Expression of immunological molecules by cardiomyocytes and inflammatory and interstitial cells in rat autoimmune myocarditis.

BACKGROUND
In a heart with myocarditis, there are cardiomyocytes, inflammatory cells, and non-inflammatory interstitial cells. Immunological molecules are thought to influence not only inflammatory cells but also cardiac function and remodeling. Whatever their origin, the cells they target and the intercellular crosstalk they mediate remain unclear. Here, we examined native gene expression of immunological molecules in normal and rat experimental autoimmune myocarditis (EAM) 18 and 90 days after immunization, using real time RT-PCR in cardiomyocytes, CD11b(+) cells, alphabetaT cells and non-cardiomyocytic non-inflammatory (NCNI) cells.


METHODS AND RESULTS
Cells were isolated by collagenase perfusion on a Langendorff apparatus and purified by passing through a stainless-steel sieve followed by magnetic bead column separation using appropriate monoclonal antibodies. Most immunological molecules were expressed in inflammatory cells. However, some were expressed in NCNI cells or cardiomyocytes. Interestingly, most of interleukin (IL)-10, monocyte chemoattractant protein (MCP)-1, or tumor necrosis factor (TNF)-alpha receptor were found in NCNI cells and most of fractalkine were found in NCNI cells and cardiomyocytes. Moreover, TNF-alpha significantly upregulated fractalkine and MCP-1 mRNA in cultivated cells from EAM hearts.


CONCLUSION
In the rat experimental myocarditis heart, inflammatory cells express many immunological molecules. Some of them are thought to influence NCNI cells or cardiomyocytes directly via receptors on these cell types. It is further suggested that fractalkine, IL-10, and MCP-1 expressed in NCNI cells or cardiomyocytes regulate inflammatory cells.


Introduction
Rat experimental autoimmune myocarditis (EAM) resembles human giant cell myocarditis [1] and recurrent forms lead to dilated cardiomyopathy (DCM) [2]. CD4 + ahT cells play important roles in initiating the disease process, while macrophages and CD4 + ahT cells infiltrate the heart during the acute phase [3,4]. Gene expression of immunological molecules in EAM heart changes diversely from acute phase to recovery phase [5]. For example, Th1 cytokine increases in the acute phase and decreases during the recovery phase. On the other hand, Th2 cytokine increases during the recovery phase [5,6]. Immunological molecules, in addition to their influence on inflammatory cells, are believed also to affect cardiac function and mediate myocardial damage [7]. It is thought that inflammatory cells and non-inflammatory cells in EAM heart engage in crosstalk by means of immunological molecules. Thus far, in vivo analysis of immunological molecules and their receptor-expressing origin cells has not been performed satisfactorily. It is important to determine which cells express these molecules or their ligands and receptors in order to understand deterioration of cardiac function and remodeling in myocarditis. 0008 In this study, we investigated native gene expression of immunological molecules by using real time RT-PCR in cardiomyocytes, ahT cells, CD11b + cells (macrophages/ granulocytes/dendritic cells in part) or non-cardiomyocytic non-inflammatory (NCNI) cells without cultivation. Cells were isolated and purified by passing through a stainlesssteel sieve and using appropriate monoclonal antibodies on a magnetic bead column separation system. Moreover, we examined the effect of tumor necrosis factor (TNF)-a on immunological molecules in cultivated cells from EAM hearts. We aimed to elucidate implications on the potential crosstalk in EAM.

Animals
Lewis rats were obtained from Charles River, Japan (Atsugi, Kanagawa, Japan) and were maintained in our animal facilities until they reached 7 weeks of age. Throughout the studies, all the animals were treated in accordance with the guidelines for animal experiments of our institute and the guide for the care and use of laboratory animals published by the US National Institutes of Health.

Induction of EAM
Whole cardiac myosin was prepared from the ventricular muscle of porcine hearts as previously described [1]. It was dissolved in PBS at a concentration of 10 mg/ml and emulsified with an equal volume of complete Freund's adjuvant supplemented with 10 mg/ml of Mycobacterium tuberculosis H37RA (Difco, Detroit, Michigan). On day 0, the rats received a single immunization at 2 subcutaneous sites with a total of 0.2 ml of emulsion for each rat. Normal rats (n = 5), EAM rats were killed on day 18 (n = 6) (acute phase), and EAM rats were killed on day 90 (n = 6) (chronic phase).

Isolation of cells, flow cytometric analysis and cell purification
Cardiomyocytes and non-cardiomyocytes in hearts of normal and myocarditis rats on days 18 and 90 were isolated after collagenase perfusion treatment for 15 -20 min using a Langendorff apparatus as reported previously [8,9]. Isolated cells in an isotonic buffer were separated serially through a 38 Am stainless-steel sieve twice and a 20 Am stainless-steel sieve twice. Cells larger than 38 Am and smaller than 20 Am were considered as cardiomyocytes and non-cardiomyocytes, respectively (Fig. 1). The cell fraction between 20 and 38 Am consisted of both cell types and was therefore discarded. The number of non-cardiomyocytes recovered from normal hearts was very small but the number from myocarditis hearts on day 18 was sufficient for statistically significant analysis. Phenotype was analyzed in both samples and cell purification was performed only in samples obtained from myocarditis hearts on day 18.
TCRa/h + T cells, TCRa/h À /CD11b + cells, and TCRa/h À / CD11b À cells (NCNI cells) were sorted using PE-conjugated monoclonal antibodies, anti-PE micro beads (Miltenyi Biotec, Bergisch Gladbach, Germany) and a MACS magnetic cell sorting system (Miltenyi Biotech) [10]. In brief, cells were serially labeled with PE-conjugated anti-TCRa/h and anti-PE micro beads and separated with a MS column in a magnetic field. The positive fraction was further purified using another MS column (purity, 99.4 T 0.5%, n = 3). Contaminated cells in the negative fraction were removed using a LD column and utilized for the succeeding sorting. The TCRa/h À cells were serially labeled with PE-conjugated anti-CD11b and anti-PE micro beads, then separated with a MS column in a magnetic field. The positive fraction was further purified using another MS column (purity, 97.0 T 2.3%, n = 3). Contaminated cells in the negative fraction were removed using a LD column (final contamination of positive cells, 2.2 T 1.1%, n = 3, Fig. 2).

Immunostaining
Cytospin preparations of purified cell fractions were made and slides were stained with May -Giemsa stain or naphthol AS-D chloroacetate esterase and a-naphthylbutyrate esterase stain. Cytospin slides prepared from NCNI cell fraction were immunostained with mouse monoclonal anti-a smooth muscle actin antibodies (SIGMA, Saint Louis, MO), rabbit anti-Factor VIII related antigen antibodies (Zymed Laboratories, San Francisco, CA) and rabbit anti-rat collagen III antibodies (Monosan, Uden, Netherlands). Sections were incubated for 60 min at 37 -C in a humidified chamber with this antibody. The slides were washed in TBS three times. Immunodetection was performed using biotinylated anti-rabbit and anti-mouse immunoglobulins followed by alkaline phosphatase conjugated streptavidin and a Fast Red chromogen (kit LSAB2; DAKO Corp., Carpinteria, CA) for red staining. The sections were lightly counterstained with Mayer's hematoxylin. Negative control slides were incubated with either mouse IgG2a or normal rabbit serum instead of the primary antibody. Sections of EAM heart were stained with hematoxylin and eosin stain. For immunostaining of OPN, paraffin sections of EAM heart on day 18 were cut at 6 Am, deparaffinized with xylene, hydrated with decreasing concentrations of ethanol and heated in a hot water bath for 40 min at 95 -C in ChemMate Target retrieval solution (Daco Corp.). The sections were probed with rabbit anti-human OPN polyclonal antibody (1 : 50 dilution; IBL, Gunma, Japan) for 60 min in a humidified chamber. After several washings with TBS, immunodetection was performed using biotinylated anti-rabbit and anti-mouse immunoglobulins followed by alkaline phosphatase conjugated streptavidin and a Fast Red chromogen. The sections were lightly counterstained with Mayer's hematoxylin.

RNA isolation from heart and reverse transcription
We prepared 1 to 5 Â 10 6 purified cells and total RNA was isolated from each purified cell fraction (cardiomyocytes; n = 5, CD11b + cells; n = 5, ahT cells; n = 5 and NCNI cells; n = 6) of EAM hearts on day 18 (acute phase) after immunization and from each cardiomyocytes fraction of normal hearts (n = 5) and EAM hearts on day 90 (chronic phase n = 6) using Trizol (LifeTechnologies, Tokyo, Japan) [5]. To confirm that gene expression of cardiomyocytes after collagenase preparation is similar to native cardiomyocytes, we prepared purified cardiomyocytes (n = 4) and whole heart (n = 4), which are homogenated immediately in Trizol, on days 0, 14, 21 and 28 and then total RNA was isolated from each sample. cDNA was synthesized from 2 -5 Ag of total RNA with random primers and murine Moloney leukemia virus reverse transcriptase in a final volume of 20 Al. Osaka, Japan), for each primer (Table 1) from 1 Al of cDNA according to the following amplification protocol: 35 cycles at 94 -C for 60 s, 58 -C for 90 s, and 73 -C for 120 s. Amplified cDNAs were directly inserted into the pGEM-T easy vector and the recombinant plasmids were isolated after transforming with Escherichia coli JM109 competent cells using the MagExtractor plasmid kit (TOYOBO, Osaka, Japan). The plasmids were diluted with DNAse-free water in a siliconised tube including 10 ng/Al MS2 RNA (Roche, Indianapolis, IN) to prevent adherence to the tube wall.

Quantitative RT-PCR
cDNA was diluted 100-fold with DNAse-free water in a siliconised tube (including 10 ng/Al MS2 RNA to prevent adherence to the tube wall) and 5 Al was then used for realtime PCR. cDNA and diluted plasmid were amplified with the same primer used for making the plasmid and Light-Cycler-FastStart DNA Master SYBR Green I (Roche, Indianapolis, IN). RNAs without reverse transcription were used as a negative control. After an initial denaturation step of 10 min at 95 -C, a three-step cycle procedure was used (denaturation 95 -C, 10 s, annealing 62 -65 -C, 10 s and extension 72 -C, 13 s) for 40 cycles. The LightCycler Software calculated a standard curve using five plasmid standards. The standard curve was created by plotting the cycle numbers at which the fluorescent signals entered the log-linear phase vs the concentrations of the standards. The absolute copy numbers of all the samples were calculated by the LightCycler software using this standard curve [5].

Non-cardiomyocytic cell culture with TNF-a
On day 18, non-cardiomyocytic (NC) cells were isolated from the hearts of EAM rats via collagenase preparation and were cultured for three weeks on 35-mm well dishes in 2 ml of RPMI medium supplemented with 10% FCS. These cultivated NC cells were suggested to contain mainly fibroblasts, smooth muscle cells, endothelial cells and CD11b + cells and to express enough mRNA of TNF-a receptor, as determined by gene expression analysis ( Table  2). After reaching confluency, NC cells were stimulated by adding TNF-a (Pepro Tech, London, England) (no TNF-a group, n = 6; 80 ng/ml of TNF-a group, n = 6; and 160 ng/ml of TNF-a group, n = 6). After culture for 24 h at 37 -C, NC Table 1 List of primers Sense primer Antisense primer cells were collected and total RNA was isolated as described above. The absolute copy numbers of g-actin, fractalkine, monocyte chemoattractant protein (MCP)-1 and osteopontin mRNA were measured by quantitative real-time PCR.

Statistical analysis
Data obtained from quantitative RT-PCR are expressed as mean T standard error of the mean (SEM). Data obtained from flow cytometric analysis are expressed as mean T standard deviation (SD). Differences between groups of purified cells were determined by one-way ANOVA and Bonferroni_s multiple comparison test. A value of P 0.05 was considered statistically significant.

Flow cytometry of non-cardiomyocytes
The number of CD4 + ahT cells and CD11b + cells increased on day 18 as compared with the control. On day 18, 65% of non-cardiomyocytes were CD11b + cells and ED- Results are expressed as the meanTSD. 1 + cells (macrophages/granulocytes) and 13% of them were CD3 + T cells which were almost all TCRah + and CD4 + . CD161 + NK cells were hardly found in the non-cardiomyocytes (Table 3).

Staining of purified cell
Images of cells stained with May -Giemsa stain or naphthol AS-D chloroacetate esterase and a-naphthylbutyrate esterase stain in ahT cell and CD11b + cell fractions were compatible with typical images of T cells or macrophage/granulocytes. However, images of cells in a NCNI cell fraction differed from those of cardiomyocytes, T cells or macrophage/granulocytes. Cells in a NCNI cell fraction showed positive staining for a smooth muscle actin, Factor VIII related antigen or collagen III (Fig. 3).

Gene expression of immunological molecule in separated cells from EAM hearts
Gene expression of a-cardiac myosin, which should be found only in cardiomyocytes, was in fact detected only in a cardiomyocyte fraction and not in the other fractions (Table 4). T cell specific marker CD3 was detected in a ahT cell fraction but CD3 gene expression in the other fractions was under 5% presumably due to ahT cells contaminating those fractions and a few gyT cells in a NCNI cell fraction. CD11b, which should be found in macrophages and granulocytes, was expressed in a CD11b + cell fraction, but its gene expression in a ahT cell fraction was under 8% and that in the other fractions was under 3%, again presumed to be due to contaminating CD11b + cells. The fibroblast-specific marker (collagen type III) was detected in a NCNI cell fraction but its expression in the other fractions was under 1%. The endothelial cell-specific marker (von Willebrand factor) was detected in a NCNI cell fraction but its expression in the other fractions was under 10%. Because gene expression in smooth muscle cells is similar to that in cardiomyocytes and macrophages, it is difficult to find a specific marker for smooth muscle cells. However, generally speaking, because calponin [11] and caldesmo [12], which smooth muscle cells contain, were detected in a NCNI cell franction, most smooth muscle cells were presumed to be in that fraction.
Gene expression of cardiac myosin isoform and ANP was examined in purified cardiomyocytes after collagenase preparation and native cardiomyocytes, namely whole hearts immediately homogenated in Trizol, on days 0, 14, 21 and 28 to confirm that gene expression in cardiomyocytes was not influenced by collagenase preparation. Their ratios of copy numbers at each day to copy numbers at day 0 were almost equivalent for purified cardiomyocytes and native cardiomyocytes (Fig. 4). Table 4 Absolute copy numbers of specific cell marker mRNA in day 18 EAM hearts Copy numbers of mRNA/microgram of total RNA (copy numbers of mRNA/the most copy numbers of mRNA) Cardiomyocytes (n =5)  Gene expression of immunological molecules was examined in fractions of purified cells (cardiomyocytes, CD11b + cells, ahT cells and NCNI cells) from day 18 EAM hearts (Table 5). Interleukin (IL)-2 and interferon (IFN)-g were detected in a ahT cell fraction alone and IFN-g receptor was detected in CD11b + cell, ahT cells, NCNI cell fractions and, to a lesser degree, in a cardiomyocyte fraction. IL-10 ( Fig. 5A) and MCP-1 (Fig. 5B) in a NCNI cell fraction were significantly more than the other fractions and IL-10 receptor was detected in CD11b + cell and ahT cell fractions. TNF-a was detected in a CD11b + cell fraction and TNF-a receptor was detected in a NCNI cell fraction and, to a lesser degree, in CD11b + cell and cardiomyocyte fractions. Osteopontin was detected in CD11b + cell and NCNI cell fractions. One of its ligands, CD44, was found mainly in a ahT cell fraction, smaller amounts were found in CD11b + cell and NCNI cell fractions, and very little was found in a cardiomyocyte fraction. Most of programmed death-1 (PD-1) was detected in a ahT cell fraction and PD-1 ligand mainly in a CD11b + cell fraction. Fractalkine was detected in NCNI cell and cardiomyocyte fractions (Fig. 5C) and its receptor CX3CR1 in ahT cell and CD11b + cell fractions (Fig. 5D). Major histocompatibility complex (MHC) Class I and MHC Class II molecules were detected mostly in ahT cell and CD11b + cell fractions respectively, but both were detected in all fractions. Expression of the genes for fractalkine, IFN-g receptor, TNF-a receptor, CD44, MHC Class I and MHC Class II, thought to be present in cardiomyocytes, was examined in cardiomyocyte fractions of normal and day 18 and 90 EAM hearts (Table 6). Fractalkine, TNFa receptor, CD44, MHC Class I and MHC Class II expression significantly increased in a fraction of cardiomyocytes of acute myocarditis hearts.

Effect of TNF-a on gene expression of immunological molecules in cultivated NC cells from EAM hearts
Because NCNI cells strongly expressed TNF-a receptor ( Table 2), we examined the effect of TNF-a on fractalkine, MCP-1 and osteopontin which they expressed. In cultivated NC cells containing fibroblasts, smooth muscle cells, endothelial cells and CD11b + cells, fractalkine (160 ng/ml of TNF-a group, 20.0 T 4.8-fold, P < 0.001) and MCP-1 (160 ng/ml of TNF-a group, 63.2 T 8.0-fold, P < 0.0001) were significantly upregulated by TNF-a ( Fig. 6A and B). However, we could not find the significant influence of gene expression of osteopontin by TNF-a in cultivated NC cells (Fig. 6C).

Immunostaining of osteopontin
Osteopontin expression was observed in some mononuclear cells infiltrating into EAM hearts on day 18 (Fig.   7B). However, osteopontin expression was hardly detected in normal heart (Fig. 7C).

Discussion
In this study, gene expression of some immunological molecules (fractalkine, IL-10, MCP-1, CD44, IFN-g receptor, TNFa receptor, etc.) was found in non-inflammatory cells. This suggests that they may be mediators not only for inflammatory cell activity but also for originally constitutive cells in the heart-cardiomyocyte and NCNI cells.
Fractalkine has been identified as a novel chemokine that exhibits cell adhesion and chemoattractive properties in the central nervous system [13]. It is interesting that fractalkine mRNA is found in purified cardiomyocytes and NCNI cells. This result supports previous reports that endothelial cells of the coronary vasculature and endocardium were the principal source of fractalkine and some fractalkine immunoreactivity was also found on the myocytes [14]. Normal cardiomyocytes express some fractalkine, but during the acute phase of inflammation expression increases. TNF-a produced by CD11b + cells in EAM hearts strongly upregulated fractalkine mRNA in cultivated NC cells. It has been reported that fractalkine expression is markedly induced by inflammatory cytokines, TNF-a, IL-1 and IFN-g in primary cultured endothelial cells [15,16]. Because cytokines as TNF-a, IL-1h or IFN-g increase in EAM hearts [5], fractalkine mRNA in cardiomyocytes of EAM hearts may be upregulated. CX3CR1, receptor of fractalkine, was found in CD11b + cells and ahT cells. This suggests that cytokines produced by inflammatory cells increase fractalkine on residential cells and then it attracts and activates inflammatory cells in EAM hearts. Recently, it was reported that fractalkine was secreted by central nervous system neurons and astrocytes [17]. Neuronal  fractalkine expression in virus encephalitis plays important roles in macrophage recruitment and neuroprotection in the central nervous system [18]. Fractalkine may play a similar role in the heart. IL-10 was expressed unexpectedly in NCNI cells. Before this study, we thought that cells expressing IL-10 were Th2 cytokine-secreting T cells. We could detect IL-2 and IFN-g gene expression in ahT cells, but IL-10 gene expression was barely detected in them. Gene expressions of IL-4 and IL-13 (Th2 cytokines), which are produced only by hemopoietic cells, were not found in EAM whole hearts on days 18 and 28 by using quantitative RT-PCR [data not shown]. Therefore, it is thought that ahT cells in acute EAM hearts secrete Th1 but not Th2 cytokine. Previously, we reported that IL-10 producing cells in peripheral blood increased on day 28 in EAM, however, on day 28, most parts of the infiltrating cells disappeared and the fibrosis remained in the myocardium [19]. In rheumatoid arthritis, IL-10 is released by fibroblast-like synoviocytes [20]. It may be that, in EAM hearts, fibroblasts, not T cells, secrete most of IL-10 and it modulate CD11b + and ahT cells expressing IL-10 receptors by paracrine mechanisms. It has been reported that IL-10 gene therapy ameliorated EAM and other models of autoimmune inflammatory disease [21], therefore NCNI cells producing IL-10 may play an important role for EAM.
It has been reported that MCP-1 is secreted by activated lymphocytes, macrophages, endothelial cells, smooth muscle cells and platelet-derived growth factor-activated fibroblasts [22 -24]. Our previous immunohistochemistry analysis in EAM showed that large monocytes were stained by anti-MCP-1 antibody [25]. This study suggests that they are not macrophages but NCNI cells, which attract macrophages into myocarditis hearts by paracrine mechanisms. It has been reported that MCP-1 secretion was regulated by the proinflammatory cytokines, IL-1 and TNF-a in colonic subepithelial myofibroblasts [26]. In this study, we also showed that MCP-1 mRNA in cultivated NC cells from EAM hearts were upregulated by TNF-a. Increase of TNFa and IL-1h produced by macrophages in EAM hearts [5] upregulates expression of MCP-1 gene in NCNI cells and it may attract macrophage.
We previously reported that large amounts of osteopontin mRNA are expressed in early EAM [5]. Osteopontin protein was detected in EAM hearts by immunohistochemistry. Osteopontin is an extracellular matrix protein as well as a cytokine that contributes to the development of Th1 immunity [27]. Osteopontin is secreted by CD11b + cells and NCNI cells. It may activate T cells expressing CD44 as one of the ligands for osteopontin [28]. It may also modulate CD11b + cells, NCNI cells and to a lesser degree probably cardiomyocytes in myocarditis. Osteopontin is also a potentially important mediator of AII regulation of cardiac fibroblast behavior in the cardiac remodeling process [29].
PD-1 deficiency causes a variety of autoimmune diseases [30] and dilated cardiomyopathy with severely impaired contraction and death by congestive heart failure [31]. It has been reported that the parenchymal cells of heart, lung and kidney constitutively express PD-1 ligand [30]. Cytokine such as IFN-g or other inflammatory stimuli induces PD-1 ligand expression [32]. In this study, CD11b + cells in EAM hearts expressed both IFN-g receptor and PD-1 ligand. CD11b + cells stimulated by IFN-g may regulate ahT cells activation in EAM by PD-1 -PD1 ligand binding. However, more evidence is needed to support this concept.
CD4 + T cells play important roles in EAM. MHC Class II expressing cells that can bind CD4 are mainly CD11b + cells. Some papers reported that MHC Class I and MHC Class II expression increases in myocarditis hearts or DCM [33 -35]. This suggest that their expression in cardiomyocytes in acute myocarditis increased, which may mean that lymphocytes have contact with cardiomyocytes in acute myocarditis more closely than in normal heart or in chronic myocarditis.
This method of isolating and purifying the subgroups of cells from myocarditis hearts by using stainless-steel sieves at 4 -C is thought not to damage cells in contrast to purification by centrifuge in Percoll [36]. Actually, the gene expression of cardiac myosin isoform and ANP [5], which is found only in cardiomyocytes, was the same in purified cardiomyocytes as in unpurifed whole hearts. Therefore we assume that the gene expression in purified cell fractions is practically unchanged by these preparations.
It was reported that immunological factors might be of greater prognostic importance than the more conventional assessments of the hemodynamic and clinical status [7]. It is clinically significant to define the mechanisms by which these factors influence cardiac function and remodeling. Our study indicated that inflammatory cells and non-inflammatory cells influenced each other. It is to be hoped that a clarification of the mechanism lead to therapies for myocarditis and DCM by regulation of immunological factors.
We present here a possible crosstalk by immunological molecules among constitutive cells in EAM hearts. They play important roles not only among inflammatory cells but also among non-inflammatory cells containing cardiomyocytes, fibroblasts, endothelial cells and smooth muscle cells.