Correspondence to: Dr. Sawa KOSTIN, M.D., Ph.D., Max-Planck-Institute for Heart and Lung Research (W.C. Kerckhoff Institute), Parkstrasse 1, 61231 Bad Nauheim, Germany. Tel.: +49 6032 705 420; Fax: +49 6032 705 458 E-mail: email@example.comL.M. POPESCU, M.D., Ph.D., Department of Cellular and Molecular Medicine, “Carol Davila” University of Medicine and Pharmacy, P.O. Box 35–29, Bucharest 35, Romania. E-mail: LMP@jcmm.org
The existence of a novel type of interstitial cells in the heart, interstitial Cajal-like cells (ICLCs), had been described for the first time in 2005. Their identification was mainly based on ultrastructural criteria: very long (tens up to hundreds of micrometres) and moniliform prolongations, which are extremely thin (less than 0.2 μm), below the resolving power of light microscopy. Myocardial ICLCs were also identified by methylene-blue vital staining, silver impregnation, and immunoreactivity for CD 34, vimentin, CD117/c-kit, etc. Although a series of studies provided evidence for the existence of ICLCs in human atria and rat ventricles, further investigations in other laboratories, using additional techniques, are required to substantiate the consistency of these findings. Here we provide further evidence for the existence of ICLCs in human and mammalian hearts (by transmission and scanning electron microscopy, as well as confocal laser scanning microscopy). Noteworthy, we confirm that ICLCs communicate with neighbouring cells via shedding (micro)vesicles. Although these so-called ICLCs represent a distinct type of cells, different from classical interstitial cells of Cajal, or fibroblasts, their role(s) in myocardium remain(s) to be established. Several hypotheses are proposed: (i) adult stromal (mesenchymal) stem cells, which might participate in cardiac repair/remodelling; (ii) intercellular signalling (e.g. via shedding microvesicles); (iii) chemo-mechanical transducers and (iv) players in pacemaking and/or arrhytmogenesis, and so on.
Interstitial cells of Cajal (ICC) have extensively been characterized in the gastrointestinal tract (for reviews, see [1–6]). However, recently questions were raised about the presence of ICC outside the musculature of the gastrointestinal tract , and such cells were named interstitial Cajal-like cells (ICLCs) [8–11]. ICLCs have been identified in numerous organs: pancreas , gallbladder [13–15], urinary tract [16–22], myometrium [11, 23–28], fallopian tube [11, 29, 30], mammary gland , placenta , prostate  and mesentery , including the vasculature [34–38]. Noteworthy, very recently Faussone-Pellegrini and co-workers , based on immunoelectron microscopy, identified ICLCs in the gastrointestinal muscle of human beings, beyond the ‘classical’ ICC.
The very first demonstration of ICLC presence in the heart had been reported by Hinescu and Popescu in 2005 . Despite a series of papers that followed [10, 40–43] showing the existence of ICLCs in various parts of the heart, more studies are certainly needed, especially using other techniques to substantiate independently the consistency of these findings. Therefore, the present study aims to investigate the nanometre details of ICLCs in human and rat myocardium using transmission (TEM) and scanning electron microscopy (SEM) as well as confocal laser scanning microscopy (CLSM).
Material and methods
Adult rat hearts were obtained from five Wistar rats (90 days old). The animals were deeply anesthetized with pentobarbital (20 mg/kg body weight). The abdominal aorta was cannulated and the right atrium was opened. After 2 min. of retrograde myocardial perfusion through the abdominal catheter (80 mmHg perfusion pressure) with oxygen-saturated Tyrode's solution, hearts were perfused for 15 min. with 2% glutaraldehyde in 0.1 mmol natrium cacodylate for TEM as described, or with 2% paraformaldehyde in 0.1 mmol/l phosphate-buffered saline (PBS) for immunohistochemistry. The institutional Ethical Committee approved the study.
We have investigated human myocardium from three donor hearts that for technical reasons were not used for heart transplantation. Specimens were either fixed in 4% freshly prepared paraformaldehyde, cryoprotected with 20% sucrose, quick-frozen in methylbutane at -130°C and stored at −80°C, or were fixed in 3% glutaraldehyde. The institutional Ethical Committee approved the study.
Transmission electron microscopy
Specimens pre-fixed in glutaraldehyde were post-fixed in 2% osmium tetroxide for 1 hr. After dehydration in graded concentrations of alcohol and propylene oxide, they were embedded in Epon, following routine procedures. Semithin (1 μm) sections were stained with toluidine blue and viewed in a Leica DM microscope (Wetzlar, Germany). Ultrathin sections were stained with uranyl acetate and lead citrate, and viewed and photographically recorded under a Philips CM 10 electron microscope (Eindhoven, The Netherlands) as described.
Scanning electron microscopy
Glutaraldehyde pre-fixed specimens were washed in 0.1 mmol natrium cacodylate buffer (pH 7.4) and immersed in 6 N NaOH for 10–15 min. at 60°C to digest collagen fibres and basement membrane as previously described . The specimens were washed thoroughly in PBS and post-fixed in 1% osmium tetroxide for 1 hr. They were finally dehydrated in graded concentrations of alcohol, dried by the tbutylalcohol freeze drying method, coated with gold using an ioncoater and examined in a Hitachi S-800 electron microscope as described previously .
Immunohistochemistry and confocal laser scanning microscopy
Frozen tissue cryosections (10 μm thick) were placed on gelatin-coated slides and incubated for 15 min. with 100 mmol glycine and 0.1% carboxy-lated bovine serum albumin (Aurion, Wageningen, The Netherlands) in PBS, pH 7.4. Then the cryosections were incubated overnight with the primary antibody against vimentin (clone V9; Sigma, St Louis, MO, USA) in a moist chamber. After washing in PBS, the preparations were incubated with biotinylated donkey antimouse IgG followed by streptavidin-Cy-2 (Biotrend Cologne, Germany). The nuclei were stained with 0.002% 7-aminoactino-mycin D (Molecular Probes, Carlsbad, CA, USA). Omission of the primary antibody served as a negative control.
The immunolabelled sections were examined with a Leica TCN-NT laser microscope, equipped with argon/krypton and helium/neon laser. Series of confocal sections were taken through the depth of the tissue samples at 0.5-μm intervals. In order to improve image quality and to obtain a high signal/noise ratio each image from the series was signal averaged. After data acquisition, the images were transferred to a Silicon Graphics Indy or Octane workstations (Silicon Graphics, Sunnyvale, CA, USA) for image restoration and reconstruction using Imaris, the multi-channel image processing software (Bitplane, Zürich, Switzerland). The principles of this method and numerous images obtained with this technique have been previously published [46, 47]. In double-labelling experiments, cryosections were incubated overnight with the primary polyclonal antibody against c-kit (Dako, Glostrup, Denmark) followed by donkey anti-rabbit IgG coupled with Cy2 (Dianova, Hamburg, Germany). After repeated washes in PBS, the preparations were incubated with antivimentin antibodies directly coupled with Cy3. Myofibrils were stained for F-actin with Alexa 633-conjugated phalloidin. The nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) (Molecular Probes). All preparations were mounted in Mowiol (Hoechst, Frankfurt, Germany). Classical immunocytochemistry was performed on fomalin-fixed paraffin-embedded tissue sections by the avidin-biotin peroxidase complex method, as previously reported . The primary antibodies used were: CD34 – monoclonal, 1: 100, clone QBEND 10 (Biogenex, San Ramon, CA, USA) and vimentin, monoclonal, 1: 100, clone V-9 (Biogenex).
Figure 1 shows the general aspect (‘silhouette’) of a typical fusiform cell body with two very long (tens of micrometers) cyto-plasmic prolongations placed between cardiomyocytes. Note the characteristic appearance of ICLC processes: unequal thickness, sudden narrowing after coming off the cell body and dilated portions alternating with slender segments. Thus, the moniliform aspect of ICLC processes is evident. The dilated portions contain usually: caveolae, elements of endoplasmic reticulum and/or mitochondria. Fragments of tortuous ICLC processes (intersected by the plane of the ultrathin section) appear in Fig. 1, right upper part.
Higher magnifications (Fig. 2) provide additional evidence, with knobs along ICLC processes. Noteworthy, ICLC processes are located in the vicinity of nerve fibres, immunocompetent cells, and fibroblasts, eventually.
Figure 3 shows some morphological characteristics details of ICLC processes. It is obvious (Fig. 3A) that the thickness of the three ICLC processes (a–c) is far below the resolving power of light microscope (0.2 μm) and therefore, such ‘structures’ of 40–50 nm diameter are visible only by electron microscopy. This may explain the fact that, in order to be constantly observed, ICLC processes had to “wait” for the technology of high-resolution TEM. It is to be mentioned that Fig. 3 presents labyrinthic network of ICLC processes. Typically, ICLC processes show caveolae, short membrane associated dense bands, and relatively numerous close appositions.
Figures 4 and 5 present the location of ICLC convoluted processes between cardiac miocytes and blood capillaries. Moreover, Fig. 5A shows clearly that the tentacular ICLC processes may surround (completely) other cardiac myocytes, or blood capillaries, or both. Hereby, ICLC processes suggest the existence of a bi(three)-dimensional network. This impression is strengthened by the scanning electron micrograph in Fig. 6. The three-dimensional image shows, unequivocally, the silhouette of a very long, slender bipolar ICLCs, which has close contacts with at least two blood capillaries and at least four ICLCs. Compare the three dimensional of Fig. 6 with the bi-dimensional view of Fig. 5. Another ICLC (Fig. 6), upper left corner, is located parallel to car-diomiocytes and blood vessels.
Currently, there is not generally accepted panel of antibodies for ICLC immunophenotyping. However, customary markers include CD117/c-kit, CD34, vimentin, α smooth muscle actin, etc. We previously reported that polyclonal antibodies used against CD117/c-kit resulted in a weakly and inconstantly positive immunostaining of ICLCs [10, 40]. Thus, we tested the presence of CD34 and vimentin antigens. The images of Fig. 7A and B suggest that, from the point of view of usual light-microscope immunohistochemistry, most of CD34+ cells and vimentin positive cells might be superposed. A network-like distribution of such ICLCs seems in accordance with TEM data. Figure 7C obtained by CLSM reveals a vimentin positive ICLCs. Vimentin localization is predominant along the processes, as previously reported in cell culture .
Figure 8 shows the images obtained by CLSM concerning the existence of c-kit+ (Fig. 8A) or vimentin (Fig. 8B) cells in human left ventricle myocardium. However, Fig. 8C indicates that only a few cells (three to four) appear positive for both c-kit and vimentin, concomitantly. This confirms again our previous report about c-kit immunopositivity in rat ventricle .
Figure 9 shows a typical bipolar ICLCs in the narrow space between three cardyomyocites. At least 24 shedding microvesicles (red arrows) budding from the plasma membrane of ICLCs are clearly visible. Such vesicles of about 200–400 nm were considered for a long time to be artefacts, but now are recognized as specific lepto-structures that are distinct from the exosomes released upon exocy-tosis of multivesicular bodies [48–50]. Exosomes have diameters of only 30–100 nm . Anyway, shedding microvesicles like those shown in Fig. 9 are no longer artefacts or ‘little more than garbage bags’, containing materials that cell needed to get rid of [48, 50].
Shedding vesicles could participate in important biological processes and their significant number at the surface of an ICLC in situ(Fig. 9) suggests that these lepto-structures may transport, as cargo, important information to other cells over nano(micro) distances.
Table 1 shortens the ‘Discussion’ and clearly shows that the ICLCs of human or mammalian myocardium represent a distinct cell type, different from classical (intestinal) ICC or fibroblasts. Moreover, a comparison of ICLCs with myofibroblasts is not necessary since myofibroblasts do not exist in healthy tissues .
Table 1. Comparative characteristics of ICC, ICLCs and fibroblasts
*Fibroblasts in situhave no caveolae [10, 22, 39, 40, 55]. Although several authors claimed that fibroblasts have caveolae, in fact caveolae could be found in human or mammalian fibroblasts, but only in vitro, in cultured cells, when the phenotype is changed.
It should be noted that, at present, although the ICLC existence as a distinct cell population is unequivocal, the (exact) functions of ICLCs in myocardium are unknown. Therefore, it seems attractive to enumerate several possibilities:
1ICLCs are involved in intercellular signalling since they have a strategic position, very closed to cardiomyocytes, blood capillaries and nerve endings and, in addition, ICLCs form a 3D-network (Fig. 5). At least two mechanisms could be considered: (i) a paracrine and/or juxtacrine secretion of small signal molecules and (ii) shedding microvesicles (see Fig. 9), which play unique roles in the ‘horizontal’ transfer of important macromolecules (e.g. proteins or RNAs) among neighbouring cells, necessary for the rapid phenotype adjustments in a variety of conditions .
2ICLCs seem to be mechanoreceptors/transducers as previously discussed [40, 42]. This is mainly suggested by their extremely long cytoplasmic processes (tens up to hundreds micrometers; see Figs. 1, 3, 6 and 9). It is noteworthy, considering the microscopic anatomy or histology, that only some specialized nerve cells have cell processes longer than ICLCs (the length does matter!).
3ICLCs could participate in pacemaking and/or arrhythmo-genesis. For instance, very recently we described a significant population of ICLCs in myocardial sleeves of pulmonary veins  and another group found c-kit+ cells in similar locations . Although, the mechanisms of atrial fibrilation are still the subject of controversies , many authors favour ‘the pulmonary vein wave’ hypothesis [59–63]. Indeed, Anderson , wrote: ‘Precise knowledge of the source of the abnormal rhythms originating within the pulmonary veins would contribute immensely to clarifying our understanding of this vexatious arrhythmia’.
4In myocardium, ICLCs might be considered as adult stro-mal mesenchymal stem cells. ICLCs might also be viewed as an unexplored population of uncommitted c-kit+ resident cells . Undeniably, the identification of cardiac progenitor/stem cells (sometimes c-kit+) is currently a hot subject [66–72]. If ICLCs are progenitor/stem cells in myocardium, then the variable c-kit posi-tivity (Table 1) may reflect different functional/pathological circumstances. Naturally, ICLCs could be a therapeutic target for regenerative cardiovascular medicine [65, 73–75].
In conclusion, instead of believing that fibroblasts are perhaps the most underestimated cell population in the heart , we suggest that ICLCs are really neglected, although they might be key players in understanding cardiac (ultra-)structure–function relationships.