TNF-α Up-regulates Renal MIF Expression in Rat Crescentic Glomerulonephritis

TNF-a

Up-regulates Renal

MIF

Expression

in Rat

Crescentic

Glomerulonephritis

Hui

Y. Lan,*

Niansheng

Yang,*

Christine

Metz,t Wei

Mu,*

Qing Song,* David

J.

Nikolic-Paterson,* Michael Bacher,t

Richard

Bucala,t and Robert

C.

Atkins*

*Department

of

Nephrology, Monash Medical

Centre, 246

Clayton

Road, Victoria 3168,

Australia

tThe

Picower

Institute for Medical Research,

350

Community Drive,

Manhasset,

New

York

11030,

U.S.A.

ABSTRACT

Background: Macrophage migration inhibitory factor (MIF) is apotentproinflammatorymediator that partic-ipates in the pathogenesis of endotoxemia and experi-mental crescenticglomerulonephritis. However, very lit-tle isknown about howMIFproduction isregulated in disease.Wethereforeexamined whethertumornecrosis factora(TNF-a), aknown inducer ofMIFexpressionby macrophages in vitro, up-regulates local and systemic MIT expressioninamacrophage-mediated rat model of crescenticglomerulonephritis.

Materials and Methods: Anti-glomerular basement membrane (GBM) glomerulonephritis was induced in groups of six primed rats. Animals were treated with 1

mg/kg soluble TNF-a receptor (TNFbp) or saline from thetimeof disease induction until theywere killed on Days 1, 7, or14.Renal MIF expression was assessed byin situ hybridization, immunohistochemistry, and ELISA, andcompared with macrophage accumulation and indi-cesof renaldamage.

Results: Although TNFbp treatment on Day 1 of the disease hadonlyapartialeffect upon theup-regulation

ofglomerularMWFexpression,onDays7 to 14italmost completely abrogated the increase in glomerular and interstitial MWFmRNA and protein expression. In addi-tion, TNFbptreatmentsignificantly inhibitedMIF secre-tionby culturedglomeruli and reducedserum MIF lev-els. The inhibitionof renalMIFexpressionwasparalleled

bya significant inhibition ofglomerularand interstitial macrophage infiltration (p < 0.001 versussaline treat-ed),asignificantsuppressionofrenal injury (proteinuria andserum creatinine), and amarked reduction in his-tologic damage (glomerular hypercellularity, crescent

formation, and interstitial fibrosis; all p < 0.01 versus salinetreated).

Conclusions: Thisstudydemonstrates for the firsttime thatTNF-a up-regulates local MIF expression byboth

infiltratingmacrophages and residentkidney cellsinrat crescentic glomerulonephritis. Inaddition, TNF-a regu-lates systemic MIF production. Thus, TNF-a, together withMIW, mayplayapathological rolein immunologi-callyinduced renal disease.

INTRODUCTION

Macrophage migration inhibitory factor (MIF)

wasoriginally described as aproduct of activated Tcells thatinhibits themigrationofmacrophages in vitro (1,2). Now, however, it is evident that

MIF is not exclusively produced by T cells. Re-cent studies demonstrate that MIF is

constitu-Addresscorrespondence and reprintrequeststo:Hui Y.

Lan, DepartmentofNephrology,Monash MedicalCentre,

246Clayton Road, Clayton, Victoria3168,Australia.

136

tivelyexpressed byavariety of celltypes, includ-ing monocytes/macrophages, anterior pituitary cells, and renal tubularepithelial cells (3-5). An-tibody-blocking studies have shownakey

patho-genic role for increasedMIFproductionin exper-imental endotoxemia (6). Glucocorticoids atlow levelsup-regulate MIFexpression via a

counter-regulatory mechanism, whereby MIF can then

act tooverridetheglucocorticoideffects (7,8).In

addition, in vitro studies have shown that the proinflammatorycytokinestumornecrosisfactor

H. Y. Lan etal.: TNF-aYUpregulates MIFinGlomerulonephritis

a (TNF-a) and interferon y (IFN--y) up-regulate macrophage MIF production, and that con-versely, MIF induces production of TNF-a and IFN-y (4,9), which suggests the presence of an amplifying proinflammatory loop response. However, there is little in vivo evidence for the functional significance of this amplification path-way.

We have recently shown that MIF produc-tion is markedly up-regulated in a rat model of crescenticglomerulonephritis. IncreasedMIF ex-pression correlated significantly with macro-phageinfiltration, loss of renal function, and de-velopment of renal lesions, including glomerular crescent formation (5). Since TNF-a has been implicated in the pathogenesis of this disease (10,1 1), we sought to determine whether TNF-a is aphysiological inducer ofMIFproduction dur-ing the disease process. This was examined by blocking the action of TNF-a in rat crescentic glomerulonephritis by treatment with TNF bind-ingprotein (TNFbp), a soluble form of recombi-nant human TNF-a receptor type 1, and by as-sessing the effect uponMIFexpression and renal injury.

MATERIALS

AND

METHODS

Nephrotoxic Serum

Rabbit anti-rat glomerular basement membrane (GBM) nephrotoxic serum was raised by re-peatedimmunizationof New Zealand white rab-bits with purified, particulate rat GBM. The sera was pooled, decomplemented at 560C for 30 min, absorbed against rat red blood cells at 40C

overnight, and stored at -200C.

TNFBinding Protein

A dimeric form of recombinant human TNF-a receptor type I, termed the TNF binding protein (TNFbp), was generously supplied by Amgen, Colorado.

Experimental Design

Accelerated anti-GBM glomerulonephritis was induced in inbred male Sprague-Dawley rats (150-200 g) as previously described (12,13). Briefly, animals were immunized s.c. with 5 mg normal rabbitIgGinFreund's complete adjuvant and injected i.v. with 10 ml/kg rabbit anti-rat GBM serum (12.5 mg IgG/ml) 5 days later

(termedDay 0). Groups of six rats wereinjected intraperitoneally with 1 mg/kg body weight of TNFbpeverysecond day from2hr before admin-istrationofanti-GBM serum until being killedon Days 1, 7, or 14. Groups of six control animals had the same schedule, but received saline (ve-hicle) in place of TNFbp. In addition, one group of six normal rats was also studied. The study was approved by the Animal Experimentation Ethics Committee of Monash Medical Centre.

Renal Function

Blood samples and 24-hr urine collections were taken on Days 0, 1, 7, and 14. Urinary protein excretion wasdeterminedwith the Manual Pon-ceau Red method. Serum creatinine was mea-sured with the standard Jaffe rate reaction (Al-kaline picrate). All analyses were performed at the Department of Biochemistry, Monash Med-ical Centre.

Histopathology

Kidney tissues forhistological examination were fixed in 10% formalin and 4-gm paraffin sec-tions were stained with hematoxylin and eosin or periodic acid-Shiff (PAS). Coded slides were assessedas previously described (12,13). Twenty glomerularcross sectionsperanimalwere exam-ined under high power to determine glomerular hypercellularity (0-3) and the percentage of glo-meruli withcrescentformation, while the degree of cortical tubulointerstitial fibrosis (0-4) was also determined.

Immunohistochemistry

Single and double immunostaining was per-formed on

4-Am

paraffin sections of formalin-fixed kidney with a newly developed method of microwave-based, multiple immunoenzymatic staining (14). Briefly, sections were dewaxed, placed in 0.01 M sodium citrate buffer, pH 6.0, andheated for2 x 5minina microwave oven at 2450 MHz/800 W. This procedure facilitates re-trieval of cytoplasmic antigens. Sections were

then incubated with mouse ED1 monoclonal antibody (MAb) (anti-CD68, which recognizes monocytes and macrophages) (15) or mouse anti-humanMIFMAb, which alsorecognizes rat

MIF (16), using a three-layer peroxidase anti-peroxidase (PAP) staining protocol, and devel-oped with 3,3-diaminobenzidine, therebygiving a brown product. For double immunostaining,

138 MolecularMedicine, Volume 3, Number 2, February1997

sections were microwave heated for a second time after completion of the three-layer PAP staining with the ED1 MAb to denature bound immunoglobulin molecules and prevent anti-bodycross-reactivity. Sections were then labeled with the anti-MIF MAb by means of a three-layer alkaline phosphatase-anti-alkaline phos-phatase (APAAP) method and developed with Fast Blue BB Base (Sigma Chemical Co., St. Louis, MO), which yielded a blue product. Sec-tions were counterstained with PAS reagent without hematoxylin and coverslipped in an aqueous mounting medium. An isotype-matched mouse antibody 73.5 against human CD45R was used as the negative control.

In Situ Hybridization

A 420bp Xba l/BamHI fragment of mouse MIF cDNAin theBluescript SK+ vector (Stratagene, La Jolla, CA) (17), which is highly homologous to the rat nucleotide sequence (18), was used to prepare sense and antisense MIF cRNA probes labeled with digoxigenin-11-UTP by in vitro transcription according to the manufacturer's protocol (Boehringer Mannheim, Mannheim, Germany). The incorporation of digoxigenin-labeled nucleotides was assayed by spotting serial dilutions of probes onto Hybond nylon mem-brane (Amersham, Little Chalfont, Buckingham-shire, U.K.) and detecting thebound probe with alkaline phosphatase-conjugated Fab fragments of anti-digoxigenin IgG with 4-nitroblue tetra-zolium chloride and X-phosphate substrates

according to the manufacturer's instructions (Boehringer Mannheim). Only probes detectable at a concentration of 0.1 pg/,ul or lower were used.

Insituhybridizationwasperformedon4-gm

paraffin sections of formalin-fixed tissue by using a newly developed microwave-based technique (19). After dewaxing, sections were treated with a microwave oven for 2 X 5 min as outlined

above, incubated with 0.2 M HCI for 15 min, followed by 1% Triton X-100 for 15 min, and then digested for20 minwith 10,ug/ml protein-ase-K at370C (Boehringer Mannheim). Sections then were washed in 2 X SSC, prehybridized, and then hybridized with 0.3 ng/,ul DIG-labeled sense or anti-sense cRNA probe overnight at 420C in a hybridization buffer containing 50% deionized formamide, 4 X SSC, 2 X Denhardt's solution, 1 mg/ml salmon sperm DNA (Sigma), and 1 mg/ml yeast tRNA. Sections were washed in 0.1 X SSC at42°C and the hybridized probe

detectedby means of alkaline

phosphatase-con-jugated sheep anti-digoxigenin IgG and color de-velopment with NBT/X-phosphate. The sections were counterstained with PAS and mounted in

anaqueousmedium. Some sectionswere subse-quently microwave oven treated and used for

three-layer PAP staining with the anti-MIF or

ED1 MAb, as described above.

Quantitation of in Situ Hybridization and Immunohistochemistry Staining

The quantitation of positively stained cells in tissue sections was performed as previously de-scribed (5). Briefly, the number of cells labeled by the MIF cRNA probe or anti-MIF MAb were

counted in at least 50 consecutive glomerular cross sections (gcs) in each animal. In double-stained sections, the number ofED1+,

MIF+,

and

ED1+MIF+ cells was counted in 50 glomerular cross sections. Macrophageaccumulationwithin the tubulointerstitium was assessedby counting ED1+ cells in 20 consecutive high-power fields (X40) by means of a 0.02 mm2 graticule fitted into the eyepiece of the microscope. Counting progressed from the outer cortex to the inner cortex,avoiding only largevessels andglomeruli. In addition, the number of labeled tubules was scored in at least 1000 consecutive cortical tubu-lar cross sections. Data from groups of six ani-mals areexpressedas the mean ± standard error of the mean (SEM) per glomerular cross section or per square millimeter (mm2).

MIF SecretionbyCultured Glomeruli Glomeruli were purified by differential sieving. Briefly, a half kidney from each animal was placed in RPMI 1640 with 5 % fetal calf serum (FCS), diced finely, gently pressed through a

250-Aim

wire mesh, and poured through a

106-,umand thena75-,umwiremesh.Glomeruli remaining on the top of a 75-ptm mesh were washedextensivelywith RPMI/5%FCS and col-lected. Isolated glomeruli were >95% pure as assessed by phase-contrast microscopy. Glomer-uliwere cultured at 3000glomeruli/mlin RPMI 1640/5% FCS for 24 hrandthe supernatantwas harvested. TheconcentrationofMIF in superna-tant samples was quantitated by a sandwich ELISA (7).

Serum MIF Levels

H. Y. Lan etal.: TNF-ae UpregulatesMIF inGlomerulonephritis 139

TABLE 1. Effect of TNFbptreatment onrenal injury inanti-GBMdisease

Serum Glomerular Glomerular Interstitial

Proteinuria Creatinine Hypercellularity Crescents Fibrosis

(mg/24hr) (,imol/L) (0-3) (%) (0-4)

Normal 9 ±2 42.7± 1.0 0 0 0

Saline treated

Day 1 65 ± 22a 47.8 ± 6.0 0.82± 0.10 0 0.8 ± 0.2

Day 7 204± 22c 59.3 ± 4.4b 1.30± 0.17 10.3 ± 2.1 2.3± 0.3

Day 14 288 ± 14c 67.2± 2.4c 1.40 ± 0.06 17.3 ± 2.6 3.2 ± 0.3

TNFbp treated

Day 1 43 ± 4a 49.0 ±4.4 0.71 ± 0.07 0.3 ± 0.3 0.5 ± 0.2

Day 7 125 ± 18cf 47.2 ± 5.1 0.99 ± 0.12 0.3 ± 0.2f 1.2 ± 0.2d

Day 14 190 ± 16cf 46.0 ± 1.8e 0.89 ± 0.07e 0.9± 0.3f 1.2 ± 0.2e

Data shownas mean + SEM. ap< 0.05, bp<0.01; cp< 0.001 versus normal;dp <0.05;ep< 0.01,fp< 0.001 comparedto time-matched saline-treated animals.

interfered with the MIF ELISA. Five-microliter serumsamples from allanimalswereresolved by SDS polyacrylamide gel electrophoresis, blotted onto nitrocellulose membrane, incubated se-quentially with rabbit anti-murine MIF poly-clonal antibody and horse radish peroxidase-conjugated goat anti-rabbit antibody, and devel-oped with chloronaphthol/H202, as previously described (4). Recombinant MIF was used as a control to compare levels of MIFexpression.

Statistical Analyses

One-wayanalysis ofvariance (ANOVA) from the

Complete Statistical Analysis program (CSS, Statsoft, Tulsa, OK, USA) was used to analyze differences in MIF expression, macrophage ac-cumulation, and renal functional in normal animals, TNFbp-treated, and saline-treated anti-GBMdisease. Differencesinhistology were com-pared by the nonparametric Mann-WhitneyUtest.

RESULTS

RenalFunction andHistopathology

The induction of anti-GBM disease in saline-treated animals resulted in severe proteinuria and a significant increase in serum creatinine levels (Table 1). Saline-treated animals showed

significant glomerularhypercellularity and cres-cent formation plus marked interstitial fibrosis onDay 14 ofanti-GBM disease. Incontrast,

TN-Fbp treatment caused a significant reduction in proteinuria andmaintained normal levels of se-rum creatinine (Table 1). Consistent with these measurements, TNFbp-treated animals showed only mildglomerular hypercellularity and inter-stitial fibrosis, whereas crescent formation was almostentirely abolished.

Glomerular MIF Expression and

Macrophage Accumulation

Insituhybridization and immunohistochemistry staining showed that a small number of glomer-ularvisceral andparietal epithelial cells

constitu-tively expressedMIFmRNA (3-10 cells/gcs) and protein (0-3cells/gcs) (Fig. la, b). The induction of anti-GBM disease in saline-treated animals causedarapidincrease inthe number of glomer-ular cells expressing MIFmRNA and proteinon Day 1, and this was further increased on Days 7 and 14 (Figs. 1 and 2). This up-regulation of

glo-merular MIF expressionwas paralleledby macro-phage accumulation over the 14-day period (Fig.

ic).

TNFbp treatment caused only a partial reduction inglomerular MIF expressionon Day 1 of disease compared with expression in saline-treated animals, and this didnotaffectglomerular macrophage accumulation (Fig. la-c). However,

TNFbptreatmentalmostcompletelypreventedany furtherincrease inMIF mRNA or protein expres-sion during the rest of the course of the disease, which wasparalleledby inhibition ofglomerular

140 Molecular Medicine, Volume 3,Number 2,February 1997

Tubulointerstitium

60,i (a) MI F mRNA

U)

0

C 45.

cn a)

0

+I 30

15.

20-i) 15

0)

U)

10.

U-

5-Ul)

L-)

CY)

CO)

ai)

0

ll

100, (d) MIF mRNA

-

75-a)

D0 50.

2 25.

(b) MIF protein

cn

U)

Ui-v T

75 (e) MIF protein

50

4,,,,,,,,,,,,,,

~~~~~~...

...

* 25_

0-3001 (f Macrophages

E

E

200-C.)

a 100.

w

0 1 7 14

Days after induction of disease

FIG. 1. TNFbptreatment

inhibits glomerular and tu-bularMIF expression and macrophage accumulation in

ratanti-GBM glomerulone-phritis

(a) Quantitation ofglomerular MIFmRNAexpressionbyin situhybidization; (b) quantita-tionofglomerular MIFprotein expression by immunohisto-chemistry; (c) quantitationof glomerularED1-positive

macro-phages; (d) quantitation of

tu-bular MIF mRNA expression by in situhybidization; (e) quanti-tation of tubularMIF protein expressionby immunohisto-chemistry; and (f) quantitation of ED 1-positive macrophages within the tubulointerstitium. Saline-treated animals are rep-resented by solid lines and closed circles, TNFbp-treated animals, bydotted lines and open circles. Dataareexpressed

asthe mean ± SEM from

groupsofsixanimals. *p < 0.05, **p < 0.01, and ***p < 0.001, compared with saline-treated animals.

SinceTNF-aisknowntoinducemacrophage

MIFproduction invitro, we used double

immu-nohistochemistry staining to examine the

num-ber of macrophages expressing MIF mRNA and protein in anti-GBM disease (Fig. 2). Quantita-tion of double staining demonstrated that the number ofglomerular ED1+MIF+ macrophages

was significantly inhibited by TNFbp treatment

(Fig. 3). While this was dueinpart toan overall

reduction in the number of glomerular

macro-phages seen with TNFbp treatment, there was

also a reduction in the percentage ofglomerular macrophages positiveforMIF (25.8 ± 4.5 versus

58.0 ± 3.4% in TNFbp- and saline-treated

ani-mals, respectively; p < 0.001).

Glomerular MIF production was also

quan-titatedby measuringMIFsecretionby purified glo-meruli during a 24-hr culture period. Substantial

levels ofMIFwere secretedby glomeruli cultured

from normal ratkidney (Fig. 4). Thiswas

signif-icantlyincreased onDay 14of anti-GBM disease

in saline-treated animals. TNFbp treatment

causeda significantreduction inglomerularMIF

secretion compared with that in saline-treated animals. In addition, TNFbptreatmentproduced

atrend towards a reduction in the basal level of

FIG. 2. TNFbptreatment inhibitsglomerular and tubular MIFexpression andmacrophage accumula-tion inratanti-GBMglomerulonephritis

Combined in situhybridization andimmunohistochemistryshow MIFmRNAexpression (blue) byintrinsickidney cells andbyED 1-positive macrophages (brown) onDay 7 of disease with (a) salineor (b) TNFbp treatment. Dou-bleimmunohistochemistry shows colocalization of strong MIFexpression (blue) andED1-positive macrophages (brown) onDay 14of disease with (c) saline or (d) TNFbp treatment. Macrophages (brown) expressingMIF mRNAorprotein areindicatedbyarrowheads. StrongMIFexpression and thedevelopmentof crescentformation and tubulitis areseen. Sectionswere counterstained with PAS minus haematoxylin. Original magnification, X400.

Glomerulus

U

V EF

---w

...a...10

nJ

142 MolecularMedicine, Volume 3,Number 2, February 1997

3000-E

m

2500-a

_'

2000-0

.C

2

1500-a) 0

c

1000-0 0

LL

500-n

7 14

Days after induction ofdisease

FIG. 3. TNFbptreatmentinhibits macrophage MIF expression in anti-GBM glomerulone-phritis

Quantitationof double imrnmunohistochemistry stain-ing shows the number ofMIF+ED1+ macrophages in saline- (solid bars) and TNFbp- (dotted bars) treated animals. Each barrepresents groupsofsix

animals (mean + SEM). **p < 0.01, ***p < 0.001.

Daysafter induction of disease

FIG. 4. TNFbptreatmentinhibits glomerular MIF sectioninratanti-GBM glomerulone-phritis

Secretion of MIFby culturedglomeruli overa 24-hr periodwas quantitated byELISA in normal rats

(Day 0) andinsaline- (solid circles) and TNFbp-(opencircles, dotted line) treated animals. Data are

expressedas themean ± SEMfromgroups of six

animals. ***p < 0.001, compared with saline-treated animals; a, p < 0.05, comparedwith normal animals (Day 0).

glomerular MIF secretion seen in normal rats,

but this did notreach statistical significance.

Tubular MIF Expression and Interstitial Macrophage Accumulation

Tubular epithelial cells are the main source of MIFinnormalkidney, withupto30%of cortical

tubules being positive for MIF mRNA and

pro-tein expression. As in glomeruli, there was a

rapid up-regulation in tubular MIF mRNA and proteinonDay 1 ofsaline-treated anti-GBM

dis-ease, which was further increased over the courseof the disease (Figs. 1 and 2a, c). A

signif-icant interstitial macrophage infiltration

oc-curred in parallel with the increase in MIF

ex-pression (Fig. If). Tubular MIF expression was

prominent within focal tubulointerstitial lesions and the development oftubulitis. Double-label-ing studies showed that most interstitial ED1+ macrophages were present inand around tubu-lointerstitial lesions withstrongtubular MIF

ex-pression (Fig. 2a, c).

Consistent with the effectsseeninglomeruli,

TNFbp treatmentcausedonlya partialreduction

intubular MIF expression on Day 1, but it

pre-ventedanyfurther increase in tubular MIF mRNA andprotein expression overDays 7and 14 of the disease (Fig. Id, e). This was associated with a

marked suppression of both interstitial

macro-phageinfiltration and a complete inhibition of

tu-bulointerstitial damage (Table 1, Figs. 1 and 2).

Serum MIF Levels in Anti-GBM Disease

To compare the effects of TNFbp treatment on

systemic MIF production during anti-GBM

dis-ease,weexaminedserumMIFlevelsbyWestern blotting. MIF was detected in normalrat serum

and MIFlevelswere substantiallyincreased

dur-ingthecourseof saline-treated anti-GBM disease

(Fig. 5).Incontrast,TNFbptreatmentresultedin

undetectable levels ofserum MIF onDay 1 and

markedly reducedserumMIFlevelsoverDays 3 to 14, compared with saline-treated animals.

DISCUSSION

This studyisthe firsttodemonstrate thatTNF-a

up-regulates both local and systemic MIF

pro-duction in an immunologicallyinduced disease. In previous studies, we described an excellent

correlation between up-regulation of renal MIF

expressionand theprogression ofrenalinjuryin rat crescentic glomerulonephritis (5).

Further-more, it has recently been reported that treat-ment of this disease model with a neutralizing

anti-MIF antibody substantially blocks

macro-phage accumulation and progressive renal

in-jury, demonstratingapathogenicrole for MIF in

macrophage-mediated glomerulonephritis (20). Blockade of TNF-a action by administration of

8-C,)

co 0

-..

6-C,,

0

LL

4-0 2-w

0-i -T1

1

a

IT

o~~~~~~~~~

'1%-6

14

-H. Y. Lan etal.: TNF-aUpregulatesMIF in Glomerulonephritis 143

DAYI DAY3 DAY7 DAY14

rmMIF Normal S T S T S T S T

FIG. 5. TNFbp treatmentreduces serum MIF levels during the anti-GBMglomerulonephritis

Through Western blot analysis, the presenceof se-rum MIF protein was detected innormal rats, which isthe same size as the recombinant protein. TNFbp treatmentcaused a marked reduction in the level of serum MIFprotein overthe entire disease time course, compared with that insaline-treated animals.

the TNFbp causedaprofound suppression of re-nal MIF production, which was associated with inhibition of macrophage infiltration, reduction inrenal injury, prevention of renalfunction im-pairment,and histological damage.Thisconfirms the importance of TNF-a inthe development of anti-GBM glomerulonephritis (10,11). Further-more, itsuggests thatup-regulation of MIF pro-duction is a pathological manifestation of pro-gressive renal functional and histologic injury, including glomerular crescent formation. How-ever, TNF-a may also act to promote macro-phage accumulation throughother mechanisms. For example, TNF-a can induce expression of interleukin- 1, monocyte chemoattractant pro-tein-1,and intercellular adhesion molecule-1,all of which have been shown tocontribute to leu-kocytic infiltration and subsequent renal injury in ratanti-GBM disease (13,21,22).

Theability ofTNFbptreatmenttoinhibit mac-rophageMIFmRNAand protein expressionduring crescentic glomerulonephritis confirms the rele-vanceofin vitrostudiesinwhichTNF-a wasfound toinduce macrophageMIFproduction (4). In

ad-dition, the ability ofTNFbp to inhibit MIIF mRNA andprotein expressionbyintrinsicrenalcells, such asglomerular and tubular epithelialcells, demon-stratesthatTNF-a canalso induce MIFexpression in cell types other than leukocytes. Furthermore,

TNFbp treatment reduced serum MIF levels.

Al-though the cellular source of circulating MIF has yet to be identified, this study has shown that TNF-a up-regulates the systemic MIF response in inflammatory disease. It should be pointed out that whileTNFbp treatment suppressedboth lo-cal and systemic MIF expression, levels of MIF

expressionwere notreduced below thoseseenin

normalanimals, demonstrating that TNF-a does not regulate constitutive MIF expression.

These data support the existence of an am-plification loop between TNF-a and MIF that exerts a potent proinflammatory effect in vivo. Thispathway may operate through the transcrip-tion factor NF-KB. Expression of theTNF-agene is strongly induced byNF-KB, andTNF-a stimu-lationof target cells resultsinactivationofNF-KB (23,24). Furthermore, an NF-KB site has been identified within the promoter of themouse MIF gene ( 17). Thus,it isfeasiblethatbothTNF-aand MIF act to mutually up-regulate each other through the activation of NF-KB. This may also be the mechanism of the counter-regulatory re-lationship between MIF and glucocorticoids, since glucocorticoids have recently been shown to exert theiranti-inflammatory action by inhibiting activa-tion of NF-KB (25). Nevertheless, the presence of endocrine-responsive elements within the MIF genepromoter also indicates that the regulation of MIF transcription maybe more complex (17).

In summary, this study has demonstrated that TNF-a-induced up-regulation of local and systemic MIF expression may be one ofthe im-portant mechanisms mediating renal injury in

experimentalcrescenticglomerulonephritis. Fur-thermore, thisprovidesthe first in vivoevidence thatTNF-aandMIFacttogetherina

proinflam-matoryamplification loopinthepathogenesis of immune/inflammatory disease.

ACKNOWLEDGMENTS

This study wassupported byNH &MRC (950912 and 960106) and NIH (AI-21359) grants.

REFERENCES

1. David JR. (1966) Delayed hypersensitivityin vitro: Its mediation by cell-free substances formed bylymphoid cell-antigen interaction.

Proc. Natl. Acad. Sci. U.S.A. 56: 72-77.

2. BloomBR, Bennett B. (1966) Mechanism of a reaction in vitro associated with delayed-type hypersensitivity. Science 153: 80-82. 3. NishnoT, BernhagenJ,ShijkiH,Calandra T,

Dohl K, Bucala R. (1995) Localization of macrophage migration inhibitory factor (MIF) to secretory granules within the cor-ticotrophic and thyrotrophic cells of the pi-tuitarygland. Mol. Med. 1: 781-788.

Bu-144 Molecular Medicine, Volume3, Number2,February 1997

cala R. (1994) The macrophage is an impor-tant and previously unrecognized source of macrophage migration inhibitory factor. J. Exp. Med. 179: 1895-1902.

5. Lan HY, Mu W, Yang N, et al. (1996) De novorenalexpressionofmacrophage migra-tion inhibitory factor (MIF) during the de-velopment of rat crescentic glomerulone-phritis. Am. J. Pathol. 149: 1119-1127. 6. BernhagenJ, CalandraT, Mitchell RA, et al.

(1993) MIF is a pituitary-derived cytokine thatpotentiateslethal endotoxaemia. Nature

365: 756-759.

7. Calandra T, Bernhagen J, Metz CN, et al. (1995) MIF as a glucocorticoid-induced modulator of cytokine production. Nature

376: 68-71.

8. BacherM, Metz CN, CalandraT, etal. (1996) An essential regulatory role for macrophage migrationinhibitory factorinT-cellactivation. Proc. Natl.Acad. Sci. U.S.A. 93: 7849-7854. 9. Pozzi LA, Weiser WY. (1992)Human

recom-binant migration inhibitory factor activates human macrophages to kill tumor cells. Cell Immunol. 145: 372-379.

10. Hruby ZW, Shirota K, Jothy S, Lowry RP. (1991) Antiserum against tumor necrosis fac-tor-alpha anda protease inhibitor reduce im-muneglomerularinjury.KidneyInt.40:43-51.

11. Baud L, Fouqueray B, Philipp C. (1994)

In-volvement oftumornecrosis factor-alphain glomerular injury. Springer Semin. Immuno-pathol. 16: 53-61.

12. Lan HY, Paterson DJ,Atkins RC. (1991) Ini-tiation and evolution of interstitial leuko-cyticinfiltration inexperimental glomerulo-nephritis. KidneyInt. 40: 425-433.

13. Lan HY, Nikolic-Paterson DJ, Mu W, Van-nice JL, Atkins RC. (1995)

Interleukin-1

re-ceptor antagonist halts the progression of established crescentic glomerulonephritis in

the rat. Kidney Int. 47: 1303-1309.

14. Lan HY, Mu W, Nikolic-Paterson DJ, Atkins

RC. (1995) Anovel, simple, reliable, and sen-sitive method for multiple immunoenzyme staining: Use of microwave oven heating to blockantibodycrossreactivityandretrieve an-tigens. J. Histochem. Cytochem. 43: 97-102.

15. Dijkstra CD, Dopp EA, Joling P, Kraal G. (1985) The heterogeneity of mononuclear phagocytes in lymphoid organs: Distinct macrophage subpopulationsintherat recog-nized by monoclonal antibodies ED1, ED2

and ED3. Immunology 54: 589-599.

16. Bacher M, Meinhardt A, Lan HY, et al. (1997) MIF expression in experimentally-inducedendotoxemia.Am. J.Pathol. 150: 235-246.

17. Mitchell R, Bacher M, Bernhagen J, Push-karskaya T, Seldin MF, Bucala R. (1995) Cloning and characterization of thegenefor mousemacrophagemigrationinhibitory fac-tor (MIF). J. Immunol. 154: 3863-3870. 18. Sakai M, Nishihira J, Hibiya Y, Koyama Y,

Nishi S. (1994) Glutathione bindingrat liver 13kprotein is the homologue of the

macro-phage migration inhibitory factor. Biochem. Mol. Biol. Int. 33: 439-446.

19. Lan HY, Mu W, Ng Y-Y,Nikolic-PatersonDJ, Atkins RC. (1996) A simple, reliable, and sensitive method of nonradioactive in situ hybridization: Use of microwave heating to

improve hybridization efficiency and pre-serve tissue morphology. J. Histochem. Cyto-chem. 44: 281-287.

20. Lan HY, Bacher M, Mu W, et al. (1996) Blockade of macrophage migration inhibi-toryfactor (MIF) inhibits rat crescentic glo-merulonephritis (GN). (abstract) J. Am. Soc. Nephrol. 7: 1707.

21. Nishikawa K, Guo YJ, Miyasaka M, et al. (1993) Antibodies to intercellular adhesion molecule

1/lymphocyte

function-associated antigen 1 prevent crescent formation in rat autoimmune glomerulonephritis. J. Exp. Med. 177: 667-677.

22. Tang WW, Qi M, Warren JS. (1996) Mono-cytechemoattractant protein 1 mediates glo-merular macrophage infiltration in anti-GBM Ab GN. Kidney Int. 50: 665-671. 23. Beg AA, Finco TS, Nantermet PV, Baldwin

AS Jr. (1993) Tumor necrosis factor and in-terleukin-1 leadto phosphorylation and loss of I kappa B alpha: a mechanism for NF-kappa B activation. Mol. Cell Biol. 13:

3301-3310.

24. TredeNS,TsytsykovaAV, ChatilaT, Goldfeld AE, Geha RS. (1995) Transcriptional activa-tion of the human TNF-alpha promoter by superantigen in human monocytic cells: Role ofNF-kappaB. J.Immunol. 155: 902-908. 25. Auphan N, DiDonato JA, Rosette C, Helm-berg A, Karin M. (1995) Immunosuppres-sion by glucocorticoids: Inhibition of NF-kappa B activity through induction of I

kappa B synthesis. Science 270: 286-290.