Induction of haem oxygenase causes cortical non haem iron

Document Sample
Induction of haem oxygenase causes cortical non haem iron Powered By Docstoc
					Journal of Neurochemistry, 2007, 100, 532–544                                                           doi:10.1111/j.1471-4159.2006.04230.x

Induction of haem oxygenase-1 causes cortical non-haem iron
increase in experimental pneumococcal meningitis: evidence that
concomitant ferritin up-regulation prevents iron-induced oxidative

Hao Ren,* Stephen L. Leib,* Donna M. Ferriero,  Martin G. Tauber* and Stephan Christen*
*Institute for Infectious Diseases, University of Berne, Berne, Switzerland
 Departments of Neurology and Pediatrics, University of California, San Francisco, California, USA

Abstract                                                               the accumulation of bilirubin detected in HO-1-positive cells)
Desferrioxamine inhibits cortical necrosis in neonatal rats with       completely prevented the infection-associated non-haem iron
experimental pneumococcal meningitis, suggesting that iron-            increase. The same cells also displayed markedly increased
induced oxidative damage might be responsible for neuronal             ferritin staining, the increase of which occurred independently
damage. We therefore examined the spatial and temporal                 of HO activity. At the same time, no increase in DNA/RNA
profile of changes in cortical iron and iron homeostatic pro-           oxidation was observed in infected animals (as assessed by in
teins during pneumococcal meningitis. Infection was associ-            situ detection of 8-hydroxy[deoxy]guanosine), strongly sug-
ated with a steady and global increase of non-haem iron in the         gesting that ferritin up-regulation protected the brain from iron-
cortex, particularly in neuronal cell bodies of layer II and V,        induced oxidative damage. Thus, although pneumococcal
and in capillary endothelial cells. The non-haem iron increase         meningitis leads to an increase of cortical non-haem iron,
was associated with induction of haem oxygenase (HO)-1 in              protective mechanisms up-regulated in parallel prevent iron-
neurones, microglia and capillary endothelial cells, whereas           induced oxidative damage. Cortical damage does not appear
HO-2 levels remained unchanged, suggesting that the non-               to be a direct consequence of increased iron, therefore.
haem iron increase might be the result of HO-1-mediated                Keywords: bacterial meningitis, bilirubin, brain injury,
haem degradation. Indeed, treatment with the haem oxyge-               hydroxyl radical, inflammation, iron.
nase inhibitor tin protoporphyrin (which completely blocked            J. Neurochem. (2007) 100, 532–544.

We have previously shown that cortical neuronal injury in                 Iron is taken up by the brain by receptor-mediated endo-
neonatal experimental pneumococcal meningitis is inhibited             cytosis of circulatory transferrin-bound iron by brain endo-
by the iron chelator desferrioxamine (DFO) (Auer et al.                thelial cells and choroid epithelial cells, and is then
2000), suggesting that cortical injury may be the result of            distributed to the different brain cells by mechanisms that
iron-induced oxidative damage. Although iron is the most
abundant metal in the brain and essential for brain function
(e.g. energy production, myelin formation), it can potentially         Received June 28, 2006; revised manuscript received September 4,
promote the formation of highly reactive hydroxyl radicals             2006; accepted September 5, 2006.
                                                                          Address correspondence and reprint requests to Dr Stephan Christen,
via Fenton chemistry (Halliwell 1992). Increased formation             Institute for Infectious Diseases, Friedbuehlstrasse 51, CH-3010 Berne,
of hydroxyl radicals causes oxidative damage to lipids,                Switzerland. E-mail:
proteins and DNA, which can ultimately lead to the death of               Abbreviations used: DAB, diaminobenzidine; DFO, desferrioxamine;
affected cells. Formation of 8-hydroxydeoxyguanosine                   GFAP, glial fibrillary acid protein; H-ferritin, heavy-chain ferritin sub-
(8OHdG) lesions in DNA by hydroxyl radicals, for example,              unit; HIF, hypoxia-inducible factor; HO, haem oxygenase; IB4, isolectin
                                                                       B4; L-ferritin, light-chain ferritin subunit; NPBI, non-protein-bound
is known to lead to activation of poly(ADP-ribose) poly-               iron; PBS, phosphate-buffered saline; 8OH(d)G, 8-hydroxy(de-
merase and p53, thereby promoting programmed cell death                oxy)guanosine; SnPP, tin protoporphyrin IX; TUNEL, terminal deoxy-
in ischaemic neurones (Yu et al. 2003).                                nucleotidyl transferase-mediated dUTP-biotin in situ nick end labelling.

                                     Ó 2006 The Authors
532                                  Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
                                                                               Haem oxygenase-1 induction in pneumococcal meningitis      533

are not completely understood, but may in part involve the              proteinase K from Sigma (St Louis, MO, USA). The T-PERÔ tissue
diffusion of extracellular non-transferrin-bound iron (Moos             protein extraction reagent, enhanced chemiluminescent reagent and
and Morgan 1998). Within brain cells most of the iron is                metal-enhanced DAB kit were from Pierce (Rockford, IL, USA).
either incorporated into haem proteins (e.g. cytochromes of             Analytical-grade water was prepared using a Milli-Q water purifica-
                                                                        tion system (Millipore, Billerica, MA, USA).
the electron transport chain) or sequestered by the intra-
cellular iron-storage protein ferritin. Ferritin is composed of a
                                                                        Infection, clinical evaluation and tissue processing
total of 24 heavy (H-) or light (L-) chain subunits in different        All animal experiments were approved by the Animal Care and
proportions depending on the cell type, and stores up to 4500           Experimentation Committee of the Canton of Berne, Switzerland,
molecules of iron (Connor et al. 2001). Sequestration of iron           and carried out according to the Guide for the Care and Use of
by ferritin or other iron-binding proteins prevents it from             Laboratory Animals (ILARCLS, 1996). Eleven-day-old nursing
engaging. However, iron can be released from ferritin, for              Wistar rats were infected intracisternally with 10 lL  5 · 106 cfu/
example, by certain conditions such as a decrease in pH or              mL Streptococcus pneumoniae (serotype 3) and returned to their
increase in superoxide anion, leading to a transient increase           dam as described previously (Pfister et al. 2000). Animals that were
in the non-protein-bound iron (NPBI) pool. An increase in               mock-infected with an equal amount of sterile saline served as
this pool leads to homeostatic regulation of the proteins               uninfected controls.
                                                                           At 18 h after infection, pups were assessed clinically using the
involved in iron transport, uptake and intracellular storage
                                                                        following scale of scores: 1, coma; 2, does not turn upright; 3, turns
(such as ferritin) so as to prevent iron-induced oxidative
                                                                        upright within 30 s; 4, minimal ambulatory activity, turns upright in
damage (Torti and Torti 2002).                                          less than 5 s; 5, normal. A small volume of CSF ( 30 lL) was
   Free iron can also be generated from haem. Haem oxygenase            drawn from the cisterna magna, and 5 lL quantitatively cultured on
(HO; EC catalyses the degradation of haem to Fe2+,           blood agar plates. The remaining material was centrifuged and
carbon monoxide and biliverdin (Maines 1997). Biliverdin is             stored at ) 80°C for biochemical analysis. Out of a total of 63
subsequently reduced to bilirubin by biliverdin reductase.              infected animals, three had seizures and were excluded from the
Under basal conditions, the major HO activity in the brain is           study. Animals who died spontaneously were also excluded from the
afforded by the constitutively expressed HO-2 isoform.                  study. Treatment with antibiotics was omitted to exclude potential
Expression of a splice variant of HO-2, named HO-3, has                 effects on iron homeostasis.
been described in the brain, but the existence and functional              At a mean ± SD of 22.6 ± 0.9 h after infection, a time point at
                                                                        which cortical neuronal injury is well developed in this model (Leib
relevance of HO-3 has recently been questioned (Ryter et al.
                                                                        et al. 2001), animals were deeply anaesthetized with pentobarbital
2006). HO-2 levels are largely uninducible, whereas stress
                                                                        (100 mg/kg intraperitoneally), thoracotomized, and blood collected
conditions such as hyperthermia (Ewing and Maines 1991),                from the left cardiac ventricle using heparin-coated syringes. For
trauma (Fukuda et al. 1996) or cerebral hypoxia–ischaemia               quantitative biochemical analyses, animals were perfused trans-
(Bergeron et al. 1997) cause the up-regulation of HO-1 (also            cardially with ice-cold phosphate-buffered saline (PBS) at a rate of
known as heat-shock protein 32) in the brain.                           20 mL/min using a motor-driven perfusion pump. After removal of
   Whether HO-1 is induced in pneumococcal meningitis has               the brain, cortices were dissected out, immediately frozen on dry ice,
not yet been investigated. We therefore examined the spatial            and stored at ) 80°C until analysis. In most experiments, one
and temporal profile of changes in iron and iron homeostatic             hemisphere was used for iron analysis and the other for western blot
proteins (HO-1, HO-2, H- and L-ferritin) and assessed iron-             analysis. For iron histochemistry and immunohistochemical analy-
induced oxidative damage by immunohistochemical analysis                ses, animals were first flushed with PBS (to remove red blood cells)
                                                                        and then perfused with 4% paraformaldehyde. Brains were removed,
of 8OH(d)G in the cortex of animals with pneumococcal
                                                                        postfixed overnight, and embedded in paraffin. Coronal sections
meningitis. We provide evidence that pneumococcal menin-
                                                                        8 lm thick were cut on a microtome and processed as described
gitis is associated with an HO-1-dependent increase in non-             previously (Schaper et al. 2002).
haem iron in the cortex, but that this increase does not cause
oxidative damage and is thus unlikely to be the cause of                Drug treatment
neuronal damage in the cortex. Alternative mechanisms by                SnPP was dissolved in dimethylsulphoxide/water (1 : 1), kept
which DFO could provide neuroprotection are discussed.                  protected from light, and injected subcutaneously at 60 mg/kg
                                                                        (Kim and Rivier 2000) every 8 h starting at the time of infection.
                                                                        Control animals received a corresponding volume of vehicle only.
Materials and methods
                                                                        Iron histochemistry
Materials                                                               Deparaffinized sections were stained for non-haem iron by the
Hydrochloric acid, hydrogen peroxide and Triton X-100 were from         DAB-intensified Perls’ reaction essentially as described by Nguyen-
Merck (Darmstadt, Germany), ferrous ammonium sulphate                   Legros et al. (1980). Briefly, sections were incubated for 30 min in a
from Fluka (Buchs, Switzerland), tin protoporphyrin IX (SnPP) from      solution consisting of one part 10% (w/v) potassium ferrocyanide
Frontier Scientific (Logan, UT, USA), and bathophenanthroline            and one part 10% (v/v) hydrochloric acid, followed by rinsing with
disulphonate, Chelex 100, diaminobenzidine (DAB), dithiothreitol,       analytical-grade water (Wang et al. 2002). Sections were then
FITC-conjugated isolectin B4 (IB4), potassium ferrocyanide and

Ó 2006 The Authors
Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
534 H. Ren et al.

immersed for 15 min in 0.05% (w/v) DAB dissolved in 0.5% (v/v)          method as described previously (Schaper et al. 2002), or appropriate
hydrogen peroxide. DAB staining was viewed on a Zeiss Axiophot          fluorescent secondary antibodies: Cy3-labelled goat anti-rabbit
microscope (Gottingen, Germany) and captured into OpenLab               (Milan Analytica, La Roche, Switzerland), Alexa 488-labelled goat
(Improvision, Coventry, UK) using a model C4742-95-12SC digital         anti-rabbit (Molecular Probes, Eugene, OR, USA) or Alexa 555-
camera (Hamamatsu, Hamamatsu City, Japan). Images were                  labelled goat anti-mouse. Sections used for DAB staining were
acquired under identical microscope and camera settings. To test        pretreated with 3% (v/v) hydrogen peroxide to inactivate endog-
for specificity of the staining for iron, control sections were          enous peroxidase activity. To control for non-specific immunoreac-
incubated with potassium ferrocyanide dissolved in water instead of     tivity, control sections were incubated in the absence of primary
hydrochloric acid.                                                      antibody. No staining was observed under these conditions. Bright-
                                                                        field and epifluorescence images were captured as described above.
Quantitative non-haem iron measurements
Non-haem iron levels were determined using the Ferene-S method          IB4 histochemistry
(Higgins 1981) using a commercially available kit (DiaSys,              Microglia and blood vessels were visualized by IB4 histochemistry
Holzhelm, Germany). Frozen cortices were homogenized 1 : 2.5            as described previously (Pennell et al. 1994). Brain sections were
(w/v) in ice-cold Tris-buffered saline (100 mM, pH 7.4) previously      first preincubated with 0.4% (v/v) Triton X-100 in PBS for 48 h at
passed over a Chelex column to remove any adventitious iron,            room temperature and then incubated for 2 h with 10 lg/mL FITC-
centrifuged at 14 000 g for 15 min at 4°C, and the supernatant          labelled IB4 dissolved in PBS containing 0.1 mM each of CaCl2,
analysed for non-haem iron in accordance with the manufacturer’s        MgCl2 and MnCl2. Sections were washed three times with PBS and
instructions. Briefly, 1 mL precipitating agent (0.8 M acetate buffer,   mounted in Prolong antifade medium (Molecular Probes) before
pH 4.5, 0.09 M thiourea) was added to 0.1 mL homogenate super-          visualization by epifluorescence microscopy.
natant, and the absorbance at 595 nm measured after incubation for
20 min at room temperature (22.5°C). Some 0.25 mL Ferene-S              Western blot analysis
solution (45 mM ascorbic acid, 0.6 mM Ferene-S, 20 mM thiourea)         Frozen cortices were weighed and homogenized 1 : 9 (w/v) in ice-cold
was then added to the mixture, incubated for 10 min and the             T-PERÔ supplemented with protease inhibitor cocktail tablets (Roche
absorbance at 595 nm measured again. The amount of non-haem iron        Diagnostics, Basle, Switzerland), centrifuged for 5 min at 10 000 g,
was determined after subtracting the first from the second reading,      and supernatants stored at ) 80°C. Protein concentration of cleared
and comparison to a standard curve prepared with ferrous ammonium       lysates was determined by a microbicinchoninic acid assay (Pierce).
sulphate. The detection limit of the assay was 0.02 lg/mL.              Samples containing 20 lg protein were separated on sodium dodecyl
Measurements in plasma and CSF were performed using 96-well             sulphate–polyacrylamide gels (13%) after boiling for 5 min in
plates instead of cuvettes, using the same proportion of sample to      2 · sample buffer containing 200 mM dithiothreitol. Separated
reagents. Haemolysed samples were excluded from the analysis.           proteins were transferred to polyvinylidene difluoride membranes,
                                                                        blocked for 1 h with 5% non-fat dry milk, washed in Tris-buffered
NPBI measurements in CSF                                                saline containing 0.2% Tween-20, and incubated overnight at 4°C
NPBI in CSF was measured by the bathophenanthroline method              with primary antibodies against HO-1, H-ferritin or L-ferritin (all used
(Nilsson et al. 2002). Briefly, CSF was centrifuged at 10 000 g for      at 1 : 500), followed by incubation with secondary horseradish
5 min, and 25 lL supernatant combined with 1 lL 250 mM EGTA             peroxidase-conjugated antibody against rabbit IgG (1 : 10 000;
in a half-well microtitre plate (Corning Life Sciences, Acton, MA,      Sigma). Immunoreactive bands were visualized by enhanced chemi-
USA). After determination of the plate blank at 535 nm, 1 lL            luminescence. Membranes were stripped and reprobed with mouse
50 mM bathophenantroline disulphonate was added, and the plate          monoclonal b-actin antibody (1 : 5000; Sigma) as a loading control.
incubated for 15 min at room temperature. The plate was then read       Band intensities of scanned blots were evaluated densitometrically
at 535 nm for Fe2+ content. To measure total iron, 1 lL 100 mM          using Scion Image (Scion Corporation, Frederick, MD, USA).
ascorbate was added and the plate read again after incubation for
15 min at room temperature. Blank values were subtracted and            Immunofluorescence analysis of DNA/RNA oxidation
concentrations determined by comparison to a standard curve             To analyse DNA/RNA oxidation in situ, sections were incubated
prepared with ferrous ammonium sulphate.                                with monoclonal antibody 15A3 (QED Bioscience, San Diego, CA,
                                                                        USA), which recognizes 8-hydroxyguanine residues in both DNA
Immunohistochemistry                                                    (8OHdG) and RNA (8OHG) (Park et al. 1992), the major nucleic
Deparaffinized sections were blocked with 5% (w/v) bovine serum          acid oxidation product formed by redox-active iron (Henle et al.
albumin and incubated for 1 h at 37°C or overnight at 4°C with one      1996; Honda et al. 2005). Rehydrated brain sections were first
of the following antibodies: rabbit affinity-purified anti-HO-1 or        treated for 30 min with 10 lg/mL proteinase K at 37°C, blocked for
anti-HO-2 (both from Stressgen, Victoria, BC, Canada) at 1 : 100;       30 min with 5% (w/v) bovine serum albumin, incubated overnight
H- and L-subunit-specific rabbit anti-ferritin (generously provided      at 4°C with 15A3 at 1 : 50, followed by Alexa 488-labelled goat
by Drs James R. Connor and John L. Beard, Pennsylvania State            anti-mouse (Molecular Probe) at 1 : 500. To distinguish DNA from
University, Hershey, PA, USA) at 1 : 100; mouse monoclonal anti-        RNA oxidation, consecutive sections were incubated for 1 h at 37°C
bilirubin (clone 24G7; Dojindo, Kumamoto, Japan) at 1 : 50; or          with either Rnase-free Dnase I, Dnase-free Rnase (both from Roche
mouse monoclonal anti-glial fibrillary acid protein (GFAP) (Chem-        Diagnostics) or both, as described previously (Nunomura et al.
icon, Temecula, CA, USA) at 1 : 200. Immunoreactivity was               1999). To minimize artifactual oxidation, preparation of brain
visualized with metal-enhanced DAB using the streptavidin–biotin        sections and staining procedures were all performed under argon

                                      Ó 2006 The Authors
                                      Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
                                                                                    Haem oxygenase-1 induction in pneumococcal meningitis   535

gas. As a positive control, sections were preincubated for 30 min             microglia/oligodendrocytes in the corpus callosum were
with hydrogen peroxide (3%) and iron (0.1 mM ferric citrate and 0.1           readily detected (closed arrowheads in Fig. 1a, panel i), the
ferrous chloride) in methanol before analysis of DNA/RNA                      cortex of uninfected control animals was mostly negative for
oxidation, as described previously (Nunomura et al. 1999).                    non-haem iron, consistent with previous findings (Connor
                                                                              et al. 2001). In contrast, infected animals (panel ii) exhibited
Statistical analysis
                                                                              a marked increase in Perls’ staining, in particular in somata of
Data were analysed using GraphPad Prism (GraphPad, San Diego,
CA, USA). Comparisons between two groups were made by either                  layer II and V neurones (cf. closed arrows in panel iv), and in
the Student’s t-test or Mann–Whitney test (for categorized data such
as clinical scores). Comparisons between more than two groups
were carried out by one-way ANOVA followed by Tukey’s multiple
comparison test. In all cases, a two-tailed p-value < 0.05 was
considered statistically significant. Graphical data are presented as
mean ± SEM, whereas data in text are reported as mean ± SD.


Clinical parameters of pneumococcal meningitis
All animals infected with S. pneumoniae developed menin-
gitis, as evidenced by lethargy and obtundation, positive CSF
bacterial titres at 18 h after infection (7.3 ± 0.6 log10 cfu/
mL), significant weight loss () 0.2 ± 0.9 vs. 2.5 ± 1.2 g of
weight gain in uninfected controls; p < 0.001) and a
diminished clinical score (3.7 ± 0.6 vs. 5 ± 0 in uninfected
controls; p < 0.001). Fewer than 5% of infected animals died

Effect of pneumococcal meningitis on cortical non-haem
To determine whether pneumococcal meningitis is associated
with an increase in iron in the cortex, brain sections were
stained for non-haem iron using the DAB-enhanced Perls’
reaction (Nguyen-Legros et al. 1980). While Perls’-positive

Fig. 1 Effect of pneumococcal meningitis on cortical non-haem iron.
(a) Representative photomicrographs of non-haem iron deposition
(DAB-intensified Perls’ staining) in cortical cross-sections from an
uninfected control animal (i) and an animal infected for 22 h with S.
pneumoniae (ii) (original magnification · 20 objective, scale bar
50 lm). (i) Closed arrow heads denote Perls’-positive oligodendro-
cytes/microglia in the corpus callosum. Roman numerals indicate
cortical layers. (ii) Note the marked increase in iron cortical staining in
the infected animal. Open arrow denotes the Perls’-positive endot-
helial cell layer of a parenchymal blood vessel. Apart from spurious
staining (dark black spots) by contaminating red blood cells, staining
was specific for non-haem iron, as no staining was observed when
hydrochloric acid was omitted from the staining procedure. (iii, iv)
Higher-magnification images of framed regions in (i) and (ii) respect-
ively (original magnification · 40 objective, scale bar 100 lm). Closed
arrows in (iv) denote Perls’-positive neurones in cortex from an
infected animal. (b) Time course of cortical non-haem iron production.
Non-haem iron levels were determined in cortical homogenates pre-
pared at the indicated times after infection by using the Ferene-S
assay. Values are mean ± SEM of 5–7 animals per time point.
*p < 0.05, ***p < 0.001 versus t ¼ 0 using one-way ANOVA followed
by Tukey’s multiple comparison test.

Ó 2006 The Authors
Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
536 H. Ren et al.

endothelial cells of parenchymal blood vessels (open arrow
in panel ii). Increased staining was observed throughout the
cortex and was not limited to injured areas, suggesting that
the non-haem iron increase is a global response to infection.
Other areas in the brain (dentate gyrus of the hippocampus,
hypothalamus and striatum) also exhibited increased Perls’
staining, but to a lesser degree than in the cortex. Iron
staining was also observed in leucocytes (primarily granulo-
cytes) infiltrating the subarachnoid space and ventricles.
   To corroborate the histochemical data and gain insight
into the mechanism by which non-haem iron in the cortex
might increase, non-haem iron levels were determined
quantitatively at several time points after infection
(Fig. 1b). Non-haem iron levels steadily increased after
infection, being significantly increased at 16 and 22 h
compared with values in uninfected controls. [Non-haem
iron levels 22 h after mock infection did not differ from
those in uninfected postnatal day 11 rat pups and were
therefore combined (t ¼ 0).] In contrast, plasma and CSF
non-haem iron levels at 22 h after infection were no                 Fig. 2 Effect of SnPP treatment on cortical non-haem iron levels.
different from those in mock-infected controls [0.87 ±               Uninfected control or infected animals were treated with either SnPP
0.30 lg/mL (n ¼ 11) vs. 0.85 ± 0.21 lg/mL (n ¼ 6), and               (60 mg/kg subcutaneously, every 8 h starting with infection) or vehi-
                                                                     cle, and cortical non-haem iron measured at 22 h after infection.
0.29 ± 0.12 lg/mL (n ¼ 12) vs. 0.28 ± 0.10 lg/mL (n ¼
                                                                     Values are mean ± SEM of 6–14 animals per treatment group.
5), respectively]. Furthermore, the iron was not derived
                                                                     **p < 0.01, ***p < 0.001; ns, non-significant using one-way ANOVA
from the bacteria injected, as the non-haem iron content of
                                                                     followed by Tukey’s multiple comparison test.
the inoculum was below the detection limit of the Ferene-S
assay. These results suggest that the infection-associated
cortical increase in non-haem iron originates from within
the brain parenchyma.                                                animals, whereas HO-1 was markedly induced in neuronal
                                                                     somata (especially in layer II and V), blood vessels and glial
Effect of SnPP treatment on non-haem iron levels                     cells of infected animals (Fig. 3a). In contrast, HO-2
To determine whether HO contributed to the non-haem iron             immunoreactivity was readily detected in neurones and
increase, we evaluated the effect of SnPP treatment on               capillary endothelial cells under basal conditions, but did
cortical non-haem iron levels. SnPP treatment completely             not increase with infection. Fluorescence double-labelling
inhibited the infection-associated increase in non-haem iron         experiments with IB4 and anti-GFAP established that the
at 22 h after infection, but had no effect on basal levels in        HO-1-positive glial cells were microglia (closed arrowheads
uninfected animals (Fig. 2). SnPP neither had an effect              in Fig. 3b, panel i).
(p > 0.05) on plasma nor CSF non-haem iron levels                       Western blotting experiments corroborated the immuno-
(0.93 ± 0.23 vs. 0.87 ± 0.30, and 0.32 ± 0.10 vs. 0.29 ±             histochemical data and showed a time-dependent increase in
0.12 lg/mL in untreated animals), nor on spontaneous deaths          HO-1 protein during infection (Fig. 3c), reaching a signifi-
(%p > 0.87), weight gain () 2.8 ± 0.3 vs. ) 3.0 ± 0.5 g),            cant 5–6-fold increase at 22 h after infection. In contrast,
clinical score (4 ± 0 vs. 4 ± 0) or CSF bacterial titres             HO-2 levels did not change during infection. These data
(7.4 ± 0.4 vs. 7.3 ± 0.4 log10 cfu/mL) in infected animals,          suggest that induction of HO-1 is the main mediator of
strongly suggesting that HO-catalysed haem degradation is            increased cortical non-haem iron in pneumococcal menin-
responsible for the non-haem iron increase during pneumo-            gitis.
coccal meningitis.
                                                                     Determination of HO activity by in situ bilirubin
Effect of pneumococcal meningitis on HO-1 and HO-2                   detection
protein expression                                                   To gain further evidence to support this hypothesis, HO
Because SnPP can inhibit both the inducible (HO-1) and               activity was assessed in conjunction with HO-1 staining by
constitutive (HO-2) isoforms of HO, we determined the                in situ detection of bilirubin (Huang et al. 2005) (Fig. 4).
cellular distribution pattern and temporal expression profile of      Bilirubin staining (green channel) was virtually absent under
the two isoforms during infection. There was only weak               basal conditions (Fig. 4a), but marked accumulation of
expression of HO-1 protein in the cortex of uninfected control       bilirubin was detected in cortical neurones (inset), glial cells

                                   Ó 2006 The Authors
                                   Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
                                                                               Haem oxygenase-1 induction in pneumococcal meningitis          537

                                                                        Fig. 3 Effect of pneumococcal meningitis on cortical HO-1 and HO-2
                                                                        protein expression. (a) Representative photomicrographs of HO-1 and
                                                                        HO-2 immunostaining in cortical sections of uninfected (i, iii) and
                                                                        infected (ii, iv) animals (original magnification · 40 objective, scale bar
                                                                        100 lm). Marked induction of HO-1 immunostaining was apparent in
                                                                        neurones (closed arrows), glial cells (arrowheads) and capillary
                                                                        endothelial cells (open arrows) 22 h after infection, whereas basal HO-
                                                                        2 staining in neurones (closed arrows) and capillary endothelial cells
                                                                        (open arrows; inset shows the positively stained endothelial cell layer
                                                                        lining a large artery at the base of the brain) remained unchanged. (b)
                                                                        Representative immunofluorescence photomicrographs of cortical
                                                                        sections of infected animals double-stained for HO-1 (in red), and
                                                                        either microglia/endothelial cells (using FITC-labelled IB4) or astro-
                                                                        cytes (using an antibody against GFAP) (in green). HO-1 co-localized
                                                                        (appearing in yellow) with IB4-positive microglia (closed arrowheads)
                                                                        and endothelial cells (indicated by open arrows in insets), but did not
                                                                        co-localize with GFAP-positive astrocytes (in green, open arrowheads)
                                                                        (original magnification · 40 objective, scale bar 100 lm; original
                                                                        magnification of insets · 20 objective, scale bar 100 lm). (c) Time
                                                                        course of HO-1 and HO-2 protein expression. Cortical homogenates
                                                                        were prepared at the indicated times and subjected to western blot
                                                                        analysis for HO-1 and HO-2. Protein bands were detected at their
                                                                        expected molecular weights. The two different bands for each time
                                                                        point correspond to two independent, individual animals. Bands were
                                                                        evaluated densitometrically, adjusted with respect to b-actin, and
                                                                        expressed as the percentage of the level expressed in uninfected
                                                                        control animals (100%). Values are mean ± SEM of 5–7 animals per
                                                                        time point. *p < 0.05 versus uninfected control (t ¼ 0), using one-way
                                                                        ANOVA followed by Tukey’s multiple comparison test.

                                                                        Effect of pneumococcal meningitis on H- and L-ferritin
                                                                        protein expression
                                                                        Haem degradation through HO-1 has been shown to be
                                                                        associated with up-regulation of ferritin protein, thereby
                                                                        preventing iron-induced oxidative damage (Balla et al.
                                                                        1992). Depending on the cell type, ferritin consists of
                                                                        different proportions of the H- and L-ferritin subunits
                                                                        (Connor et al. 2001). We therefore determined the cellular
                                                                        distribution and temporal expression profile of the two
                                                                        subunits during pneumococcal meningitis.
                                                                           In uninfected control animals, basal H-ferritin staining was
                                                                        readily observed throughout the cortex (Fig. 5a). Similar to
                                                                        HO-1, infection caused a marked increase in H-ferritin
                                                                        staining in neuronal somata, parenchymal blood vessels and
                                                                        glial cells of the cortex. Although prominent L-ferritin
and cerebral endothelial cells of infected animals (Fig. 4b),           immunoreactivity was readily detected in microglia/oligo-
indicating that infection induced an increase in HO activity            dendrocytes of the corpus callosum (not shown), L-ferritin
in these cells. Double-staining with HO-1 (red channel)                 staining was comparatively low in the cortex of uninfected
revealed that most bilirubin-positive cells in the cortex of            control animals (Fig. 5a, panel iii), in line with previously
infected animals were positive for HO-1 (cf. merged image in            reported data (Cheepsunthorn et al. 1998). However,
Fig. 4h). SnPP treatment completely abolished the infection-            L-ferritin staining was markedly increased in activated
induced accumulation of bilirubin in these cells (Fig. 4c).             microglia (as judged by their distribution and morphology)
Together with the other data shown so far, the results provide          located in the cortex of infected animals (closed arrows in
strong evidence that induction of HO-1 is indeed the cause of           Fig. 5a, panel iv). Blood vessel staining was also increased,
the non-haem iron increase.                                             but to a lesser extent. Double-labelling experiments with IB4

Ó 2006 The Authors
Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
538 H. Ren et al.

Fig. 4 In situ detection of cortical bilirubin and effect of SnPP. Rep-    (inset), glial cells (closed arrowheads) and capillary endothelial cells
resentative immunofluorescence photomicrographs of cortical sec-            (open arrows), which was completely inhibited by SnPP treatment. (f)
tions of uninfected (left column), infected (middle column) and infected   In line with the well known effect that, while SnPP inhibits HO-1
animals treated with SnPP (right column) double-stained for bilirubin      activity, it also induces the enzyme (Vreman et al. 1996), HO-1 protein
(using monoclonal antibody 24G7, in green, a–c) and HO-1 (in red,          was up-regulated even more in infected animals treated with SnPP. (h)
d–f). In (a–c) original magnification · 63 objective, scale bar 200 lm.     Overlay of bilirubin and HO-1 staining showing that most bilirubin-
Merged images (g–i) show overlays of the corresponding bilirubin and       positive cells were HO-1 positive. Closed arrows in (f) and (i) denote
HO-1 staining. Infection led to accumulation of bilirubin in neurones      an (HO-1-positive, but bilirubin-negative) neuron.

and anti-GFAP established that the glial cells positive for                protein were up-regulated to a similar extent as in untreated
H- and L-ferritin were microglia (closed arrowheads in                     infected animals (Fig. 6), indicating that the two ferritin
Fig. 5b, panels i and iii). These experiments thus reveal that             subunits were up-regulated for the most part independently
ferritin subunits are up-regulated in the cell types (i.e. neu-            of HO-1-mediated haem degradation. These results suggest
rones, microglia and capillary endothelial cells) that also                that pneumococcal meningitis is not associated with a large
display induction of HO-1.                                                 increase in free iron in the cortex.
   The kinetics of H- and L-ferritin up-regulation was similar
to that of the increase in non-haem iron or induction of HO-1              Effect of pneumococcal meningitis on CSF NPBI and
(Fig. 5c). H-ferritin was significantly increased by  4-fold               cortical DNA/RNA oxidation
at 16 h after infection, whereas L-ferritin was already                    To find evidence in support of this notion, NPBI levels in the
significantly increased by 8 h. Thus, ferritin is up-regulated              CSF were determined by the bathophenanthroline method.
concomitantly with haem degradation.                                       This method measures iron that is only weakly associated
                                                                           with biological ligands (mostly in the form of low molecular
Effect of SnPP treatment on ferritin up-regulation                         weight chelates), and thus potentially redox-active. As with
Because H- and L-ferritin can be up-regulated by both iron-                non-haem iron, total NPBI in the CSF of infected animals
dependent (Eisenstein et al. 1991) and iron-independent                    (1.6 ± 2.0; n ¼ 14) did not differ from that in mock-infected
(Torti and Torti 2002) mechanisms, we studied whether                      control animals (1.8 ± 2.5 lM; n ¼ 9). In both infected and
inhibition of HO activity would affect up-regulation of the                uninfected animals, NPBI was > 95% in the reduced state
two subunits. Although SnPP treatment completely blocked                   (i.e. Fe2+).
the infection-associated increase in non-haem iron and                        The major adduct formed during iron-induced oxidation
bilirubin in the cortex (cf. Figs 2 and 4), H- and L-ferritin              of DNA and RNA, 8OH(d)G, was determined in the cortex by

                                        Ó 2006 The Authors
                                        Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
                                                                               Haem oxygenase-1 induction in pneumococcal meningitis            539

                                                                        Fig. 5 Effect of pneumococcal meningitis on cortical H-ferritin and
                                                                        L-ferritin protein expression. (a) Representative photomicrographs of
                                                                        H- and L-ferritin immunostaining in cortical sections of uninfected (i, iii)
                                                                        and infected (ii, iv) animals (original magnification · 40 objective, scale
                                                                        bar 100 lm). H-ferritin immunoreactivity markedly increased in
                                                                        neurones (closed arrows), glial cells (arrowheads) and capillary
                                                                        endothelial cells (open arrows) upon infection (ii), whereas L-ferritin
                                                                        was predominantly induced in activated microglia, judged by their
                                                                        morphology (iv, closed arrowheads). (b) Representative immunofluo-
                                                                        rescence photomicrographs of cortical sections of infected animals
                                                                        double-stained for either H- or L-ferritin (in red), and either microglia/
                                                                        endothelial cells (using FITC-labelled IB4) or astrocytes (using an
                                                                        antibody against GFAP) (in green). H- and L-ferritin co-localized
                                                                        (appearing in yellow) with IB4-positive microglia (closed arrowheads)
                                                                        and endothelial cells (indicated by open arrows in insets), but did not
                                                                        co-localize with GFAP-positive astrocytes (in green, open arrowheads)
                                                                        (original magnification · 40 objective, scale bar 100 lm; original
                                                                        magnification of insets · 20 objective, scale bar 100 lm). (c) Time
                                                                        course of H- and L-ferritin protein expression. Cortical homogenates
                                                                        were prepared at the indicated times and subjected to western blot
                                                                        analysis for H- and L-ferritin. Protein bands were detected at their
                                                                        expected molecular weights. The two different bands for each time
                                                                        point correspond to two independent, individual animals. Bands were
                                                                        evaluated as described in the legend to Fig. 3. Values are mean ±
                                                                        SEM of 6–7 animals per time point. *p < 0.05, **p < 0.01, ***p < 0.001
                                                                        versus uninfected controls (t ¼ 0 h) using one-way ANOVA followed
                                                                        by Tukey’s multiple comparison test.

                                                                        (Nunomura et al. 1999) caused a substantial increase in DNA/
                                                                        RNA oxidation in sections derived from both mock-infected
                                                                        and infected animals (Figs 7c and d respectively), indicating
                                                                        that the technique per se is capable of detecting oxidative

                                                                        In this study, we have shown that pneumococcal meningitis is
                                                                        associated with a raised level of non-haem iron in the cortex, a
                                                                        brain region affected by necrotic neuronal damage. The
                                                                        infection-associated increase in non-haem iron is the result of
                                                                        HO-mediated haem degradation. Co-localization experiments
                                                                        strongly suggest that induction of HO-1 is the cause of this
                                                                        increase. However, the non-haem iron increase occurred
                                                                        throughout the cortex and was not restricted to focal areas of
                                                                        cortical damage, indicating that injury is not simply a
                                                                        consequence of increased iron. Furthermore, cortical levels
                                                                        of 8OH(d)G, the major oxidative DNA/RNA lesion formed by
                                                                        redox-active iron, did not increase upon infection. The
                                                                        concomitant up-regulation of ferritin independent of HO
                                                                        activity strongly suggests that oxidative DNA/RNA damage
immunofluorescence (Fig. 7), as an indirect marker of redox-             was prevented by the sequestration of liberated iron by ferritin.
active iron. Pneumococcal meningitis was neither associated             The results presented in our study therefore strongly suggest
with increased nuclear or mitochondrial DNA oxidation                   that cortical damage is not the direct result of increased non-
(revealed by the Dnase treatment) nor with increased                    haem iron, not even as an underlying causative factor.
cytoplasmic RNA oxidation (revealed by the Rnase treat-                    Several lines of evidence indicate that the origin of the
ment). Pretreatment of sections with hydrogen peroxide/iron             non-haem iron is the brain parenchyma. Thus, the non-haem

Ó 2006 The Authors
Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
540 H. Ren et al.

                                                                             whereas HO-2 levels remained unchanged. Second, HO-1
                                                                             was induced in the same cells as those exhibiting increased
                                                                             iron staining (i.e. neurones, microglia and capillary endo-
                                                                             thelial cells). Finally, bilirubin production (as an index of HO
                                                                             activity) co-localized with HO-1 and was completely inhib-
                                                                             ited by SnPP.
                                                                                The origin of the haem still remains to be determined. HO
                                                                             has traditionally been viewed as an enzyme located in the
                                                                             endoplasmic reticulum (which would require haem to be
                                                                             intracellular), but more recent reports have raised the
                                                                             possibility that HO-1 might also be located in other subcel-
                                                                             lular domains beside the endoplasmic reticulum, such as in
                                                                             caveloae (Ryter et al. 2006). It is therefore currently not clear
                                                                             whether the haem degraded during pneumococcal meningitis
                                                                             originates from an intracellular pool (e.g. damaged haem
                                                                             proteins) or from an extracellular source (e.g. haemoglobin
                                                                             released by extravasated red blood cells). Regardless of its
                                                                             source, it is important to mention that haem is a potent catalyst
                                                                             of lipid peroxidation and other cytotoxic reactions under
                                                                             certain conditions such as low pH (Everse and Hsia 1997).
                                                                             The hypothesis has recently been put forward that a low level
                                                                             of HO-1 induction limits brain injury in intracerebral
                                                                             haemorrhage, whereas excessive up-regulation after a large
                                                                             haematoma may have a harmful effect (Keep et al. 2005). The
                                                                             induction of HO-1 may thus limit haem-induced toxicity in
                                                                             pneumococcal meningitis.
                                                                                Apart from potentially detoxifying harmful haem, HO-
                                                                             mediated haem degradation also leads to the formation of
                                                                             bilirubin and carbon monoxide, two potent cytoprotective
Fig. 6 Lack of effect of SnPP on infection-associated up-regulation of       molecules (Ferris et al. 1999; Otterbein et al. 2003). Over-
H- and L-ferritin. (a) H- and L-ferritin protein levels were determined by   expression of HO-1 inhibits cortical damage as a result of
western blotting in cortical homogenates of infected animals treated         permanent focal cerebral ischaemia in mice (Panahian et al.
with either SnPP or vehicle. (b) Results are expressed as a                  1999), and neurones overexpressing HO-1 are protected from
mean ± SEM percentage of the level expressed in corresponding                oxidative stress in vitro (Chen et al. 2000). The mechanisms
infected animals corrected for b-actin (100%). Results represent data
                                                                             by which this occurs are not completely understood, but
from 5–7 animals per group. ns, No significant difference using
                                                                             probably involve the action of the potent antioxidant
Student’s t-test.
                                                                             bilirubin (Stocker et al. 1987) and signalling function of
                                                                             carbon monoxide (Mancuso 2004). Bilirubin, for example,
iron increase in the cortex ( 4.5 lg/g tissue) is greater than              inhibits hydrogen peroxide-induced cell death of cultured
the concentration of non-haem iron in plasma, which did not                  cortical neurones at low nanomolar concentrations (Dore
change during infection. Non-haem iron or NPBI levels did                    et al. 1999), and hyperbilirubinaemia has been shown to
also not increase in the CSF. These results do not support the               protect against focal cerebral ischaemia in rats (Kitamura
view that increased cortical non-haem iron resulted from                     et al. 2003). HO-1 may also be neuroprotective indirectly
translocation of plasma non-haem iron across the blood–                      through its induction in the vasculature. Vascular processes
brain barrier. Similarly, infiltrating leucocytes, although                   play an important role in the development of neuronal injury
Perls’ positive, are an unlikely source of the non-haem iron,                in pneumococcal meningitis (Pfister et al. 2000), and induc-
as leucocyte infiltration is limited to the subarachnoid space                tion of HO-1 has been shown to protect vascular endothelium
and ventricles, and spares the brain parenchyma in the acute                 from oxidative damage by inducing ferritin and sequestering
phase of the disease. The experiments with SnPP clearly                      potentially harmful iron (Balla et al. 1992). Carbon monox-
show that the non-haem iron is the result of HO-mediated                     ide acts as a vasorelaxant, much like nitric oxide, and
haem degradation. Our data strongly suggest that induction                   counteracts the vasoconstrictive effect of endothelin (Morita
of HO-1 is responsible for this non-haem iron increase for                   and Kourembanas 1995), known to contribute to the
several reasons. First, the non-haem iron increase during the                development of cortical damage in pneumococcal meningitis
course of infection was paralleled by induction of HO-1,                     (Pfister et al. 2000). We are presently investigating whether

                                         Ó 2006 The Authors
                                         Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
                                                                               Haem oxygenase-1 induction in pneumococcal meningitis   541

Fig. 7 Effect of pneumococcal meningitis
on cortical DNA/RNA oxidation. Consecu-
tive cortical sections of uninfected control
animals and animals infected for 22 h were
subject to either no pretreatment, treatment
with Rnase (for the specific detection of
DNA oxidation) or treatment with Dnase (for
the specific detection of RNA oxidation) and
subsequently incubated with monoclonal
antibody 15A3, which recognizes both
8OHdG and 8OHG (original magnification
· 20 objective, scale bar 100 lm; insets
· 100, scale bar 500 lm). Pneumococcal
meningitis caused neither increased nuc-
lear (see insets) or mitochondrial DNA oxi-
dation (e, f), nor cytosolic RNA oxidation
(g, h). Pretreatment with Dnase and Rnase
combined inhibited staining almost entirely
(not shown), while combined H2O2 and iron
treatment readily increased DNA/RNA oxi-
dation in sections from both uninfected
control and infected animals (c, d).

HO-1 activity determines the outcome of cortical injury in              our data do not favour the view that cortical damage is the
pneumococcal meningitis. Preliminary data show that inhi-               result of iron toxicity. First, non-haem iron increased
bition of HO-1 activity by SnPP exacerbates cortical damage             throughout the cortex and was not restricted to areas of
(H. Ren, S. Leib, S. Christen, unpublished data), suggesting            focal injury, indicating that non-haem iron alone does not
that either the degradation of haem and/or the formation of             cause injury. Furthermore, NPBI levels in the CSF did not
HO-1-derived products may exhibit neuroprotection in                    increase upon infection. CSF NPBI measurements serve as a
experimental pneumococcal meningitis, as has been shown                 surrogate marker for extracellular redox-active iron (Nilsson
for other paradigms of brain injury such as kainate-induced             et al. 2002; Ogihara et al. 2003), as tissue homogenization of
excitotoxicity (Huang et al. 2005). This hypothesis clearly             tissues leads to artifactual release of iron (Kakhlon and
deserves further study.                                                 Cabantchik 2002), which prevents its use for determining
   The initial question raised in this study was whether                changes in redox-active iron. However, redox-active iron in
cortical damage is a consequence of iron toxicity. However,             the extracellular space can be determined by microdialysis.

Ó 2006 The Authors
Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
542 H. Ren et al.

Using an adult rabbit model, we previously showed that                inhibited by SnPP (Carraway et al. 1998). There is clear
extracellular NPBI levels in the cortex do not increase during        evidence that up-regulation of ferritin, especially the H-
pneumococcal meningitis (Bottcher et al. 2004). Further-              subunit, limits cellular oxidative damage and apoptosis.
more, only a modest increase in the formation of hydroxyl             Thus, heterozygous mice deficient in H-ferritin exhibit
radicals (assessed by measuring hydroxylation of phenylal-            increased oxidative damage in the brain (Thompson et al.
anine added to the microdialysis fluid) was observed. This is          2003), and ferritin protects the endothelium from hydrogen
in contrast to pure cerebral ischaemia models, in which both          peroxide-mediated cytotoxicity (Balla et al. 1992).
brain interstitial space and CSF NPBI iron levels markedly               As pointed out in the introduction, the present study was
increase and are associated with increased hydroxyl radical           initiated because of the inhibitory effect of the iron chelator
production (Nilsson et al. 2002). These results indicate that,        DFO on cortical damage. However, if increased iron is not
in contrast to cerebral ischaemia, pneumococcal meningitis is         the cause of cortical brain damage, how then does DFO limit
not associated with an increase of redox-active iron in the           brain injury in pneumococcal meningitis (Auer et al. 2000)?
extracellular space.                                                  As an alternative to scavenging redox-active iron, it is
   To determine whether haem degradation led to an increase           possible that DFO induces neuroprotective pathways via
of intracellular redox-active iron, we determined 8OH(d)G             another mechanisms than preventing iron-induced damage.
levels by immunofluorescence. In contrast to cerebral isch-            Thus, DFO is known to inhibit the activity of the iron-
aemia models (Cui et al. 1999), pneumococcal meningitis was           dependent prolyl hydroxylases, a process that leads to
not associated with increased formation of 8OHdG or 8OHG in           stabilization and activation of the transcription factor hypoxia
the cortex. These results are in line with our previous data that     inducible factor (HIF)-1, which in turn induces gene
pneumococcal meningitis does not lead to increased protein            expression of cytoprotective proteins such as erythropoietin,
oxidation or nitrotyrosine formation in the cortex (Schaper           vascular endothelial growth factor or HO-1 (Ratan et al.
et al. 2002). Taken together, these results do not favour the         2004). Pretreatment of animals with DFO before hypoxia–
view that cortical damage is the result of iron-induced               ischaemia inhibits the development of neuronal injury, an
oxidative damage. This notion is further supported by the fact        effect probably mediated by the induction of HIF-1 (Berger-
that only very few cells stain for terminal deoxynucleotidyl          on et al. 2000). Consistent with the notion that DFO may
transferase-mediated dUTP-biotin in situ nick end labelling           inhibit cortical damage by inducing HIF-1-dependent target
(TUNEL) in the cortex of animals with pneumococcal                    genes, DFO treatment caused up-regulation of HO-1 and
meningitis (Gianinazzi et al. 2003), although DNA oxidation           ferritin in uninfected rat pups, and appeared to enhance the
is known to lead to the strand breaks that are detected by            up-regulation of HO-1 and ferritin in animals with pneumo-
TUNEL staining.                                                       coccal meningitis; at the same time it had no effect on the
   It seems likely that coordinated up-regulation of H- and                                                                      ¨
                                                                      cortical non-haem iron increase (H. Ren, S. Leib, M. Tauber,
L-ferritin prevents a rise in iron-induced oxidative DNA              D. Ferreiro and S. Christen, unpublished results). Whether
damage in pneumococcal meningitis by rapidly sequestering             DFO inhibits cortical damage by this mechanism requires
iron liberated during haem degradation. Thus, HO-1 staining           further study. In addition to a direct effect on parenchymal
overlapped with increased protein expression of H- and                cells, a neuroprotective effect could also be mediated
L-ferritin, both with regard to brain region and cell type.           indirectly via an effect on the vascular endothelium.
Similar to the increase in non-haem iron, a raised level of
H- and L-ferritin expression was observed throughout the
cortex and was not restricted to focal areas of damage.
Furthermore, enhanced ferritin protein expression was inde-                                                     ¨
                                                                      We thank Corinne Siegenthaler and Jurg Kummer for technical
pendent of HO activity, suggesting that ferritin up-regulation        assistance with animal experiments, and Dr Manuela Frese-Schaper
occurred independently of an increase in ‘free iron’.                 for providing tissue sections. This study was supported by the Swiss
Although iron is known to post-transcriptionally up-regulate          National Science Foundation (32-066845.01, 632-066057.01, and
                                                                      31-108236) and the National Institute of Neurological Disorders and
H- and L-ferritin by promoting the dissociation of iron-
                                                                      Stroke (NS33997 and NS35902).
regulatory protein-1 from preformed mRNA, thereby
enabling its transcription, it is well known that both H-ferritin
(Miller et al. 1991; Torti and Torti 2002) and L-ferritin             References
(Hintze and Theil 2005) can also be up-regulated at the                                                      ¨
                                                                      Auer M., Pfister L. A., Leppert D., Tauber M. G. and Leib S. L. (2000)
transcriptional level by cytokines (e.g. tumour necrosis                    Effects of clinically used antioxidants in experimental pneumo-
factor-a or interleukin-1b) or factors that activate the                    coccal meningitis. J. Infect. Dis. 182, 347–350.
antioxidant response element. HO-independent up-regulation            Balla G., Jacob H. S., Balla J., Rosenberg M., Nath K., Apple F., Eaton
of ferritin has been shown in other studies. Thus, similar to               J. W. and Vercellotti G. M. (1992) Ferritin: a cytoprotective
                                                                            antioxidant strategem of endothelium. J. Biol. Chem. 267, 18 148–
the situation in pneumococcal meningitis, the up-regulation                 18 153.
of pulmonary ferritin by lipopolysaccharide is also not

                                    Ó 2006 The Authors
                                    Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
                                                                                      Haem oxygenase-1 induction in pneumococcal meningitis          543

Bergeron M., Ferriero D. M., Vreman H. J., Stevenson D. K. and Sharp           Huang E., Ong W. Y., Go M. L. and Garey L. J. (2005) Heme oxyge-
      F. R. (1997) Hypoxia–ischemia, but not hypoxia alone, induces the             nase-1 activity after excitotoxic injury: immunohistochemical
      expression of heme oxygenase-1 (HSP32) in newborn rat brain.                  localization of bilirubin in neurons and astrocytes and deleterious
      J. Cereb. Blood Flow Metab. 17, 647–658.                                      effects of heme oxygenase inhibition on neuronal survival after
Bergeron M., Gidday J. M., Yu A. Y., Semenza G. L., Ferriero D. M. and              kainate treatment. J. Neurosci. Res. 80, 268–278.
      Sharp F. R. (2000) Role of hypoxia-inducible factor-1 in hypoxia-        ILARCLS (1996) Guide for the Care and Use of Laboratory Animals.
      induced ischemic tolerance in neonatal rat brain. Ann. Neurol. 48,            National Academy Press, Washington, D.C.
      285–296.                                                                 Kakhlon O. and Cabantchik Z. I. (2002) The labile iron pool: charac-
Bottcher T., Ren H., Goiny M. et al. (2004) Clindamycin is neuropro-                terization, measurement, and participation in cellular processes (1).
      tective in experimental Streptococcus pneumoniae meningitis                   Free Radic. Biol. Med. 33, 1037–1046.
      compared with ceftriaxone. J. Neurochem. 91, 1450–1460.                  Keep R. F., Xi G., Hua Y. and Hoff J. T. (2005) The deleterious or
Carraway M. S., Ghio A. J., Taylor J. L. and Piantadosi C. A. (1998)                beneficial effects of different agents in intracerebral hemorrhage:
      Induction of ferritin and heme oxygenase-1 by endotoxin in the                think big, think small, or is hematoma size important? Stroke 36,
      lung. Am. J. Physiol. 275, L583–L592.                                         1594–1596.
Cheepsunthorn P., Palmer C. and Connor J. R. (1998) Cellular distri-           Kim C. K. and Rivier C. L. (2000) Nitric oxide and carbon monoxide
      bution of ferritin subunits in postnatal rat brain. J. Comp. Neurol.          have a stimulatory role in the hypothalamic-pituitary-adrenal
      400, 73–86.                                                                   response to physico-emotional stressors in rats. Endocrinology
Chen K., Gunter K. and Maines M. D. (2000) Neurons overexpressing                   141, 2244–2253.
      heme oxygenase-1 resist oxidative stress-mediated cell death.            Kitamura Y., Ishida Y., Takata K. et al. (2003) Hyperbilirubinemia
      J. Neurochem. 75, 304–313.                                                    protects against focal ischemia in rats. J. Neurosci. Res. 71, 544–
Connor J. R., Menzies S. L., Burdo J. R. and Boyer P. J. (2001) Iron and            550.
      iron management proteins in neurobiology. Pediatr. Neurol. 25,           Leib S. L., Clements J. M., Lindberg R. L., Heimgartner C., Loeffler
      118–129.                                                                                             ¨
                                                                                    J. M., Pfister L. A., Tauber M. G. and Leppert D. (2001) Inhibition
Cui J., Holmes E. H. and Liu P. K. (1999) Oxidative damage to the c-fos             of matrix metalloproteinases and tumour necrosis factor alpha
      gene and reduction of its transcription after focal cerebral ischemia.        converting enzyme as adjuvant therapy in pneumococcal menin-
      J. Neurochem. 73, 1164–1174.                                                  gitis. Brain 124, 1734–1742.
Dore S., Takahashi M., Ferris C. D., Zakhary R., Hester L. D., Guastella       Maines M. D. (1997) The heme oxygenase system: a regulator of second
      D. and Snyder S. H. (1999) Bilirubin, formed by activation of                 messenger gases. Annu. Rev. Pharmacol. Toxicol. 37, 517–554.
      heme oxygenase-2, protects neurons against oxidative stress injury.      Mancuso C. (2004) Heme oxygenase and its products in the nervous
      Proc. Natl Acad. Sci. USA 96, 2445–2450.                                      system. Antioxid. Redox Signal. 6, 878–887.
Eisenstein R. S., Garcia-Mayol D., Pettingell W. and Munro H. N.               Miller L. L., Miller S. C., Torti S. V., Tsuji Y. and Torti F. M. (1991)
      (1991) Regulation of ferritin and heme oxygenase synthesis in rat             Iron-independent induction of ferritin H chain by tumor necrosis
      fibroblasts by different forms of iron. Proc. Natl Acad. Sci. USA              factor. Proc. Natl Acad. Sci. USA 88, 4946–4950.
      88, 688–692.                                                             Moos T. and Morgan E. H. (1998) Evidence for low molecular weight,
Everse J. and Hsia N. (1997) The toxicities of native and modified                   non-transferrin-bound iron in rat brain and cerebrospinal fluid.
      hemoglobins. Free Radic. Biol. Med. 22, 1075–1099.                            J. Neurosci. Res. 54, 486–494.
Ewing J. F. and Maines M. D. (1991) Rapid induction of heme oxyge-             Morita T. and Kourembanas S. (1995) Endothelial cell expression
      nase 1 mRNA and protein by hyperthermia in rat brain: heme                    of vasoconstrictors and growth factors is regulated by smooth
      oxygenase 2 is not a heat shock protein. Proc. Natl Acad. Sci. USA            muscle cell-derived carbon monoxide. J. Clin. Invest. 96,
      88, 5364–5368.                                                                2676–2682.
Ferris C. D., Jaffrey S. R., Sawa A. et al. (1999) Haem oxygenase-1            Nguyen-Legros J., Bizot J., Bolesse M. and Pulicani J. P. (1980) ‘Di-
      prevents cell death by regulating cellular iron. Nat. Cell Biol. 1,           aminobenzidine black’ as a new histochemical demonstration of
      152–157.                                                                      exogenous iron. Histochemistry 66, 239–244.
Fukuda K., Richmon J. D., Sato M., Sharp F. R., Panter S. S. and Noble         Nilsson U. A., Bassen M., Savman K. and Kjellmer I. (2002) A simple
      L. J. (1996) Induction of heme oxygenase-1 (HO-1) in glia after               and rapid method for the determination of ‘free’ iron in biological
      traumatic brain injury. Brain Res. 736, 68–75.                                fluids. Free Radic. Res. 36, 677–684.
Gianinazzi C., Grandgirard D., Imboden H., Egger L., Meli D. N.,               Nunomura A., Perry G., Pappolla M. A., Wade R., Hirai K., Chiba S. and
      Bifrare Y. D., Joss P. C., Tauber M. G., Borner C. and Leib S. L.             Smith M. A. (1999) RNA oxidation is a prominent feature of
      (2003) Caspase-3 mediates hippocampal apoptosis in pneumo-                    vulnerable neurons in Alzheimer’s disease. J. Neurosci. 19, 1959–
      coccal meningitis. Acta Neuropathol. (Berl.) 105, 499–507.                    1964.
Halliwell B. (1992) Reactive oxygen species and the central nervous            Ogihara T., Hirano K., Ogihara H. et al. (2003) Non-protein-bound
      system. J. Neurochem. 59, 1609–1623.                                          transition metals and hydroxyl radical generation in cerebrospinal
Henle E. S., Luo Y., Gassmann W. and Linn S. (1996) Oxidative damage                fluid of newborn infants with hypoxic ischemic encephalopathy.
      to DNA constituents by iron-mediated fenton reactions. The                    Pediatr. Res. 53, 594–599.
      deoxyguanosine family. J. Biol. Chem. 271, 21 177–21 186.                Otterbein L. E., Soares M. P., Yamashita K. and Bach F. H. (2003) Heme
Higgins T. (1981) Novel chromogen for serum iron determinations. Clin.              oxygenase-1: unleashing the protective properties of heme. Trends
      Chem. 27, 1619–1620.                                                          Immunol. 24, 449–455.
Hintze K. J. and Theil E. C. (2005) DNA and mRNA elements with                 Panahian N., Yoshiura M. and Maines M. D. (1999) Overexpression of
      complementary responses to hemin, antioxidant inducers, and iron              heme oxygenase-1 is neuroprotective in a model of permanent
      control ferritin-L expression. Proc. Natl Acad. Sci. USA 102,                 middle cerebral artery occlusion in transgenic mice. J. Neurochem.
      15 048–15 052.                                                                72, 1187–1203.
Honda K., Smith M. A., Zhu X. et al. (2005) Ribosomal RNA in Alz-              Park E. M., Shigenaga M. K., Degan P., Korn T. S., Kitzler J. W., Wehr
      heimer disease is oxidized by bound redox-active iron. J. Biol.               C. M., Kolachana P. and Ames B. N. (1992) Assay of excised
      Chem. 280, 20 978–20 986.                                                     oxidative DNA lesions: isolation of 8-oxoguanine and its nucleo-

Ó 2006 The Authors
Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544
544 H. Ren et al.

     side derivatives from biological fluids with a monoclonal antibody     Stocker R., Yamamoto Y., McDonagh A. F., Glazer A. N. and Ames B.
     column. Proc. Natl Acad. Sci. USA 89, 3375–3379.                            N. (1987) Bilirubin is an antioxidant of possible physiological
Pennell N. A., Hurley S. D. and Streit W. J. (1994) Lectin staining of           importance. Science 235, 1043–1046.
     sheep microglia. Histochemistry 102, 483–486.                         Thompson K., Menzies S., Muckenthaler M., Torti F. M., Wood T., Torti
Pfister L. A., Tureen J. H., Shaw S., Christen S., Ferriero D. M., Tauber         S. V., Hentze M. W., Beard J. and Connor J. (2003) Mouse brains
     M. G. and Leib S. L. (2000) Endothelin inhibition improves cer-             deficient in H-ferritin have normal iron concentration but a protein
     ebral blood flow and is neuroprotective in pneumococcal menin-               profile of iron deficiency and increased evidence of oxidative
     gitis. Ann. Neurol. 47, 329–335.                                            stress. J. Neurosci. Res. 71, 46–63.
Ratan R. R., Siddiq A., Aminova L., Lange P. S., Langley B., Ayoub I.,     Torti F. M. and Torti S. V. (2002) Regulation of ferritin genes and
     Gensert J. and Chavez J. (2004) Translation of ischemic precon-             protein. Blood 99, 3505–3516.
     ditioning to the patient: prolyl hydroxylase inhibition and hypoxia   Vreman H. J., Cipkala D. A. and Stevenson D. K. (1996) Characteri-
     inducible factor-1 as novel targets for stroke therapy. Stroke 35,          zation of porphyrin heme oxygenase inhibitors. Can. J. Physiol.
     2687–2689.                                                                  Pharmacol. 74, 278–285.
Ryter S. W., Alam J. and Choi A. M. (2006) Heme oxygenase-1/carbon         Wang X. S., Ong W. Y. and Connor J. R. (2002) Increase in ferric and
     monoxide: from basic science to therapeutic applications. Physiol.          ferrous iron in the rat hippocampus with time after kainate-induced
     Rev. 86, 583–650.                                                           excitotoxic injury. Exp. Brain Res. 143, 137–148.
                                             ¨                     ¨
Schaper M., Gergely S., Lykkesfeldt J., Zbaren J., Leib S. L., Tauber      Yu S. W., Wang H., Dawson T. M. and Dawson V. L. (2003) Poly (ADP-
     M. G. and Christen S. (2002) Cerebral vasculature is the major              ribose) polymerase-1 and apoptosis inducing factor in neurotoxi-
     target of oxidative protein alterations in bacterial meningitis.            city. Neurobiol. Dis. 14, 303–317.
     J. Neuropathol. Exp. Neurol. 61, 605–613.

                                        Ó 2006 The Authors
                                        Journal Compilation Ó 2006 International Society for Neurochemistry, J. Neurochem. (2007) 100, 532–544

Shared By: