SEARCH

SEARCH BY CITATION

Keywords:

  • bladder ultrastructure;
  • myofibroblast;
  • stretch;
  • innervation

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

OBJECTIVES

To describe the ultrastructure and relationship to nerves of the myofibroblast in the human bladder lamina propria, and discuss its possible role in bladder function, including sensing stretch, as the response of the bladder to stretch has been thoroughly investigated by afferent nerve recordings, but specialized stretch sensing organs have yet to be identified.

MATERIALS AND METHODS

Flexible cystoscopic bladder biopsies were obtained from patients with detrusor hyper-reflexia and from controls. Systematic electron micrographs were obtained throughout the lamina propria, and the presence and location of cells with ultrastructural characteristics of myofibroblasts noted, together with their relation to surrounding nerves.

RESULTS

Within the lamina propria there was a layer of cells with the cytological characteristics of both fibroblasts and smooth muscle cells, that included bundles of fine cytoplasmic filaments, dense bodies, linear arrays of subsurface vacuoles, and the presence of an interrupted basal lamina. This combination of features is characteristic of the myofibroblast. These cells had close contacts with unmyelinated axonal varicosities containing a mixture of clear and large dense-cored vesicles, or clear vesicles alone.

CONCLUSIONS

There is a layer of cells with the ultrastructural characteristics of myofibroblasts within the human bladder lamina propria. Their close contacts with nerves containing both small clear, and small clear with dense-cored, vesicles implies they have both an efferent and an afferent nerve supply, possibly functioning as a bladder stretch receptor. Furthermore, because of their similarities with the interstitial cells of Cajal in the gut, which are claimed to modulate small intestinal function, we discuss other possible roles for bladder lamina propria myofibroblasts.

The basic science section covers not only molecular biology and urological oncology, but also urological physiology and experimental studies into the effect of treatment (be it pharmacological or technological) on the function of the urinary tract. There are several important papers in this issue relating to tumour markers, and to understanding the mechanism of disease in urothelial cancer and renal cancer.

The effect of stretch on bladder function has inspired a considerable amount of experimental work and has shown that not only is the phenotype of the muscle cells changed, but that also, as is shown in this section, there may well be specialised stretch-sensing cells in the bladder.

The use of technology to treat prostatic conditions has become very popular, although several technologies move in and out of favour. One such is the use of ethanol and chemoablation of the prostate. This is re-evaluated here and some of its short comings indicated. It is encouraging that there is still place for the critical evaluation of a technique to be by an experimental study.

INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

The response of the bladder to stretch has been thoroughly investigated [1]. Numerous unmyelinated, free nerves have been identified in the lamina propria of the bladder [2,3] and it has been suggested that such axons could respond to mechanical stimuli, or that there are afferent volume receptors close to them. Indeed, there is increasing evidence that such afferent nerve endings in the bladder wall are important in regulating bladder function, and they have been shown to be activated during stretch of the bladder wall in rats [4] and cats [5]. Neurotransmitters and the chemical composition of bladder contents might modify the behaviour of these nerves [4]. While no specialized stretch-sensing organs have been reported, these afferent nerves are likely to be important.

The structure of the human bladder lamina propria is well documented; it consists of loose fibroelastic connective tissue containing axons with their varicosities, fibroblasts, blood capillaries, and smooth muscle cells. The smooth muscle cells have been reported to form a distinct layer termed, by analogy with the structure in the intestine, the ‘muscularis mucosae’ although it was noted to be discontinuous and less well defined than in the gut [6]. This description has been adopted by urological texts [7].

However, recently the existence of another type of cell in the bladder lamina propria, the myofibroblast, has been identified [8]. This cell has some of the structural characteristics of both a smooth muscle cell and a fibroblast. Reportedly it occurs in several normal tissues, including stomach, testicle, prostate, liver, spleen, kidney and lung [9,10]. Furthermore, it is found and implicated in the development of a variety of pathological conditions, many of which result from an overactivity of myofibroblasts. These include burn contractures, lung fibrosis, liver cirrhosis and atherosclerotic plaques. These cells have been described in relation to the fibromatoses, and are possibly responsible for the desmoplastic reaction seen in many cancers, e.g. breast carcinoma and malignant melanoma [9,10].

The myofibroblast is also important in the inflammatory response, being a prolific producer of chemokines and cytokines [10], and has a central role in wound healing, a function which is facilitated because the myofibroblast may be contractile, aiding in reducing the amount of denuded surface area of wounded tissue. While several different phenotypes have been described, many contain smooth muscle myosin isoforms in addition to α-smooth muscle actin necessary for their contractile properties [10].

Morphologically similar cells have been described in the rabbit bladder serosa [11] after mechanical stress induced by partial outlet obstruction, and throughout the detrusor muscle, interstitium and outer fibromuscular coat of the human and guinea-pig bladder [12]. Recently descriptions have been published of similar cells, also referred to as ‘interstitial cells’, in the human bladder lamina propria [8,13]. Elsewhere in the urinary tract, such cells have been described in the guinea-pig upper urinary tract [14], where they have been attributed a role in conducting and amplifying pacemaker signals to initiate action potential discharge, and the rabbit lower urinary tract [15], where they are specialized pacemaking cells.

Herein we report the existence of a myofibroblast layer in human bladder lamina propria from healthy subjects and those with detrusor hyper-reflexia, describe the relationship of these cells with surrounding nerves, and discuss their possible role in bladder function.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

Bladder biopsies were obtained from> 40 patients with detrusor hyper-reflexia and eight with microscopic haematuria. All parts of the study were approved by the National Hospital for Neurology and Neurosurgery and The Institute of Neurology, London, Joint Medical Research Ethics Committee, and were carried out with the patients’ consent.

The biopsies were obtained and processed as previously described [3]. Systematic transmission electron micrographs were obtained throughout the lamina propria of these specimens, and the presence of cells with the ultrastructural characteristics of a myofibroblast and their relationship to surrounding nerve fibres recorded.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

In the superficial zone of the lamina propria there were several layers of flattened cells, orientated parallel to the urothelial basal lamina and separated one from another by bundles of collagen fibres. The cells of this superficial layer had the cytological characteristics of fibroblasts, but many of those more deeply placed within the lamina propria also showed features of smooth muscle cells, including bundles of fine cytoplasmic filaments associated with dense bodies, linear arrays of subsurface vacuoles, and a discontinuous basal lamina; all are characteristic features of myofibroblasts (Fig. 1a–d). They also had prominent rough endoplasmic reticulum and Golgi apparatus, notched nuclei, appeared to attach to one another at their margins, were generally large with extensive flattened cytoplasmic extensions (Fig. 1c) and were often associated with extracellular bundles of fine filaments (microtendons). Close contacts between these cells and unmyelinated axonal varicosities containing a mixture of both clear and large dense-cored vesicles (Fig. 1a), or clear vesicles alone (Fig. 1d) were identified. The clear vesicles had a median (range) diameter of 54 (53–55) nm, and the dense-cored vesicles were larger, with a median diameter of 98 (65–135) nm. Those nerve varicosities which contained only clear vesicles often had small sub-axolemmal densities, indicating possible vesicle release sites (Fig. 1d). No differences were apparent between the morphology of the myofibroblasts or their relationships with nerves in biopsies from patients with detrusor hyper-reflexia when compared with those from ‘control’ bladders. Finally, no layers of definitive smooth muscle cells that could represent a muscularis mucosa were identified, the myofibroblast layer described lying in the region in which a muscularis mucosae has been previously reported [6].

image

Figure 1. a, A myofibroblast, identified by subsurface caveolae (C), fine filaments in cross section (F) and interrupted basal lamina (BL). The surface is irregular and the cell closely associated with an axonal varicosity (A), which contains both large dense-cored and small clear vesicles. b, A comparison of the myofibroblast (M), with fine filament bundles (F), dense bodies (D), interrupted basal lamina (BL) and subsurface caveolae (C) and the fibroblast (FB), with rough endoplasmic reticulum (ER), but no basal lamina or subsurface caveolae. c, A myofibroblast, showing several lateral extensions (L), dense bodies (D) and attachment plaques (P). d, A myofibroblast with a close relationship to two nerve varicosities (V) which contain only clear vesicles, one of which has evidence of a release site (R). Another axon (A) is seen close by.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

The ultrastructural characteristics required for a cell to be identified as a myofibroblast have been clearly defined [9]. Ultrastructurally it has bundles of cytoplasmic ‘microfilaments’ with dense bodies running parallel to the long axis, a well developed rough endoplasmic reticulum and Golgi apparatus, a notched nucleus, ‘pinocytotic vesicles’, partial investment by basal lamina with points of plasmalemmal attachment, well-developed microtendons, and intercellular intermediate and gap junctions. The cells described here fulfil many of these defining criteria. Sui et al.[8] identified gap junctions between similar cells in the human bladder, but it was not possible to confirm this in the present material. The term ‘pinocytotic vesicle’ ascribes an unverifiable function to the same structural feature identified here, and thus we referred to them as subsurface caveolae, because of their close resemblance to similar structures found in definitive smooth muscle cells. Furthermore, we prefer to define the filamentous content of the cytoplasm as bundles of fine filaments rather than as ‘microfilaments’, which has a more restrictive connotation.

Whether these cells represent the same cells described by Dixon and Gosling [6] as smooth muscle cells is not clear, but while there were clear differences between the myofibroblasts described here and smooth muscle cells, it is possible that we examined cells from a different subset of the population spectrum. However, in the present study we failed to identify definitive smooth muscle cells in this layer of the bladder lining, and therefore consider it unlikely that a ‘muscularis mucosa’ exists in the human bladder.

The present study describes only the ultrastructural features of the myofibroblasts and their other phenotypic characteristics; their ability to stain for vimentin, desmin, α-smooth muscle actin and myosin isoforms needs to be investigated further, so that these cells can be more precisely classified [10].

Dixon and Gilpin [2] described the presence of vesicle-containing axonal varicosities in the lamina propria, and stated that as there were no recognisable neuro-effector sites in this region, such nerves could represent the terminations of sensory neurones. They described the occasional relationship of such axons to connective tissue cells, which lacked an investing basal lamina. While insufficient morphological detail of these cells was given to determine if they were myofibroblast-like cells, there was a close relationship between nerve fibres and the myofibroblasts described here, suggesting a functional relationship.

These associations are with nerve varicosities containing both clear and large dense-cored vesicles, and clear vesicles alone. Clear vesicles of a similar mean diameter are presumed to contain acetylcholine in cholinergic nerves [16,17] but may also occur in somatic afferent nerve terminals. The association of small clear vesicles with larger dense-cored vesicles is well described in cholinergic nerves [18] and their terminals have been reported to release acetylcholine and ATP or a related purine [19,20]. While we report two different populations of axonal varicosity, i.e. those containing small clear vesicles alone and those containing small clear vesicles associated with large dense-cored vesicles, it is possible that in the former group large dense-cored vesicles were present but not visible in some specimens because such vesicles are less numerous than the small clear vesicles, and may have not been included in the plane of section. Alternatively, it is also possible that there are two distinct populations of nerves, efferent and afferent, a suggestion supported by reports that when compared with nerve terminals among the detrusor smooth muscle cells, the suburothelial axons contain a significantly higher proportion of large granulated (dense-cored) vesicles. This may reflect their known diverse content of neuropeptides, including substance P and calcitonin gene-related peptide, and a possible sensory function [21].

The lamina propria nerves have been implicated in the response of the bladder to stretch [1,4], and given their often close association to the myofibroblasts, it is possible that they could function together in this role. The axon terminal could act as a mechanoreceptor through their close contact with the myofibroblasts, responding to a mechanical stimulus of stretch or contraction of the myofibroblasts. Furthermore, the proximity described between myofibroblasts and nerve varicosities containing only agranular vesicles suggests they could provide an efferent supply to the myofibroblasts. Therefore it is possible that the myofibroblasts and their attached and closely associated axonal varicosities could collectively function as a bladder stretch receptor, in a manner analogous to the muscle spindle of voluntary muscle, in which the sensory axon terminals are closely applied to the intrafusal myofibres and discharge in response to a change in their dimensions [22]. If the planar dimensions of the myofibroblasts were under autonomic control by neighbouring efferent axons, a mechanism for tuning the sensitivity of the bladder to applied stretch can be proposed, again by analogy with the muscle spindle, in which α-efferent control of the intrafusal fibres modulate their mechanical properties and thus the sensitivity of their central sensory innervated region to stretch. The proposed cholinergic innervation could explain why anticholinergic medication increases bladder capacity and reduces the sensations of urgency, an observation that it is difficult to account for solely by the action of anticholinergics on the parasympathetic innervation of the detrusor muscle.

Whatever the merits of this particular hypothesis, the intimate relations of the myofibroblasts to axons which are presumed to be afferent implies that there is some functional interplay between them. Further studies will be needed to determine its precise nature. Other possibilities include that the release of neuropeptides from afferent terminals could affect the function of myofibroblasts, or that interaction occurs in both directions.

Clues to the possible role of the myofibroblast in bladder function can be gained from assessing their function elsewhere; the interstitial cells of Cajal (ICC), myofibroblast-like cells located in the gastrointestinal tract, have been studied extensively. While some have suggested that ICC could act as mechanoreceptors, as we suggest here for myofibroblasts in the bladder lamina propria, this hypothesis has received little attention [23]. However, ICC have been implicated as electrical pacemakers and conductors of electrical activity in the gastrointestinal tract [23], a role which has also been attributed to myofibroblasts in the upper [14] and lower [15] urinary tract. A recent study reported that there are gap junctions between myofibroblasts in the lamina propria of the human bladder, by ultrastructural studies and staining for connexin43. This finding implies that such cells may have the capacity to form a functional syncytium in the suburothelial space [8], which, facilitated by their extensive processes, would be able rapidly to spread electrical, and possibly contractile, activity. While no contacts with detrusor muscle have been described previously [8] or in the present study, others suggested that myofibroblasts in the gastrointestinal tract and urethra are linked functionally to the smooth muscle cells of the wall of the viscus [15,24].

As well as responding to efferent neural activity, the myofibroblasts may be responsive to humoral influences from the epithelium. For example, they could have a role in the relay of signals released from the urothelium during bladder filling, e.g. ATP [8], which is released from rabbit urothelium in response to mechanical stimulation [25], might bind to P2X3 receptors expressed by sensory neurones [26], and can activate pelvic afferents in the urinary bladder [27]. The P2X3 receptor is thus involved in afferent pathways that control urinary bladder volume reflexes, and the myofibroblasts could be central to this, either being directly responsive to ATP, or with ATP modifying the response of the lamina propria afferents to myofibroblast activity. Similar pathways might be involved in the reaction of the bladder to intravesical vanilloids, which are an effective therapy in patients with detrusor hyper-reflexia [28], and act via the VR-1 receptor on afferent lamina propria nerves [26]. Recent work has also suggested that this receptor might be present on the surface of incompletely characterized ‘interstitial cells’ in the human bladder lamina propria [13], which could therefore respond directly to vanilloids.

Closer examination of ICC suggests an analogous role for myofibroblasts in modulating neurotransmission in the bladder, as they have been implicated in so doing in the gastrointestinal tract [23]. Morphological observations have shown by electron microscopy that spacing between ICC and varicose nerve terminal is as little as 20 nm. Inhibitory innervation within the canine colon originates in regions heavily populated with ICC, and others have shown that inhibitory motor neurones containing vasoactive intestinal peptide are closely associated with ICC in the submucosal pacemaker region. As noted above, ICC have been shown to be responsive to several enteric neurotransmitters including acetylcholine, vasoactive intestinal peptide, ATP, nitric oxide and substance P [23,29]. Further physiological evidence is that nitric oxide and, recently, carbon monoxide have been proposed as inhibitory neurotransmitters generated by the ICC and other myofibroblasts [29]. In the guinea pig bladder, Smet et al.[12] suggested that interstitial cells may mediate the effects of neuronally released nitric oxide, adding further credence to their possible role in neurotransmission.

There is thus ample evidence that myofibroblasts are important in smooth muscle behaviour, neurotransmission and regulation of normal function in the gastrointestinal tract, and the hypothesis that they could have similar functions in the bladder should be tested. If such cells are central to normal bladder function, their abnormal activity could be important in disease processes. When neutralising antibodies to c-kit, a protein located on ICC, are injected in neonatal animals for several days after birth, intestinal c-kit immunoreactivity disappears and intestinal smooth muscle activity becomes abnormal [29]. Furthermore, mutants of c-kit expression have abnormalities of intestinal smooth muscle function. Several well described human gastrointestinal diseases are associated with ICC abnormalities, including ulcerative colitis, intestinal pseudo-obstruction, Hirschprung's disease, congenital megacolon of piebaldism, and infantile hypertrophic pyloric stenosis. By extrapolation, it is possible that abnormalities of myofibroblast function in the bladder lamina propria could be associated with abnormal bladder function, e.g. as is seen in detrusor instability, detrusor hyper-reflexia, interstitial cystitis and unexplained urinary retention, even though we detected no morphological differences between myofibroblasts in the present patients with detrusor hyper-reflexia and those in control patients. Further studies of myofibroblast organization, structure and function within the bladder lamina propria are clearly needed.

ACKNOWLEDGEMENTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

We acknowledge with gratitude the provision of the specimens used for this study by Mr P. Dasgupta, Mr I. Hussain and Mr C. Brady, taken as part of a larger collaborative study of the innervation of the human lamina propria, and thank Mr B.C. Young and Ms K. Riley for their technical assistance. The authors also thank the trustees of St. Peter's Trust for Kidney, Bladder & Prostate Research for their financial support.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES
  • 1
    Morrison J, Steers WD, Brading A et al. Neurophysiology and neuropharmacology. In AbramsP, CardozoL, KhouryS, WeinA eds, Incontinence. Chapt 2. Plymouth: Plymbridge, 2002
  • 2
    Dixon JS, Gilpin CJ. Presumptive sensory axons of the human urinary bladder: a fine structural study. J Anat 1987; 151: 199207
  • 3
    Wiseman OJ, Brady CM, Hussain IF et al. The ultrastructure of bladder lamina propria nerves in healthy subjects and in patients with detrusor hyperreflexia. J Urol 2002; 168: 20405
  • 4
    Morrison J, Wen J, Kibble A. Activation of pelvic afferent nerves from the rat bladder during filling. Scand J Urol Nephrol 1999; 201: 735
  • 5
    Downie JW, Armour JA. Mechanoreceptor afferent activity compared with receptor field dimensions and pressure changes in feline urinary bladder. Can J Physiol Pharmacol 1992; 70: 145767
  • 6
    Dixon JS, Gosling JA. Histology and fine structure of the muscularis mucosae of the human urinary bladder. J Anat 1983; 136: 26571
  • 7
    Brookes JD. Anatomy of the lower urinary tract and male genitalia. In WalshP, RetikAB, VaughanEDJr, WeinAJ eds, Campbell's Urology, 8th edn, Vol. 1. Chapter 1. Philadelphia: Saunders, 2002: 59
  • 8
    Sui GP, Rothery S, Dupont E, Fry CH, Severs NJ. Gap junctions and connexin expression in human suburothelial interstitial cells. BJU Int 2002; 90: 118 29
  • 9
    Schurch W, Seemayer TA, Gabbiani G. The myofibroblasts. a quarter of a century after its discovery. Am J Surg Pathol 1998; 22: 1417
  • 10
    Powell DW, Mifflin RC, Valentich JD, Crowe SE, Saada JI, West AB. Myofibroblasts. 1. Paracrine cells important in health and disease. Am J Physiol 1999; 277: C1C19
  • 11
    Pampinella F, Roelofs M, Castellucci E, Passerini-Glazel G, Pagano F, Sartore S. Time-dependent remodeling of the bladder wall in growing rabbits after partial outlet obstruction. J Urol 1997; 157: 67783
  • 12
    Smet PJ, Jonavicius J, Marshall VR, DeVente J. Distribution of nitric oxide synthase immunoreactive nerves and identification of the cellular targets of nitric oxide in guinea-pig and human urinary bladder by cGMP immunohistochemistry. Neuroscience 1996; 71: 33748
  • 13
    Ost D, Roskams T, Van Der Aa F, De Ridder D. Topography of the vanilloid receptor in the human bladder: more than just the nerve fibers. J Urol 2002; 168: 2937
  • 14
    Klemm MF, Exintaris B, Lang RJ. Identification of the cells underlying pacemaker activity in the guinea-pig upper urinary tract. J Physiol 1999; 519: 86784
  • 15
    Sergeant GP, Hollywood MA, McCloskey KD, Thornbury KD, McHale NG. Specialised pacemaking cells in the rabbit urethra. J Physiol 2000; 256: 35966
  • 16
    Burnstock G. The changing face of autonomic neurotransmission. Acta Physiol Scand 1986; 126: 6791
  • 17
    Parsons SM, Prior C, Marshall IG. Acetylcholine transport, storage, and release. Int Rev Neurobiol 1993; 35: 279390
  • 18
    Gabella G. Structure of the intramural nerves of the rat bladder. J Neurocytol 1999; 28: 61537
  • 19
    Gabella G. The structural relations between nerve fibres and muscle cells in the urinary bladder of the rat. J Neurocytol 1995; 24: 15987
  • 20
    Hoyle CH, Burnstock G. Postganglionic efferent transmission in the bladder and urethra. In MaggiCA ed. Nervous Control of the Urogenital System. Chur, Switzerland: Harwood Academic Publishers, 1993: 34981
  • 21
    Brading A. Physiology of bladder smooth muscle. In TorrensM, MorrisonJFB eds, The Physiology of the Lower Urinary Tract. London: Springer Verlag, 1987: 16192
  • 22
    Landon DN. Skeletal muscle – normal morphology, development and innervation. In MastagliaFL, WaltonJN eds, Skeletal Muscle Pathology. 2nd edn. Chapter 1. London: Churchill Livingstone, 1992
  • 23
    Sanders KM. A case for interstitial cells of Cajal as pacemakers and mediators of neurotransmission in the gastrointestinal tract. Gastroenterology 1996; 111: 492515
  • 24
    Edwards FR, Hirst GD, Suzuki H. Unitary nature of regenerative potentials recorded from circular smooth muscle of guinea-pig antrum. J Physiol 1999; 519: 23550
  • 25
    Ferguson DR, Kennedy I, Burton TJ. ATP is released from rabbit urinary bladder epithelial cells by hydrostatic pressure changes – a possible sensory mechanism? J Physiol 1997; 505: 50311
  • 26
    Yiangou Y, Facer P, Ford A et al. Capsaicin receptor VR1 and ATP-gated ion channel P2X3 in human urinary bladder. BJU Int 2001; 87: 7749
  • 27
    Namasivayam S, Eardley I, Morrison JFB. Purinergic sensory neurotransmission in the urinary bladder: an in vitro study in the rat. BJU Int 1999; 84: 85460
  • 28
    Fowler CJ, Beck RO, Gerrard S, Betts CD, Fowler CG. Intravesical capsaicin for treatment of detrusor hyperreflexia. J Neurol Neurosurg Psych 1994; 57: 16973
  • 29
    Powell DW, Mifflin RC, Valentich JD et al. Intestinal subepithelial myofibroblasts. Am J Physiol 1999; 277: C183F201
ABBREVIATIONS
ICC

interstitial cells of Cajal.