis a member of a new gene family. J Immunol 141:1018–1025

Schroppel B, Zhang N, Chen P, Zang W, Chen D, Hudkins KL,Kuziel WA, Sung R, Bromberg JS, Murphy B (2004) Dif-

ferential expression of chemokines and chemokine recep-tors in murine islet allografts: the role of CCR2 and CCR5signaling pathways. J Am Soc Nephrol 15:1853–1861

Sherry B, Horii Y, Manogue KR, Widmer U, Cerami A (1992)Macrophage inflammatory proteins 1 and 2: an overview.Cytokines 4:117–130

Sung FL, Zhu TY, Au-Yeung KKW, Siow YL, Karmin O (2002)Enhanced MCP-1expression during ischemia/reperfusioninjury is mediated by oxidative stress and NF-j B. Kid Int62:1160–1170

Tesch GH, Maifert S, Schwarting A, Rolins BJ, Kelley R (1999)Monocyte chemoattractant protein 1-dependent leukocyticinfiltrates are responsible for autoimmune disease in MRL-Faslpr mice. J Exp Med 190:1813–1824

Verani A, Scarlatti G, Comar M, Tresoldi E, Polo S, Giacca M,Lusso P, Siccardi AG, Vercellli D (1997) C-C chemokinesreleased by lipopolysaccharide (LPS)-stimulated humanmacrophages suppress HIV-1 infection in both macrophagesand T cells. J Exp Med 185:805–816

Yamamoto T, Nishioka K (2003) Role of monocyte chemoattr-actant protein-1 and its receptor, CCR-2, in the pathogen-esis of bleomycin-induced scleroderma. J Invest Dermatol121:510–516

Yla-Herttuala S, Lipton BA, Rosenfeld ME, Sarkioja T,Yoshimura T, Leonard EJ, Witztum JL, Steinberg D (1991)Expression of monocyte chemoattractant protein 1 in mac-rophage-rich areas of human and rabbit atheroscleroticlesions. Proc Natl Acad Sci USA 188:5252–5256

J Mol Hist (2006) 37:101–113 113

123