Pathological mechanisms in experimental autoimmune myasthenia gravis. II. Passive transfer of experimental autoimmune myasthenia gravis in rats with anti-acetylcholine recepotr antibodies.

Passive transfer of experimental autoimmune myasthenia gravis (EAMG) was achieved using the gamma globulin fraction and purified IgG from sera of rats immunized with Electrophus electricus (eel) acetylcholine receptor (AChR). This demonstrates the critical role of anti-AChR antibodies in impairing neuromuscular transmission in EAMG. Passive transfer of anti-AChR antibodies from rats with chronic EAMG induced signs of the acute phase of EAMG in normal recipient rats, including invasion of the motor end-plate region by mononuclear inflammatory cells. Clinical, eletrophysiological, histological, and biochemical signs of acute EAMG were observed by 24 h after antibody transfer. Recipient rats developed profound weakness and fatigability, and the posture characteristic of EAMG. Striking weight loss was attributable to dehydration. Recipient rats showed large decreases in amplitude of muscle responses to motor nerve stimulation, and repetitive nerve stimulation induced characteristic decrementing responses. End-plate potentials were not detectable in many muscle fibers, and the amplitudes of miniature end-plate potentials were reduced in the others. Passively transferred EAMG more severely affected the forearm muscles than diaphragm muscles, though neuromuscular transmission was impaired and curare sensitivity was increased in both muscles. Some AChR extracted from the muscles of rats with passively transferred EAMG was found to be complexed with antibody, and the total yield of AChR per rat was decreased. The quantitative decrease in AChR approximately paralleled in time the course of clinical and electrophysiological signs. The amount of AChR increased to normal levels and beyond at the time neuromuscular transmission was improving. The excess of AChR extractable from muscle as the serum antibody level decreased probably represented extrajunctional receptors formed in response to functional denervation caused by phagocytosis of the postsynaptic membrane by macrophages. The amount of antibody required to passively transfer EAMG was less than required to bind all AChR molecules in a rat's musculature. The effectiveness of samll amounts of antibody was probably amplified by the activation of complement and by the destruction of large areas of postsynaptic membrane by phagocytic cells. A self-sustaining autoimmune response to AChR was not provoked in animals with passively transferred EAMG.

Antagonism of voltage-gated calcium channels in small cell carcinomas of patients with and without Lambert-Eaton myasthenic syndrome by autoantibodies omega-conotoxin and adenosine.

The Lambert-Eaton myasthenic syndrome (LES) is an autoimmune presynaptic disorder of peripheral cholinergic neurotransmission in which there is often an associated small cell lung carcinoma (SCC). SCC lines established from patients with and without LES exhibit a Ca2+ influx response to depolarization by K+ that is consistent with the presence of voltage-gated Ca2+ channels. Autoantibodies antagonistic to SCC Ca2+ channel activity were found exclusively in patients with LES, independent of cancer status. Depolarization-induced uptake of 45Ca2+ by SCC lines was reduced maximally after 3-4 days of exposure to serum immunoglobulins from 14 of 19 LES patients, while 53 control immunoglobulins (including patients with SCC, other tumors, other paraneoplastic syndromes, and other neurological and autoimmune diseases) were without effect. The snail neurotoxin omega-conotoxin of subtype GVIA, which is a specific antagonist of presynaptic Ca2+ channels, inhibited K+-stimulated Ca2+ uptake in a dose-dependent manner that was essentially irreversible. Adenosine, reported to be a specific antagonist of neuronal Ca2+ channels, also impaired voltage-stimulated Ca2+ influx in SCC. Use of LES patients' IgG and omega-conotoxin in further studies of SCC may facilitate identification and purification of the LES antigen(s) and yield a quantitative serological test for diagnosing this autoimmune paraneoplastic syndrome.

Ultrastructural localization of the terminal and lytic ninth complement component (C9) at the motor end-plate in myasthenia gravis.

The terminal and lytic complement component (C9) was localized at the motor end-plate in acquired autoimmune myasthenia gravis (MG) by the immunoperoxidase method, with adequate preservation of fine structure and negligible background staining. C9 was localized on short segments of the postsynaptic membrane on degenerated fragments of the junctional folds shed into the synaptic space, and on disintegrating junctional folds. An inverse relationship was noted between the structural integrity of the junctional folds and the abundance of C9 at a given end-plate region. Destruction of junctional folds by complement may induce relocation of the nerve terminal and increased spatial separation of end-plate regions on the muscle fiber. Destruction of junctional folds by the complement membrane attack complex is a cause of the acetylcholine receptor deficiency at the MG end-plate, but antibody-dependent modulation of the receptor may also contribute to deficiency of the receptor. In certain disorders other than autoimmune MG, pathological mechanisms other than complement-mediated lysis may affect the structural integrity of the postsynaptic region.

Experimental autoimmune myasthenia gravis: a sequential and quantitative study of the neuromuscular junction ultrastructure and electrophysiologic correlations.

Neuromuscular junction ultrastructure in rat forelimb digit extensor muscle was sequentially and quantitatively investigated in experimental autoimmune myasthenia gravis (EAMG). Experimental animals were immunized with highly purified eel electroplax acetylcholine receptor protein plus complete Freund's adjuvant and B pertussis vaccine; control animals received only adjuvant and vaccine. During the first 7 days (latent period) after immunization end-plate structure and neuromuscular transmission remained normal in the experimental group. Between day 7 and 11 (acute phase) mononuclear cells infiltrated those regions of muscle where the end-plates were located and there was intense degeneration of the postsynaptic regions with splitting away of abnormal junctional folds from the underlying muscle fibers. Macrophages entered the gaps thus formed and removed the degenerating folds by phagocytosis. The nerve terminals were displaced from their usual location but maintained their structural integrity. Neuromuscular transmission was blocked in many muscle fibers. Miniature end-plate potentias (MEPPs), detectable in only a few fibers, were of abnormally low amplitude. After day 11 (chronic phase) the nerve terminals returned to the highly simplified postsynaptic folds became reconstituted and again degenerated. Immature junctions with poorly differentiated postsynaptic regions and nerve sprouts near end-plates were also observed. In two animals relapsing during the chronic phase degeneration of the postsynaptic folds was more intense than in the other chronic-phase animals. The posysynaptic membrane length and length per unit area and the MEPP amplitudes were significantly decreased in all chronic phase animals and the decreases were greater in the relapsing than in the non-lapsing animals. Minor morphometric alterations were also observed in the nerve terminals. These might have been secondary to the postsynaptic changes. The postsynaptic region is the primary target of the autoimmune reaction in EAMG. The ultrastructural, morphometric and electrophysiological abnormalities of the end-plate in chronic EAMG resemble those which have been observed in human myasthenia gravis.

Lead neuropathy. 1) Morphometry, nerve conduction, and choline acetyltransferase transport: new finding of endoneurial edema associated with segmental demyelination.

Morphometric and pathologic studies along the length of the peripheral nervous system were obtained in groups of rats fed 4% lead carbonate for 3 and 6 months and in match-fed controls. The number and diameter histograms of L6 cytons of spinal ganglia and of myelinated fibers of proximal and distal portions of peroneal and sural nerve were not significantly different from the control groups. On the other hand, segmental demyelination occurred approximately as frequently in proximal as in distal parts of nerves. At 3 months approximately 1/3 of teased myelinated fibers showed changes of segmental demyelination (Condition C), or of remyelination after segmental demyelination (Condition F) or of both segmental demyelination and of remyelination (Condition D), while at 6 months more than 4/5ths of fibers showed these changes. As expected, regression lines of axonal area on number of lamellae of myelin, were less steep in nerves of rats fed on lead for 6 months as compared to controls. Axonal transport of choline acetyltransferase in lead neuropathy did not differ from that in control rats. As expected from the studies of others, conduction velocity of myelinated fibers of caudal nerve were low. A new finding was the often quite striking increase of transverse fascicular area of peripheral nerves. This was due to edema which appeared to develop at about the time of onset os segmental demyelination. Although the edema may be an epiphenomenon, it could be an important observation bearing on the development of lead neuropathy. It would be important to know next whether or not the blood nerve barrier is altered in lead neuropathy.

Ultrastructural localization of immune complexes (IgG and C3) at the end-plate in experimental autoimmune myasthenia gravis.

Rats immunized with purified torpedo acetylcholine receptor (AChR) plus adjuvants developed chronic experimental autoimmune myasthenia gravis (EAMG) after day 28. Forelimb muscles from EAMG rats 29 to 103 days after immunization and from control animals were used for the ultrastructural localization of IgG and C3. IgG was demonstrated with rabbit anti-rat IgG followed by treatment with peroxidase-labeled staphylococcal protein A; and C3 with peroxidase-labeled rabbit anti-rat C3, or with unlabeled rabbit anti-rat C3 followed by peroxidase-labeled protein A. In EAMG rats both IgG and C3 were localized on the terminal expansions of the junctional folds, where AChR is known to be located, and on detached, degenerated parts of the folds in the synaptic space. Background staining was negligible. The findings provide unambiguous evidence for a destructive autoimmune reaction involving the postsynaptic membrane in EAMG, implicate the complement system in this reaction and show that detachment of the tips of the junctional folds is one way by which immune complexes, and AChR, are eliminated from the postsynaptic membrane. The immuno-electron microscopic findings in chronic EAMG closely resemble those described in human myasthenia gravis.

Interstitial hyperosmolarity may cause axis cylinder shrinkage in streptozotocin diabetic nerve.

Maximal conduction velocity values of nerves of diabetic rats 20 weeks after streptozotocin intoxication were found to be intermediate between those of onset-control and those of end-control groups. The abnormality of conduction velocity of the streptozotocin group might therefore be attributed to a failure of maturation. Detailed electron microscopic morphometry of myelinated fibers (MFs) indicates that more than lack of maturation is involved. Whereas the number of lamellae and the perimeter of axis cylinders of myelinated fibers of the three study groups suggested that growth continues, cross-sectional area of the axis cylinders of the streptozotocin group was smaller than those of either control group. Scored evaluation of fiber shape and the measured index of circularity, which related perimeter and transverse axis cylinder area, also indicated that a selective shrinkage of axis cylinders had occurred. This selective alteration in size and shape of axis cylinders is identical to that described after hyperosmolar fixation. Compared with that of controls, the serum of streptozotocin rats is hyperosmolar. It would seem reasonable to attribute the axis cylinder changes to shrinkage. Whether an additional maturational effect is operative as well cannot be resolved from our data.

Structural and biochemical effects of essential fatty acid deficiency on peripheral nerve.

The effect of postweaning essential fatty acid (EFA) deficiency on the peripheral nerve was studied in groups of rats. At 325 days, the characteristic biochemical changes of EFA deficiency were present in isolated peripheral myelin, although to a lesser degree than reported in non-neural tissues. There was no significant difference between control and deficient groups in number or size distributions of myelinated fibers (MFs) in muscle and sensory nerves, in the incidence of teased fiber abnormalities, in rates of axonal transport of dopamine-beta-hydroxylase and acetylcholinesterase, or in conduction velocity and compound action potentials of peripheral nerve in vivo or in vitro. Four weeks after a standard sciatic crush injury, the median MF diameter in regenerated peroneal nerves was significantly smaller in EFA-deficient rats than in control rats, but this difference was no longer significant at 18 weeks. At 18 weeks, EFA-deficient and control regenerated nerves showed similar myelin periodicity and relationship of axonal area to number of myelin lamellae. We conclude that acquired EFA deficiency in the rat leads to biochemically abnormal peripheral myelin, but that this state is unaccompanied by clinical, physiological, or morphological evidence of neuropathy.

Duchenne dystrophy: ultrastructural localization of the acetylcholine receptor and intracellular microelectrode studies of neuromuscular transmission.

Despite focal degeneration and simplification of the postsynaptic region in Duchenne dystrophy, the postsynaptic acetylcholine receptor (AChR) is preserved. This is in contrast to myasthenia gravis where similar postsynaptic alterations are invariably associated with a marked decrease in AChR. There is no extrajunctional spread of AChR in Duchenne dystrophy. The amplitude and frequency of miniature end-plate potentials and the number of transmitter quanta released by nerve impulse are normal but the resting membrane potential is lower than normal. The findings indicate that the release and the postsynaptic responsiveness to acetylcholine are intact in Duchenne dystrophy.

Synaptic vesicle abnormality in familial infantile myasthenia.

In familial infantile myasthenia (FIM) the miniature end-plate potential (MEPP) amplitude is normal in rested muscle, but stimulation in vitro at 10 Hz decreases it abnormally, and neuromuscular transmission fails in a few minutes. In search of a morphologic correlate of the transmission failure, we analyzed the densities and diameters of synaptic vesicles in deep and superficial regions of nerve terminals in external intercostal muscles of three FIM patients and three nonweak controls before and after 10-Hz stimulation for 10 minutes. The densities of superficial or deep synaptic vesicles before or after stimulation in FIM were not significantly different from the corresponding control values. The diameters of superficial and deep synaptic vesicles before stimulation were significantly smaller in the three FIM patients than in the three controls. Stimulation in the FIM patients reduced the MEPP amplitude by 51 to 75%, but increased the vesicle diameter in two patients and did not change the vesicle diameter in one patient. Stimulation in the controls reduced the MEPP amplitude by only 16 to 34%, decreased the vesicle diameter in two, and did not change the vesicle diameter in one. Stimulation after treatment with 1 mg/dl hemicholinium markedly reduced the MEPP amplitude in the controls, had no further effect on the transmission defect in FIM, and had no consistent effect on vesicle diameter in FIM or controls. We conclude that synaptic vesicles are abnormally small in rested muscle in FIM, but vesicle size cannot be reliably correlated with the MEPP amplitude in FIM or controls.

Pain in peripheral neuropathy related to rate and kind of fiber degeneration.

In a series of 72 patients with disease of peripheral neurons, neuropathic painfulness of the foot was found to be related to the rate and kind of nerve fiber degeneration. Patients with acute breakdown of myelinated fibers (either by wallerian or axonal degeneration) tend to have pain more often and to a greater degree than do patients with more chronic forms of nerve fiber degeneration. Neuropathic painfulness was not found to be related simply to the ratio of remaining large and small fibers after nerve fiber degeneration. These studies do not fit the expectation of the proponents of the gate theory of pain.

Longitudinal study of neuropathic deficits and nerve conduction abnormalities in hereditary motor and sensory neuropathy type 1.

We measured neuropathic deficit (neurologic disability score [NDS]) and attributes of nerve conduction in hereditary motor and sensory neuropathy (HMSN 1) in cross-sectional evaluation of 69 patients and in longitudinal evaluation over approximately 15 years in 31 of them. Neuropathic deficit worsened by 0.6 NDS point per year in patients 5 to 14 years old at first evaluation, by 1.1 points in patients 15 to 39 years old, and by 0.9 point in patients 40 or more years old. Neuropathic deficit was greater in HMSN 1b (the disorder linked to Duffy) than in HMSN 1a (not linked to Duffy). Nerve conduction attributes changed significantly depending on attribute studied, age, and nerve. In patients evaluated serially, ulnar conduction velocity (CV) increased by a few meters per second in patients who were 5 to 14 or 15 to 39 years old at first examination, but decreased in patients who were older. In serial measurements, peroneal nerve amplitude decreased in all 3 age groups. We found an association between CV and amplitude or NDS at first and last examinations, suggesting an association between severity of the CV abnormality and neuropathic deficit. The severity of the CV abnormality in the young appears to predict later neurologic abnormality.

Muscle acetylcholine receptors complexed with autologous IgG reflect seropositivity but not necessarily in vivo binding.

The diagnosis of acquired myasthenia gravis (MG) in apparently seronegative individuals is aided by finding immunoglobulin complexed to acetylcholine receptors (AChR) and a reduction in the number of binding sites for alpha-bungarotoxin (alpha-BTx) in nerve-muscle biopsies. In this study, we found that anti-AChR antibodies in extracellular fluids can complex with cytoplasmic epitopes of AChR in the process of muscle extraction. When normal muscle was briefly exposed to antibodies (greater than or equal to 0.3 nmol/l) in the initial step of tissue homogenization (before detergent extraction), membranous AChR became complexed with IgG. This was so even with a nonmyasthenogenic monoclonal antibody specific for the alpha-subunit's presumptive cytoplasmic segment 366-389. We also found that antibodies reactive with AChR's alpha-BTx binding region can significantly lower apparent yields of alpha-BTx binding sites extracted from muscle. Thus, the finding of IgG complexed to AChR extracted from biopsied muscle does not necessarily reflect in vivo binding but, nevertheless, is a sensitive indicator of AChR seropositivity in patients suspected to have MG.

Assessment of nerve regeneration and adaptation after median nerve reconnection and digital neurovascular flap transfer.

Six patients with median nerve severance (five sutured) were studied after an interval of 1.8 to 35 years to assess residual neurologic deficit, misdirected axonal regrowth, and adaptation to faulty reinnervation. Mild motor impairment was confirmed by the smaller thenar muscle action potentials and isometric muscle twitches from supramaximal stimulation of the median nerve. Sensory impairment was supported by the increased thresholds of vibratory (p = 0.003) and touch-pressure (p = 0.004) detection thresholds of the pulp of index fingers and decreased amplitudes of sensory nerve action potentials. A tactile hemidigit localization test revealed that localization was not significantly different from that on the contralateral side but perceptual territory was increased. This increase is best explained by misdirected axon regrowth without CNS adaptation. Long-standing faculty tactile digit localization in neurovascular skin flaps from finger to thumb also was demonstrated--further evidence that CNS adaptation is imperfect when sensory nerves to digits are relocated.

Ultrastructural localization of the acetylcholine receptor in myasthenia gravis and in its experimental autoimmune model.

Peroxidase-conjugated alpha-bungarotoxin (P-BGT) was used for the ultrastructural localization of the acetylcholine receptor in end-plates in external intercostal muscles of four patients with myasthenia gravis, in forelimb digit extensor muscles of rats with advanced chronic experimental autoimmune myasthenia gravis, and in suitable human and rat controls. In control end-plates, the previously reported localization of acetylcholine receptor on the terminal expansions of the postsynaptic folds and, in traces, on the presynaptic membrane was confirmed. By contrast, in myasthenia gravis some postsynaptic regions bound no P-BGT; in other regions, the folds displayed only faint traces of the reaction product, or only some segments of the postsynaptic membrane showed the reaction product; finally, in some regions there was no apparent decrease in reaction product. In general, those postsynaptic regions showing the greatest decrease in P-BGT binding were also the simplest or showed the most degenerative changes, and the presynaptic staining was decreased in proportion to the decrease in the adjacent postsynaptic P-BGT binding. In the experimental animals, the abnormalities in the amount and distribution of the acetylcholine receptor were essentially like those in the more severely affected patients. Morphometric estimates of the postsynaptic acetylcholine receptor surface correlated well with the patients' clinical status and linearly with the miniature end-plate potential amplitude.

Passively transferred experimental autoimmune myasthenia gravis. Sequential and quantitative study of the motor end-plate fine structure and ultrastructural localization of immune complexes (IgG and C3), and of the acetylcholine receptor.

Experimental autoimmune myasthenia gravis (EAMG) was passively transferred with immunoglobulin from rats with chronic EAMG to normal recipients. IgG and C3 were localized on terminal expansions of junctional folds of end-plates by 6 hours. Segments of folds rich in acetylcholine receptor (AChR) and coated with IgG and C3 were shed into the synaptic space by 24 hours, resulting in AChR deficiency of the postsynaptic membrane. Many sensitized postsynaptic regions were destroyed by macrophages by day 2, but effective nerve-muscle contacts were reestablished by day 5. On day 10, end-plates were still structurally abnormal and showed AChR deficiency, but the animals were clinically recovered. On day 54, postsynaptic regions were still reduced in size, with slight reduction of postsynaptic AChR. Throughout the study, the miniature end-plate potential amplitude tended to vary directly with morphometric estimates of the abundance of the postsynaptic membrane reacting for AChR. Complement-mediated injury to the junctional folds and opsonization of the postsynaptic region can explain the morphologic changes. It is not yet known why phagocytic invasion of the end-plate occurs in acute EAMG and in passively transferred EAMG induced by chronic EAMG immuglobulin, but not in chronic EAMG and only rarely in the human disease.

Autoantibodies bind solubilized calcium channel-omega-conotoxin complexes from small cell lung carcinoma: a diagnostic aid for Lambert-Eaton myasthenic syndrome.

Serum autoantibodies found by radioimmunoassay in 27 of 52 patients with the Lambert-Eaton myasthenic syndrome (LES) bound specifically to a soluble omega-conotoxin binding component of a voltage-gated Ca2+ channel (VGCC) complex extracted from small cell lung carcinoma (SCC). These antibodies were not found in 43 control patients with other neurologic diseases, including myasthenia gravis, peripheral neuropathies, and amyotrophic lateral sclerosis, or in 9 patients with endocrine autoimmunity, but they were found in 2 of 21 control patients with SCC without a history of LES, 1 of whom had severe autonomic neuropathy. Seropositivity was more frequent in patients with LES who had evidence of a primary lung cancer (76%) than in those with other neoplasms or without evidence of cancer (30%). Antigens extracted from SCC tumor lines derived from patients with and without LES and from a human neuroblastoma line yielded results that were highly correlated. A control extract of colonic carcinoma (derived from a patient with LES) yielded negative results. The data implicate a tumor-associated VGCC as the autoimmunizing stimulus in a subset of patients with LES and provide the first direct evidence that the VGCC complex in SCC is a target for some LES antibodies. The serologic test described should be a useful aid in diagnosing LES.

Immune complexes (IgG and C3) at the motor end-plate in myasthenia gravis: ultrastructural and light microscopic localization and electrophysiologic correlations.

Although there is strong evidence that myasthenia gravis (MG) is caused by an autoimmune reaction to the nicotinic postsynaptic acetylcholine receptor (AChR) protein, immune complexes have never been directly demonstrated at the end-plate by immunocyto-chemistry or immunoelectron microscopy. Staphylococcal protein A (which binds to the Fc region of human IgG subclasses 1, 2, and 4) and rabbit anti-human C3 conjugated with peroxidase were used for the ultrastructural (5 patients) and light microscopic (12 patients) localization of IgG and C3, respectively, at MG end-plates. Both IgG and C3 were localized on segments of the postsynaptic membrance and fragments of degenerating junctional folds in the synaptic space. In nonmyasthenic control patients no immune complexes were evident at the end-plate. As judged by morphometric analysis of electron micrographs, the immune complexes were more abundant in the less severely affected MG patients than in the more severely affected ones. A linear correlation was demonstrated between the length of the postsynaptic membrance binding immune complexes and the amplitude of the miniature end-plate potential. The less intense reaction for immune complexes in the more severely affected MG patients can be attributed to the smaller quantity of AChR remaining at their end-plates. The findings provide unambiguous evidence for a destructive auto-immune reaction involving the postsynaptic membrance in MG. Immunopharmacologic blockade of AChR and IgG-induced modulation of AChR may also contribute to the AChR deficiency at the MG end-plates.

Linkage evidence for genetic heterogeneity among kinships with hereditary motor and sensory neuropathy, type I.

Previous reports have shown linkage of hereditary motor and sensory neuropathy, type I (HMSN I), a dominantly inherited hypertrophic neuropathy, to the locus for the Duffy blood group on the long arm of chromosome 1. Two kinships that were extensively studied and reported almost 20 years ago and used to show heterogeneity among kinships with peroneal muscular atrophy and to characterize HMSN I were investigated for linkage to various blood erythrocyte and lymphocyte (HLA) antigens. Strong evidence against linkage to the Duffy blood group locus was found for one kinship, whereas suggestive evidence for linkage was found for the other. These data imply that HMSN I is heterogeneous--that is, caused by different genetic mechanisms. The HMSN I that is not linked to the Duffy locus might be identified as HMSN IA, and the HMSN I that is linked to the Duffy locus might be designated as HMSN IB. HMSN IA was not linked to other blood types or HLA antigens. In addition, no evidence for linkage to blood types and HLA was found for spastic paraplegia with peroneal muscular atrophy and sensory loss (HMSN V).

Uremic neuropathy. III. Controlled study of restricted protein and fluid diet and infrequent hemodialysis versus conventional hemodialysis treatment.

The effect of a treatment schedule of restricted protein and fluid intake and of infrequent hemodialysis (schedule 1) as compared to a conventional hemodialysis treatment schedule (schedule 2) on the presence and severity of peripheral neuropathy has been studied in a small group of patients using a crossover design. Using the neurologic evaluation and quantitative assessment of cutaneous sensation and of nerve conduction as indices of peripheral nerve function, we have not demonstrated any worsening of peripheral nerve function from markedly curtailing the frequency of hemodialysis and modifying the diet. Because of the small size of the study, the preponderance of patients receiving the test treatment schedule first, and the possibility that nerve function slowly worsens with time, it is not possible to say with certainty that the test or the control treatment schedule might adversely affect peripheral nerve function. If such an effect is present it would appear to be small.