124
methods will eventually reveal whether it really is time to give up the classical methods that have stood us in such good stead for several decades SIDNEY
SkIUI.iMAN
References 1 Shulman, S. (1975) Re,jroduclzon arid Anlzbody Kr.~jm.\r, CRC Pi-w, Cleveland 2 Shulman, S. (1978) in Spermatozoa, ilnlzlrodze~ unrl Irzfirlzl;~ (Cohen, J. and Hendry, W. F., eds.) pp. 81-99. Blackwell Scientific Publications, Oxford 3 Riimke, P. and Hekman, A. (1975) Clzn. Hnriocnnol. M&b. 4, 473-496
4 Hekman, A. and Kiimke, P. (1976) in -Tpx&ok cd Imvw~opathology (Micscher, P. A. and Miiller-Eberhard, H. J., eds.) pp, 947-962, Grune and Stratton, New York 5 Shulman, S., Zappi, E., Ahmed, U. and Davis, J. E. (1972) htracu/aon
5, 269-278
6 Mathur, S., Williamson, H. O., Derrick, F. C., Madyastha, P. R., Melchers, J. T., Holtz, C. L., Baker, E. R., Smith, C. L. and Fudcnberg, H. [I. (1981)s. ~rnmunoi. 126, 905-909 7 Mathur, S., Williamson, H. O., Landgrebc, S., Smith, C. L. and Fudenberg, H. H. (1979) 3. Ivununoi. M~t/wd\ 30, 38 l-393 8 Shulman, S. (1980) in Mamtnl o/ Uinzcal Zmmunology (Rose, N. R. and Friedman, H., eds.) 2nd edn, pp. 907-916, Waverly Press, Baltimore 9 Ingerslev, H. J. (1979) M.3. krtzl. 24, 1-12 10 Ingerslev, H. J. and Hjort, T. (1979) I;rrlz/. SI&. 31, 496-502 11 Hamerlynck, J. and Riimke, P. (1968) ,7, &prod. EPrtzl. 17, 191-194
The immunobiology of Langerhans cells P. S. Friedmann Uepartmcnt
of Dermatology,
The Royal Victoria
Infirmary, The University Tyne, NE1 4LP, U.K.
Interest in the biology oj’ Langerhans cells has recently been .stimulatrd Friedmann, that they aw active members the immune .syrtem.
of Newcastle-Upon-Tyne, by ob.trrvations,
Newcastle-Uponrevxwed
here by Petu
?f
The cells discovered in 1868 by Paul Langerhansr have been overlooked and unstudied, probably because they were difficult to see with normal histological stains. Langerhans found that they could be stained with gold chloride and concluded they were therefore nerve cells. In 1875, Kanvier2 suggested that the cells were of lymphoid origin and over the years since then they have been regarded as epidermal cells, artefacts, melanocytes and neural elements such as Schwann cells. In 1961, Birbeck and co-workers’ described the specific organelle or granule by which Langerhans cells (LCs) are now identified by electron microscopy. Once LCs could be identified with certainty, the way was opened for progress in the analysis of their structure and function. Methods
of demonstration
and
identification
I.igh t micro.\ coj~y
Methods for the demonstration of LCs by light microscopy have improved. They can be stained with certain metals including gold, osmium, cobalt, lanthanum, mercury, nickel and chromium. They take up
supra-vital dyes such as methylene blue and brilliant such as cresyl blue, and a variety of substances formaldehyde, glutaraldehyde, ethylene-diamine and paraphenylene-diamine which can act as contact allergens and can be identified histochemically. This was the property that led Shelley and Juhlin to postulate that LCs formed a reticula-epithelial system involved in trapping external antigens4. LCs have an affinity for various catecholamines including dopamine, noradrenaline and L-DOPAS. After exposure of the skin to these substances u-z vitro, and subsequent exposure to formaldehyde vapour fluorescent products are formed in the cells. LCs may be stained by a number of histochemical methods. The most widely used involves the demonstration of formalin-resistant adenosine triphosphatase (A’l’Pasc) which is present in the outer membrane of the cell’l. In the epidermis, the stain is specific for these cells. Human LCs (cell body only) can be recognised by the presence of a-mannosidase, the guinea pig’s LCs contain aminopeptidasc and murine LCs acetyl esterase.
125
Immunol~~gicul murkers LCs possess immunological markers and receptors similar to those found on macrophages. Thus, they bind the Fc fragment of IgG’ as well as components C3b, C3d, C4b and C4d of complemenP. In addition, human and mouse LCs bear histocompatibility antigens of the D/DR” and I-A and I-E/Cl0 specificities, respectively. Moreover, they appear to be the only cells in the epidermis to produce such ‘Ia’ antigens. The presence of receptors has allowed the identification and manipulation of LCs. For example, LCs in dissociated cell suspensions may be selectively enriched by rosetting procedures and then used in assays of their functional capacities. Electron mzcroscopy The collective results of ultrastructural studies have revealed that LCs are dendritic cells with a lobulated nucleus and clear cytoplasm containing microtubules and microfilaments (concomitants of motility) but lacking tonofilaments and desmosomes (characteristic of epithelial cells). They contain specific LC granules, lysosomes and mitochondria. The
Langerhans
cell
Fig. 1. Location
of Langerhans
cells.
skin, there are density of 4-800/ mm2 Is. In guinea-pig 2 I6 and similar numbers have about 800-lOOO/mm been found in the skin of other rodents. Certain anatomical sites have reduced numbers and uneven distributions of LCs. They are absent from cornea” and are reduced to 130/mm* in, hamster cheek pouch”; mouse tail skin has a regular pattern of scales with LCs clustered at the interscale regions but deficient in the central scale zones. The average density is about 1 lo-260/mm2 and varies with strainr6.
granule
The specific LC granules described by Birbeck3 and visible only at the ultrastructural level are still regarded as the characteristic by which LCs are defined. In ultrathin sections, the granules are rod-shaped with a central dense striated band. They may have an expanded end, giving a ‘raquet-like’ appearance. Granules have been seen fused with the outer membrane of the cell and, when in that position, they may take up protein labels such as peroxidase”. The intracellular granules stain with the reaction product of osmiumzinc iodide whereas those which have made contact with the outer membrane do not. The granules have not been shown to possess enzymic activity and it is not yet established whether they are formed from the outer membrane or the Golgi apparatus. No functional activity has yet been ascribed to them. For a full review see Wolffrz. Distribution
LCs are found in mammals in most squamous epithelia including the skin, buccal mucosa, oesophagus, and vaginal and rectal mucosae. In skin, they occur mainly in the mid-epidermis (Fig. 1.) where they have been called Type 1 LCs on the basis of many dendrites, electron-dense cytoplasm and numerous Birbeck granulesr3. Type II LCs are found in the basal layer of epidermis and have few dendrites or Birbeck mitochondria and electron-dense granules, many cytoplasm. LCs or LC-like cells are also found in the dermis and in lymphoid organs including thymus, lymph nodes and spleen (for references see Ref. 14). In human epidermis they represent 3-8% of the total number of epidermal cells and are present at a
Origin
The tissue of origin of LCs has only been established recently. The chain of discovery began with the demonstration that they were not of neuraland others”,*O showed that crest origin. SilversI mouse skin deprived of its neural crest elements would still develop LCs but not Schwann cells and melanocytes. Evidence accumulated to indicate that they might be of mesenchymal origin’*; they are distributed in lymphoid tissues, they resemble the cells of histiocytosis X (which also contain Birbeck granuleszl) and hairy cell leukaemia and they have surface markers which are similar to those of the macrophage/monocyte lineage. Katz and co-workers22 first showed they were derived from a mobile pool by transplanting skin from parental mice to appropriate Fl hybrid recipients. Subsequently, the H-2 and Ia specificities of LCs in the grafts were determined at various times. It was observed that from 11 days after transplantation, there was a gradual increase in the proportion of LCs whose histocompatibility phenotype was that of the recipient. Keratinocytes, on the other hand, were always of donor phenotype. Those workers went on to show that LCs arrived from a central source, the bone marrow, rather than by local lateral migration from ad,jacent skin. They determined the histocompatibility phenotypes of LCs and epidermal keratinocytes in radiation chimaeras, i.e. irradiated mice reconstituted with allogeneic bone-marrow. By three days after reconstitution, a small percentage of LCs bearing Ia antigens of the marrow donor phenotype were detectablezi. By three months, 80% of LCs were of donor phenotype. Moreover, Fre-
126
linge? showed that epidermal la antigens were synthesised by bone-marrow-derived cells in the epidermis and were not simply passively adsorbed. ‘Thus, LCs are unequivocally derived from a central pool of precursors in bone marrow, from which they are continually renewed and replaced in the periphery. In normal marrow there are no cells containing LC granules. It is likely, therefore, that LC granules are only formed once the cell has taken up a position in which it is required to be functionally mature. It is interesting that some LCs appear able to synthesise DNA and divide once they have taken up residence in the periphery. Thus, Mackenziez5 showed that LCs in normal epidermis would incorporate ‘H-thymidine in viva. Cells stained for ATPase were also heavily labelled with silver grains in autoradiographs. Moreover, the cells were usually found in pairs which suggested that uptake of label and division had occurred in .ritu. It is not clear what stimulates L,Cs to divide in the periphery. Functional activities The two principal theories concerning the functional role of LCs are (1) they are involved in some way in the process of keratinisation or its control; (2) they are important components of the immune system which have the capacity to present antigens, control lymphoid cell traffic and perhaps play a role in the intra-thymic conditioning and differentiation of T lymphocytes. Role in keratinisation LCs appear in mouse and human fetal epidermis at the same time as the granular cell layer which is a concomitant of normal ortho-keratinisation. In the mouse tail, there are distinct scaly rings separated by interscale grooves. LCs, the granular cell layer and orthokeratinisation are confined to the interscale regions. In the scale centres, by contrast, LCs are absent and parakeratosis occurs in which squamous cells retain their nuclei and do not form a granular laye?. In fetal development, the formation of the interscale groove and the localisation of the granular layer and LCs to that region occur synchronouslyzD. If vitamin A acid is applied topically to the mouse tail, the zonal arrangement of both granular layer and LCs changes to a confluent distribution over the treated areaz6. In rats rendered deficient in vitamin A squamous metaplasia occurs in the mucosal linings of the trachea and bladder. This is accompanied by the appearance of LCs which are not normally found in those site+. Although clear associations with time and place exist between LCs and the process of keratinisation, the nature of their interrelationship remains to be established. If, however, the outer layers of skin including the LC layer are stripped off with ‘Sellotape’, squamous cells regenerate and keratinise several days before LCs are reconstitutedz8. Also, LCs
are not found in the Hassals corpuscles of thymus which are keratinised structures. Therefore, it seems unlikely that LCs exert direct control over squamouscell proliferation or kcratinisation. Role m the immune .\ystcm The resemblances between LCs and the abnormal macrophage-like cells of histiocytosis X*1 and the demonstration that LCs contain lysosome-like bodies and acid phosphatase, suggested they were of mesenchymal origin and have led to attempts to confirm their macrophage nature. SagebieP? showed they were phagocytic but less so than their neighbouring keratinocytes, engulfing less ferritin, both after local injection of ferritin in viuo and after incubation of skin s&es in ferritin solutions in vitro. However, it is possible that ferritin is too small to be a good test of phagocytosis. Although engulfed material was contained in vacuoles, none was found in association with LC granules. The possible relationship between LCs and macrophages is strengthened by the presence of surface receptors for the Fc fragment of IgG and for C3 which are also characteristic of macrophages. Morphological studies have revealed the active participation of LCs in allergic contact hypersensitivity reactions. The collected observations of several workers can be summarised in the following synthesis of events. The time course of changes in LC numbers at the site of primary sensitisation with dinitrofluorobenzene (DNFB) has been followed in mice by Bergstresser”. Within 12 h of the topical application of DNFB, the density of LCs fell from an initial level of 770/mm2 to less than 20/mm2. Over the next 12 h their numbers began to recover, reaching 90/mm2. Although a second application of DNFB was made at that point LCs continued to increase in number for a further 6 h, reaching a density of 220/mm2 by 30 h. By 36 h however (12 h after the second application of DNFB) they had more or less vanished (less than 20/mm2) but thereafter, they gradually returned to normal density. Silberberg and her co-workers studied responses to challenge with various antigens in guinea pigs that had been either actively or passively sensitised’4,30, LC numbers were expressed as their ratio to basal keratinocytes (BK). l‘hree to six hours after challenge of passively sensitised guinea pigs with dinitrochlorobenzene (DNCBj, the LC/BK ratio rose from an initial value of 0.07 to 0.09, Over the next 9-12 h the ratio fell to 0.04. From 24 h onwards, LC numbers rose back towards normal. Within 3-6 h of challenge with DNCB or ferritin3”, LCs increased in numbers in the dermis, appeared in dermal lymphatic vessels and then in the marginal sinuses and cortex of draining lymph nodes. In animals challenged with ferritin, the LCs were seen to bear ferritin both on their surface and within the cell. Close apposition of LCs to mononuclear cells was seen in the epidermis of actively sen-
127
sitised animals, whereas in passively sensitised animals, this apposition was seen in the dermis. The LCs appeared to suffer damage during this interaction in that glycogen granules appeared, the cytoplasm showed rarefaction and the perinuclear space became dilated. Some LCs were phagocytosed by macrophages while others showed ‘autophagocytosis’ of damaged organelles. The changes in LC numbers and their interaction with mononuclear cells was never seen at sites treated with irritants. Thus, LCs are highly mobile cells capable of changing their numbers rapidly and of carrying antigens to the regional lymph nodes and entering into close relations with mononuclear lymphoid cells. The functional importance of LCs in the induction of contact hypersensitivity has been studied by Streilein and colleagues in mouse tail and hamster cheek pouch in ZJZUO, both sites having low numbers of LCs. In addition, they depleted LCs further by exposure of skin to u.v.-B (290-310nm) irradiation. When DNFB was painted onto skin naturally deficient or artificially depleted of LCs, instead of active contact hypersensitivity, specific tolerance was induced. Ptak32 demonstrated that Fc bearing cells in the epidermis are specially potent inducers of contact sensitivity. Thus, hapten-coated cells, such as erythrocytes, thymocytes, spleen cells and peritoneal exudate cells, when given intravenously induced immunological tolerance. By contrast hapten-coated epiderma1 cells given intravenously induced persistant contact hypersensitivity. This was mainly a property of the cells bearing Fc receptors, presumed to be LCs. Thus, LCs are not only vital antigen-presenting cells for the induction of contact hypersensitivity, but when antigens enter without the intervention of LCs, tolerance develops. This discovery has great clinical and therapeutic importance. Their role in presentation of antigen has been examined further. They can replace other antigen-presenting cells such as monocyte/macrophages in antigeninduced stimulation of lymphocytes in uitro. Columnpurified guinea pig lymphocytes were induced to undergo proliferation by syngeneic LC-enriched stimulator cells pulsed with appropriate antigens (ovalbumin, PPD or TNP)?j. They failed to stimulate in allogeneic conditions and lost activity after treatment with anti-Ia serum and complement. Recently, human LCs have been shown to present antigens of PPD34, herpes simplex virus3s and nickel’O to purified autologous lymphocytes in vitro, thereby inducing a proliferative response. Role
in rejection
of skin
grafts
Both human and guinea pig LCs act as potent stimulator cells in mixed lymphocyte reactions’3,3”. It seems probable that the Ia antigens on LCs are important targets for the re,jection reaction against graf-
ted skin. Strcilein and co-workers?’ attempted to show that survival was prolonged if skin grafts were depleted of LCs. They transplanted mouse corneas, known to be almost completely deficient in both LCs and Ia antigens, to recipients which differed only in the I region of the major histocompatibility complex (MHC). Prolonged graft survival (up to 60 days) was observed; moreover, recipients were not primed against subsequent grafts of body wall skin which were rejected in the normal, first-set fashion. However, when body wall skin was grafted after irradiation with u.v.-B to deplete LCs, normal first-set rejection occurred and the recipients were also primed so that accelerated rejection of subsequent grafts was observed. Thus, it appeared that U.V. irradiation did not completely deplete the epidermis of Ia antigens, even though the LC population was functionally compromised as was shown by the development of tolerance induced with topical applications of DNFB. Relationship
to other
dendritic
cells
Dendritic cells lacking LC granules are found in the epidermis and are called ‘indeterminate’ cells. They also possess membrane-bound ATPase, have Ia antigens and probably Fc receptorsi’. It is possible that they are tither immature LCs which have not yet developed specific granules or that they are LCs in which granules were missed by the plane of section. Other types of dendritic cell have been described in various locations in the lymphoid tissue. These include the interdigitating reticulum cells, the follicular dendritic reticulum cells of NossalJx and the dendritic cells of Steinman and Cohn?“. The cells have similarities but also certain differences. Thus, although they all bear Ia antigens, the cells of Steinman and Cohn are ATPase negative and only LCs possess Birbeck granules. For a more comprehensive comparison see Thorbeckei/. Role
in control
of lymphoid
traffic
Streilein”” has proposed that LCs play a role in controlling T lymphocyte traffic, since during reaction to contact allergens, lymphocytes migrate into the epidermis and enter into close contact with LCs. He postulated the existence of a specialised lymphocyte traffic circuit, the skin-associated lymphoid tissue (SALT), analogous to those described for the gut (GALT) and bronchial (BALT) lymphoid circulations. It is suggested that during communication between LCs and T cells, some sort of conditioning stimulus is provided that causes T cells to ‘home’ to the epidermis. This is supported by the observation that in certain lymphoid malignancies such as mycosis fungoides, T lymphocytes have a predilection for the epidermis and infiltrating T cells have been seen in close relationship with LCs-“. Some sort of chronic antigen exposure may be responsible for the initial
128
epidermotrophic infiltration by T cells and their subsequent malignant change”‘.The abnormal T cells have been shown to function as helper cells42. Additional circumstantial evidence supporting the role of LCs in attracting T cells into the epidermis is the apparently special capacity of LCs to interact with helper ‘I’ cells. This is seen in the development of active contact sensitivity following intravenous injection of the hapten-coated LCs described above, in contrast to other cell-types which induce tolerance. The recent demonstration that 1’ cells of helper phenotype reside in areas of lymphoid tissues also inhabited by Ia-bearing dendritic cells some of which may be LCs, provides an anatomical basis for such a relationship43. A further role for LCs or LC-like, Ia-bearing, antigen-presenting cells has been postulated in the intrathymic conditioning of T cells, which renders them capable of recognising antigens presented in the appropriate context of histocompatibility44. Conclusions
In the short space of a few years, the LC has leapt to prominence in the fields of immunology and dermatology. It has been clearly established that LCs are of mesenchymal origin, derived from precursors resident in bone marrow. LCs populate many regions of the body and serve as essential outposts of the immune system. It is both frustrating and provocative that there are so few clues as to the nature of the Birbeck granule, the specific organelle by which LCs are identified. Apart from the now clearly established function of antigen presentation which results in the induction of contract sensitivity, hypotheses are burgeoning which propose a variety of additional activities for these cells. If these ideas stand the tests of experiment and scrutiny, then LCs may be implicated in some of the most fundamental intricacies of intercellular communication within the immune system.
References 1 Langerhans, P. (1868) Virchows Arch., (P&d And.) 44, 325-337 2 Ranvier, L. (1875) in 7razt~ ~uhnique n’%zrtologzr (Savy Edition, Paris) 3 Birbeck, M. S., Breathnach, A. S. and Everall, J. I). (1961) ,j’. Inmt. Drrmatol. 37, 51-63 4 Shelley, W. B. and Juhlin, L. (1976) .Nntuw(Lon&nz) 261, 46-47 5 Falck, B., Agrup, G., Jacobsson, S., Rorsman, H., Rosengren, E., Sachncr, K. and iigren, M. (1976) J. Invu.~l. I)rrmnlo/. 66, 265 6 Wolff, K. and Winkelmann, R. K. (1967),7. Inw1. f)rrrrrnlol. 48, 50-54 7 Stingl, G, Wolff-Schreiner, E. C. H., Pichler, W. J., Gschnait, F., Knapp, W. and Wolff, K. (1977) .N&ru jZ,ono’on) 268, 245-246 8 Burke, K. and Gigli, I. (1980) ,7. Inwl. 11urmn~ol. 75, 46-51
9 Rowden, G., Lewis, M. G. and Sullivan, A. L. (1977) .No~~P (London) 268, 247-248 10 Stingl, G., Tamaki, K. and Katz, S. I. (1980) Irnmunol. Kw 53, 149-174 11 Wolff, K. and Schreiner, R. (1970) 3. I nui,t. Dimnotol. 54, 37-47 12 Wolff, K. (1972) in Ourrnl !‘vrr,hlnn.r in D~rma~olqy 4, 79-145. S. Karger, Basle 13 Breathnach, A. S. (1977) Bnt.,7. &m&o/. 97, suppl. 15, 14 14 Silberberg, I., Bacr, R. L., Rosenthal, S. A., Thorbecke, G. J. and Bcrezowskv. V. 11975) C/I Irrimz~nol. 18. 435-453 ii. K. and Wolff: K. (1967) 3. Irwr\l. 15 Brown, J., Wi&lmann, Dermatol.
49, 386-390
16 Wolff, K. and Winkelmann, R. K. (1967) .7, Inurst. Drrmutol. 48, 504 17 Bergstresser, P. K., Toews, G B. and Streilcin, J. W. (1980) 3. Inuesl. Dcrmlol. 75, 73-77 18 Sil&, W. K. (1957) Amrr.J. And. 100, 225-239 19 Breathnach, A. S., Silvers, W. K., Smith, .J. and Heyner, S. (1968) 3. Inurst. Dwrmtol. 50, 147-l 60 20 Reams, W. M. and Tomkins, S. P. (1973) flu. U~ol. 31, 114-123 21 Turiaff, J. and Basset, F. (1965) Bull. Sfx. IMP//. Ho). (P&s) 116, 1197-1208 K. and Sachs, D. H. (1979) .Valurr jlon~on) 22 Katz, S. I., Tamaki, 282,324-326 23 Tamaki. K. and Katz, S. I. (198O)J. Inn&. I&w&ol. 75, 12-13 24 Frelinger, J. A. and Frelinger, J. G. (1980) J. Zmnt.I)ermnlol. 75,68-70 25 Mackenzie, I. C. (1975) Amr.3. And. 144, 127-136 26 Yong-Chuan Wong and Buck, K. C. (1971) 3. Inur.~l. Ihnntol. 56, 1 l-17 27 Schweizer, J. and Marks, F. (1977) 3 Inur~t. Drnnntoi. 69, 198-204 R. K. (1968)J. Inurst. 28 Lessard, R., Wolff, K. and Winkelmann, I1crmnlol. 50, 171-179 29 Sagebiel, R. W. (1972) ,7. Inrw11. Dpmvnlol. 58, 47-54 30 Silberberg-Sinakin, I., Thorbecke, G. J., Bacr, R. L., Rosenthal, S. A. and Berezowsky, V. (1976) CII. Ivurrunol. 25, 137-151 31 Streilein, J. W., ‘l‘oews, C. B. and Bergstrcsscr, P. R. (1980)3. Iriuu,t. Drrmntol. 75, 17-21 32 Ptak, W., Rozycka, I)., Askenasc, P. W. and Gershon, K. K. (1980)~7. f?q. Men’. 151, 362-375 33 Stingl, G., Katz, S. I., Clement, L., Green, I. and Shevach, E. M. (1978)J. Irnmunol. 121, 2005-2013 34 Braathen, L. R. and Thorsby, E. (1980) S~an12’.3. Immwzoi. 11, 401-408 L. Ii., Berle, E., Mobech-Hanssen, U. and Thorsby, 35 Braathen, E. (1980) Actn0um. I;pnuroi. (Stockhoirn) 60, M-387 36 Braathen, L. R. (1980) Br. J. 0vma&~l. 103, 5 17-526 G. J., Silberberg-Sinakin, I. and Flotte, T. (1980) 37 Thorbeckc, 7. Irw.\t.
Derrrrztol.
75, 32-43
38 Nossal, G. J. V., Abbott, A., Mitchell, J. and Lummus, Z (1968) J. Sufi. Md. 127, 227-290 39 Steinman, R. M. and Cohn, %. A. (1973) ,j’. Eup. A4rri. 137, 1142-1162 40 Strcilcin, W. J. (1978)J. Inrv~t. I1unuzloi. 71, 167-171 41 Rowden, G. and Lewis, M. G (1976) I,‘r. 3. D~m~~lo/. 95, 665-672 42 Broder, S., Uchiyama, T. and Waldmann, ‘I‘. A. (1979) (hnrrr Trd. Kq. 63, 607-6 12 43 Janossy, G., Tidman, N., Selby, W. S., Thomas, J. A., Granger, S., Kung, P. C. and Goldstein, G. (1980) .Nnturr (London) 288, 81-84 44 Longo, D. L and Schwartz, R. H. (1980) .rVnlrtrr (Lwu/un,n) 287, 44-46
methods will eventually reveal whether it really is time to give up the classical methods that have stood us in such good stead for several decades SIDNEY
SkIUI.iMAN
References 1 Shulman, S. (1975) Re,jroduclzon arid Anlzbody Kr.~jm.\r, CRC Pi-w, Cleveland 2 Shulman, S. (1978) in Spermatozoa, ilnlzlrodze~ unrl Irzfirlzl;~ (Cohen, J. and Hendry, W. F., eds.) pp. 81-99. Blackwell Scientific Publications, Oxford 3 Riimke, P. and Hekman, A. (1975) Clzn. Hnriocnnol. M&b. 4, 473-496
4 Hekman, A. and Kiimke, P. (1976) in -Tpx&ok cd Imvw~opathology (Micscher, P. A. and Miiller-Eberhard, H. J., eds.) pp, 947-962, Grune and Stratton, New York 5 Shulman, S., Zappi, E., Ahmed, U. and Davis, J. E. (1972) htracu/aon
5, 269-278
6 Mathur, S., Williamson, H. O., Derrick, F. C., Madyastha, P. R., Melchers, J. T., Holtz, C. L., Baker, E. R., Smith, C. L. and Fudcnberg, H. [I. (1981)s. ~rnmunoi. 126, 905-909 7 Mathur, S., Williamson, H. O., Landgrebc, S., Smith, C. L. and Fudenberg, H. H. (1979) 3. Ivununoi. M~t/wd\ 30, 38 l-393 8 Shulman, S. (1980) in Mamtnl o/ Uinzcal Zmmunology (Rose, N. R. and Friedman, H., eds.) 2nd edn, pp. 907-916, Waverly Press, Baltimore 9 Ingerslev, H. J. (1979) M.3. krtzl. 24, 1-12 10 Ingerslev, H. J. and Hjort, T. (1979) I;rrlz/. SI&. 31, 496-502 11 Hamerlynck, J. and Riimke, P. (1968) ,7, &prod. EPrtzl. 17, 191-194
The immunobiology of Langerhans cells P. S. Friedmann Uepartmcnt
of Dermatology,
The Royal Victoria
Infirmary, The University Tyne, NE1 4LP, U.K.
Interest in the biology oj’ Langerhans cells has recently been .stimulatrd Friedmann, that they aw active members the immune .syrtem.
of Newcastle-Upon-Tyne, by ob.trrvations,
Newcastle-Uponrevxwed
here by Petu
?f
The cells discovered in 1868 by Paul Langerhansr have been overlooked and unstudied, probably because they were difficult to see with normal histological stains. Langerhans found that they could be stained with gold chloride and concluded they were therefore nerve cells. In 1875, Kanvier2 suggested that the cells were of lymphoid origin and over the years since then they have been regarded as epidermal cells, artefacts, melanocytes and neural elements such as Schwann cells. In 1961, Birbeck and co-workers’ described the specific organelle or granule by which Langerhans cells (LCs) are now identified by electron microscopy. Once LCs could be identified with certainty, the way was opened for progress in the analysis of their structure and function. Methods
of demonstration
and
identification
I.igh t micro.\ coj~y
Methods for the demonstration of LCs by light microscopy have improved. They can be stained with certain metals including gold, osmium, cobalt, lanthanum, mercury, nickel and chromium. They take up
supra-vital dyes such as methylene blue and brilliant such as cresyl blue, and a variety of substances formaldehyde, glutaraldehyde, ethylene-diamine and paraphenylene-diamine which can act as contact allergens and can be identified histochemically. This was the property that led Shelley and Juhlin to postulate that LCs formed a reticula-epithelial system involved in trapping external antigens4. LCs have an affinity for various catecholamines including dopamine, noradrenaline and L-DOPAS. After exposure of the skin to these substances u-z vitro, and subsequent exposure to formaldehyde vapour fluorescent products are formed in the cells. LCs may be stained by a number of histochemical methods. The most widely used involves the demonstration of formalin-resistant adenosine triphosphatase (A’l’Pasc) which is present in the outer membrane of the cell’l. In the epidermis, the stain is specific for these cells. Human LCs (cell body only) can be recognised by the presence of a-mannosidase, the guinea pig’s LCs contain aminopeptidasc and murine LCs acetyl esterase.
125
Immunol~~gicul murkers LCs possess immunological markers and receptors similar to those found on macrophages. Thus, they bind the Fc fragment of IgG’ as well as components C3b, C3d, C4b and C4d of complemenP. In addition, human and mouse LCs bear histocompatibility antigens of the D/DR” and I-A and I-E/Cl0 specificities, respectively. Moreover, they appear to be the only cells in the epidermis to produce such ‘Ia’ antigens. The presence of receptors has allowed the identification and manipulation of LCs. For example, LCs in dissociated cell suspensions may be selectively enriched by rosetting procedures and then used in assays of their functional capacities. Electron mzcroscopy The collective results of ultrastructural studies have revealed that LCs are dendritic cells with a lobulated nucleus and clear cytoplasm containing microtubules and microfilaments (concomitants of motility) but lacking tonofilaments and desmosomes (characteristic of epithelial cells). They contain specific LC granules, lysosomes and mitochondria. The
Langerhans
cell
Fig. 1. Location
of Langerhans
cells.
skin, there are density of 4-800/ mm2 Is. In guinea-pig 2 I6 and similar numbers have about 800-lOOO/mm been found in the skin of other rodents. Certain anatomical sites have reduced numbers and uneven distributions of LCs. They are absent from cornea” and are reduced to 130/mm* in, hamster cheek pouch”; mouse tail skin has a regular pattern of scales with LCs clustered at the interscale regions but deficient in the central scale zones. The average density is about 1 lo-260/mm2 and varies with strainr6.
granule
The specific LC granules described by Birbeck3 and visible only at the ultrastructural level are still regarded as the characteristic by which LCs are defined. In ultrathin sections, the granules are rod-shaped with a central dense striated band. They may have an expanded end, giving a ‘raquet-like’ appearance. Granules have been seen fused with the outer membrane of the cell and, when in that position, they may take up protein labels such as peroxidase”. The intracellular granules stain with the reaction product of osmiumzinc iodide whereas those which have made contact with the outer membrane do not. The granules have not been shown to possess enzymic activity and it is not yet established whether they are formed from the outer membrane or the Golgi apparatus. No functional activity has yet been ascribed to them. For a full review see Wolffrz. Distribution
LCs are found in mammals in most squamous epithelia including the skin, buccal mucosa, oesophagus, and vaginal and rectal mucosae. In skin, they occur mainly in the mid-epidermis (Fig. 1.) where they have been called Type 1 LCs on the basis of many dendrites, electron-dense cytoplasm and numerous Birbeck granulesr3. Type II LCs are found in the basal layer of epidermis and have few dendrites or Birbeck mitochondria and electron-dense granules, many cytoplasm. LCs or LC-like cells are also found in the dermis and in lymphoid organs including thymus, lymph nodes and spleen (for references see Ref. 14). In human epidermis they represent 3-8% of the total number of epidermal cells and are present at a
Origin
The tissue of origin of LCs has only been established recently. The chain of discovery began with the demonstration that they were not of neuraland others”,*O showed that crest origin. SilversI mouse skin deprived of its neural crest elements would still develop LCs but not Schwann cells and melanocytes. Evidence accumulated to indicate that they might be of mesenchymal origin’*; they are distributed in lymphoid tissues, they resemble the cells of histiocytosis X (which also contain Birbeck granuleszl) and hairy cell leukaemia and they have surface markers which are similar to those of the macrophage/monocyte lineage. Katz and co-workers22 first showed they were derived from a mobile pool by transplanting skin from parental mice to appropriate Fl hybrid recipients. Subsequently, the H-2 and Ia specificities of LCs in the grafts were determined at various times. It was observed that from 11 days after transplantation, there was a gradual increase in the proportion of LCs whose histocompatibility phenotype was that of the recipient. Keratinocytes, on the other hand, were always of donor phenotype. Those workers went on to show that LCs arrived from a central source, the bone marrow, rather than by local lateral migration from ad,jacent skin. They determined the histocompatibility phenotypes of LCs and epidermal keratinocytes in radiation chimaeras, i.e. irradiated mice reconstituted with allogeneic bone-marrow. By three days after reconstitution, a small percentage of LCs bearing Ia antigens of the marrow donor phenotype were detectablezi. By three months, 80% of LCs were of donor phenotype. Moreover, Fre-
126
linge? showed that epidermal la antigens were synthesised by bone-marrow-derived cells in the epidermis and were not simply passively adsorbed. ‘Thus, LCs are unequivocally derived from a central pool of precursors in bone marrow, from which they are continually renewed and replaced in the periphery. In normal marrow there are no cells containing LC granules. It is likely, therefore, that LC granules are only formed once the cell has taken up a position in which it is required to be functionally mature. It is interesting that some LCs appear able to synthesise DNA and divide once they have taken up residence in the periphery. Thus, Mackenziez5 showed that LCs in normal epidermis would incorporate ‘H-thymidine in viva. Cells stained for ATPase were also heavily labelled with silver grains in autoradiographs. Moreover, the cells were usually found in pairs which suggested that uptake of label and division had occurred in .ritu. It is not clear what stimulates L,Cs to divide in the periphery. Functional activities The two principal theories concerning the functional role of LCs are (1) they are involved in some way in the process of keratinisation or its control; (2) they are important components of the immune system which have the capacity to present antigens, control lymphoid cell traffic and perhaps play a role in the intra-thymic conditioning and differentiation of T lymphocytes. Role in keratinisation LCs appear in mouse and human fetal epidermis at the same time as the granular cell layer which is a concomitant of normal ortho-keratinisation. In the mouse tail, there are distinct scaly rings separated by interscale grooves. LCs, the granular cell layer and orthokeratinisation are confined to the interscale regions. In the scale centres, by contrast, LCs are absent and parakeratosis occurs in which squamous cells retain their nuclei and do not form a granular laye?. In fetal development, the formation of the interscale groove and the localisation of the granular layer and LCs to that region occur synchronouslyzD. If vitamin A acid is applied topically to the mouse tail, the zonal arrangement of both granular layer and LCs changes to a confluent distribution over the treated areaz6. In rats rendered deficient in vitamin A squamous metaplasia occurs in the mucosal linings of the trachea and bladder. This is accompanied by the appearance of LCs which are not normally found in those site+. Although clear associations with time and place exist between LCs and the process of keratinisation, the nature of their interrelationship remains to be established. If, however, the outer layers of skin including the LC layer are stripped off with ‘Sellotape’, squamous cells regenerate and keratinise several days before LCs are reconstitutedz8. Also, LCs
are not found in the Hassals corpuscles of thymus which are keratinised structures. Therefore, it seems unlikely that LCs exert direct control over squamouscell proliferation or kcratinisation. Role m the immune .\ystcm The resemblances between LCs and the abnormal macrophage-like cells of histiocytosis X*1 and the demonstration that LCs contain lysosome-like bodies and acid phosphatase, suggested they were of mesenchymal origin and have led to attempts to confirm their macrophage nature. SagebieP? showed they were phagocytic but less so than their neighbouring keratinocytes, engulfing less ferritin, both after local injection of ferritin in viuo and after incubation of skin s&es in ferritin solutions in vitro. However, it is possible that ferritin is too small to be a good test of phagocytosis. Although engulfed material was contained in vacuoles, none was found in association with LC granules. The possible relationship between LCs and macrophages is strengthened by the presence of surface receptors for the Fc fragment of IgG and for C3 which are also characteristic of macrophages. Morphological studies have revealed the active participation of LCs in allergic contact hypersensitivity reactions. The collected observations of several workers can be summarised in the following synthesis of events. The time course of changes in LC numbers at the site of primary sensitisation with dinitrofluorobenzene (DNFB) has been followed in mice by Bergstresser”. Within 12 h of the topical application of DNFB, the density of LCs fell from an initial level of 770/mm2 to less than 20/mm2. Over the next 12 h their numbers began to recover, reaching 90/mm2. Although a second application of DNFB was made at that point LCs continued to increase in number for a further 6 h, reaching a density of 220/mm2 by 30 h. By 36 h however (12 h after the second application of DNFB) they had more or less vanished (less than 20/mm2) but thereafter, they gradually returned to normal density. Silberberg and her co-workers studied responses to challenge with various antigens in guinea pigs that had been either actively or passively sensitised’4,30, LC numbers were expressed as their ratio to basal keratinocytes (BK). l‘hree to six hours after challenge of passively sensitised guinea pigs with dinitrochlorobenzene (DNCBj, the LC/BK ratio rose from an initial value of 0.07 to 0.09, Over the next 9-12 h the ratio fell to 0.04. From 24 h onwards, LC numbers rose back towards normal. Within 3-6 h of challenge with DNCB or ferritin3”, LCs increased in numbers in the dermis, appeared in dermal lymphatic vessels and then in the marginal sinuses and cortex of draining lymph nodes. In animals challenged with ferritin, the LCs were seen to bear ferritin both on their surface and within the cell. Close apposition of LCs to mononuclear cells was seen in the epidermis of actively sen-
127
sitised animals, whereas in passively sensitised animals, this apposition was seen in the dermis. The LCs appeared to suffer damage during this interaction in that glycogen granules appeared, the cytoplasm showed rarefaction and the perinuclear space became dilated. Some LCs were phagocytosed by macrophages while others showed ‘autophagocytosis’ of damaged organelles. The changes in LC numbers and their interaction with mononuclear cells was never seen at sites treated with irritants. Thus, LCs are highly mobile cells capable of changing their numbers rapidly and of carrying antigens to the regional lymph nodes and entering into close relations with mononuclear lymphoid cells. The functional importance of LCs in the induction of contact hypersensitivity has been studied by Streilein and colleagues in mouse tail and hamster cheek pouch in ZJZUO, both sites having low numbers of LCs. In addition, they depleted LCs further by exposure of skin to u.v.-B (290-310nm) irradiation. When DNFB was painted onto skin naturally deficient or artificially depleted of LCs, instead of active contact hypersensitivity, specific tolerance was induced. Ptak32 demonstrated that Fc bearing cells in the epidermis are specially potent inducers of contact sensitivity. Thus, hapten-coated cells, such as erythrocytes, thymocytes, spleen cells and peritoneal exudate cells, when given intravenously induced immunological tolerance. By contrast hapten-coated epiderma1 cells given intravenously induced persistant contact hypersensitivity. This was mainly a property of the cells bearing Fc receptors, presumed to be LCs. Thus, LCs are not only vital antigen-presenting cells for the induction of contact hypersensitivity, but when antigens enter without the intervention of LCs, tolerance develops. This discovery has great clinical and therapeutic importance. Their role in presentation of antigen has been examined further. They can replace other antigen-presenting cells such as monocyte/macrophages in antigeninduced stimulation of lymphocytes in uitro. Columnpurified guinea pig lymphocytes were induced to undergo proliferation by syngeneic LC-enriched stimulator cells pulsed with appropriate antigens (ovalbumin, PPD or TNP)?j. They failed to stimulate in allogeneic conditions and lost activity after treatment with anti-Ia serum and complement. Recently, human LCs have been shown to present antigens of PPD34, herpes simplex virus3s and nickel’O to purified autologous lymphocytes in vitro, thereby inducing a proliferative response. Role
in rejection
of skin
grafts
Both human and guinea pig LCs act as potent stimulator cells in mixed lymphocyte reactions’3,3”. It seems probable that the Ia antigens on LCs are important targets for the re,jection reaction against graf-
ted skin. Strcilein and co-workers?’ attempted to show that survival was prolonged if skin grafts were depleted of LCs. They transplanted mouse corneas, known to be almost completely deficient in both LCs and Ia antigens, to recipients which differed only in the I region of the major histocompatibility complex (MHC). Prolonged graft survival (up to 60 days) was observed; moreover, recipients were not primed against subsequent grafts of body wall skin which were rejected in the normal, first-set fashion. However, when body wall skin was grafted after irradiation with u.v.-B to deplete LCs, normal first-set rejection occurred and the recipients were also primed so that accelerated rejection of subsequent grafts was observed. Thus, it appeared that U.V. irradiation did not completely deplete the epidermis of Ia antigens, even though the LC population was functionally compromised as was shown by the development of tolerance induced with topical applications of DNFB. Relationship
to other
dendritic
cells
Dendritic cells lacking LC granules are found in the epidermis and are called ‘indeterminate’ cells. They also possess membrane-bound ATPase, have Ia antigens and probably Fc receptorsi’. It is possible that they are tither immature LCs which have not yet developed specific granules or that they are LCs in which granules were missed by the plane of section. Other types of dendritic cell have been described in various locations in the lymphoid tissue. These include the interdigitating reticulum cells, the follicular dendritic reticulum cells of NossalJx and the dendritic cells of Steinman and Cohn?“. The cells have similarities but also certain differences. Thus, although they all bear Ia antigens, the cells of Steinman and Cohn are ATPase negative and only LCs possess Birbeck granules. For a more comprehensive comparison see Thorbeckei/. Role
in control
of lymphoid
traffic
Streilein”” has proposed that LCs play a role in controlling T lymphocyte traffic, since during reaction to contact allergens, lymphocytes migrate into the epidermis and enter into close contact with LCs. He postulated the existence of a specialised lymphocyte traffic circuit, the skin-associated lymphoid tissue (SALT), analogous to those described for the gut (GALT) and bronchial (BALT) lymphoid circulations. It is suggested that during communication between LCs and T cells, some sort of conditioning stimulus is provided that causes T cells to ‘home’ to the epidermis. This is supported by the observation that in certain lymphoid malignancies such as mycosis fungoides, T lymphocytes have a predilection for the epidermis and infiltrating T cells have been seen in close relationship with LCs-“. Some sort of chronic antigen exposure may be responsible for the initial
128
epidermotrophic infiltration by T cells and their subsequent malignant change”‘.The abnormal T cells have been shown to function as helper cells42. Additional circumstantial evidence supporting the role of LCs in attracting T cells into the epidermis is the apparently special capacity of LCs to interact with helper ‘I’ cells. This is seen in the development of active contact sensitivity following intravenous injection of the hapten-coated LCs described above, in contrast to other cell-types which induce tolerance. The recent demonstration that 1’ cells of helper phenotype reside in areas of lymphoid tissues also inhabited by Ia-bearing dendritic cells some of which may be LCs, provides an anatomical basis for such a relationship43. A further role for LCs or LC-like, Ia-bearing, antigen-presenting cells has been postulated in the intrathymic conditioning of T cells, which renders them capable of recognising antigens presented in the appropriate context of histocompatibility44. Conclusions
In the short space of a few years, the LC has leapt to prominence in the fields of immunology and dermatology. It has been clearly established that LCs are of mesenchymal origin, derived from precursors resident in bone marrow. LCs populate many regions of the body and serve as essential outposts of the immune system. It is both frustrating and provocative that there are so few clues as to the nature of the Birbeck granule, the specific organelle by which LCs are identified. Apart from the now clearly established function of antigen presentation which results in the induction of contract sensitivity, hypotheses are burgeoning which propose a variety of additional activities for these cells. If these ideas stand the tests of experiment and scrutiny, then LCs may be implicated in some of the most fundamental intricacies of intercellular communication within the immune system.
References 1 Langerhans, P. (1868) Virchows Arch., (P&d And.) 44, 325-337 2 Ranvier, L. (1875) in 7razt~ ~uhnique n’%zrtologzr (Savy Edition, Paris) 3 Birbeck, M. S., Breathnach, A. S. and Everall, J. I). (1961) ,j’. Inmt. Drrmatol. 37, 51-63 4 Shelley, W. B. and Juhlin, L. (1976) .Nntuw(Lon&nz) 261, 46-47 5 Falck, B., Agrup, G., Jacobsson, S., Rorsman, H., Rosengren, E., Sachncr, K. and iigren, M. (1976) J. Invu.~l. I)rrmnlo/. 66, 265 6 Wolff, K. and Winkelmann, R. K. (1967),7. Inw1. f)rrrrrnlol. 48, 50-54 7 Stingl, G, Wolff-Schreiner, E. C. H., Pichler, W. J., Gschnait, F., Knapp, W. and Wolff, K. (1977) .N&ru jZ,ono’on) 268, 245-246 8 Burke, K. and Gigli, I. (1980) ,7. Inwl. 11urmn~ol. 75, 46-51
9 Rowden, G., Lewis, M. G. and Sullivan, A. L. (1977) .No~~P (London) 268, 247-248 10 Stingl, G., Tamaki, K. and Katz, S. I. (1980) Irnmunol. Kw 53, 149-174 11 Wolff, K. and Schreiner, R. (1970) 3. I nui,t. Dimnotol. 54, 37-47 12 Wolff, K. (1972) in Ourrnl !‘vrr,hlnn.r in D~rma~olqy 4, 79-145. S. Karger, Basle 13 Breathnach, A. S. (1977) Bnt.,7. &m&o/. 97, suppl. 15, 14 14 Silberberg, I., Bacr, R. L., Rosenthal, S. A., Thorbecke, G. J. and Bcrezowskv. V. 11975) C/I Irrimz~nol. 18. 435-453 ii. K. and Wolff: K. (1967) 3. Irwr\l. 15 Brown, J., Wi&lmann, Dermatol.
49, 386-390
16 Wolff, K. and Winkelmann, R. K. (1967) .7, Inurst. Drrmutol. 48, 504 17 Bergstresser, P. K., Toews, G B. and Streilcin, J. W. (1980) 3. Inuesl. Dcrmlol. 75, 73-77 18 Sil&, W. K. (1957) Amrr.J. And. 100, 225-239 19 Breathnach, A. S., Silvers, W. K., Smith, .J. and Heyner, S. (1968) 3. Inurst. Dwrmtol. 50, 147-l 60 20 Reams, W. M. and Tomkins, S. P. (1973) flu. U~ol. 31, 114-123 21 Turiaff, J. and Basset, F. (1965) Bull. Sfx. IMP//. Ho). (P&s) 116, 1197-1208 K. and Sachs, D. H. (1979) .Valurr jlon~on) 22 Katz, S. I., Tamaki, 282,324-326 23 Tamaki. K. and Katz, S. I. (198O)J. Inn&. I&w&ol. 75, 12-13 24 Frelinger, J. A. and Frelinger, J. G. (1980) J. Zmnt.I)ermnlol. 75,68-70 25 Mackenzie, I. C. (1975) Amr.3. And. 144, 127-136 26 Yong-Chuan Wong and Buck, K. C. (1971) 3. Inur.~l. Ihnntol. 56, 1 l-17 27 Schweizer, J. and Marks, F. (1977) 3 Inur~t. Drnnntoi. 69, 198-204 R. K. (1968)J. Inurst. 28 Lessard, R., Wolff, K. and Winkelmann, I1crmnlol. 50, 171-179 29 Sagebiel, R. W. (1972) ,7. Inrw11. Dpmvnlol. 58, 47-54 30 Silberberg-Sinakin, I., Thorbecke, G. J., Bacr, R. L., Rosenthal, S. A. and Berezowsky, V. (1976) CII. Ivurrunol. 25, 137-151 31 Streilein, J. W., ‘l‘oews, C. B. and Bergstrcsscr, P. R. (1980)3. Iriuu,t. Drrmntol. 75, 17-21 32 Ptak, W., Rozycka, I)., Askenasc, P. W. and Gershon, K. K. (1980)~7. f?q. Men’. 151, 362-375 33 Stingl, G., Katz, S. I., Clement, L., Green, I. and Shevach, E. M. (1978)J. Irnmunol. 121, 2005-2013 34 Braathen, L. R. and Thorsby, E. (1980) S~an12’.3. Immwzoi. 11, 401-408 L. Ii., Berle, E., Mobech-Hanssen, U. and Thorsby, 35 Braathen, E. (1980) Actn0um. I;pnuroi. (Stockhoirn) 60, M-387 36 Braathen, L. R. (1980) Br. J. 0vma&~l. 103, 5 17-526 G. J., Silberberg-Sinakin, I. and Flotte, T. (1980) 37 Thorbeckc, 7. Irw.\t.
Derrrrztol.
75, 32-43
38 Nossal, G. J. V., Abbott, A., Mitchell, J. and Lummus, Z (1968) J. Sufi. Md. 127, 227-290 39 Steinman, R. M. and Cohn, %. A. (1973) ,j’. Eup. A4rri. 137, 1142-1162 40 Strcilcin, W. J. (1978)J. Inrv~t. I1unuzloi. 71, 167-171 41 Rowden, G. and Lewis, M. G (1976) I,‘r. 3. D~m~~lo/. 95, 665-672 42 Broder, S., Uchiyama, T. and Waldmann, ‘I‘. A. (1979) (hnrrr Trd. Kq. 63, 607-6 12 43 Janossy, G., Tidman, N., Selby, W. S., Thomas, J. A., Granger, S., Kung, P. C. and Goldstein, G. (1980) .Nnturr (London) 288, 81-84 44 Longo, D. L and Schwartz, R. H. (1980) .rVnlrtrr (Lwu/un,n) 287, 44-46
