We have conducted an immunocytochemical analysis to investigate the presence of the recently described vascular cell adhesion molecule-1 (VCAM-1) in human kidney, using the anti-VCAM-1 monoclonal antibody 1.4C3. In normal control tissue VCAM-1 was present on some (but not all) parietal epithelial cells lining Bowman's capsule. Forty-nine of fifty clinical biopsy specimens were characterised by the additional presence of VCAM-1 on proximal tubular cells. This was most marked in biopsies of patients with interstitial nephritis or systemic vasculitis with crescentic nephritis, but was also observed in biopsies with minimal change, IgA or lupus nephropathy, or from patients with diabetic nephropathy, amyloid, or gout. Proximal tubule VCAM-1 correlated significantly with the number of transferrin-receptor-positive leukocytes (r = 0.607, p less than 0.0001) in the interstitium, but not with expression of HLA-DR by tubular cells. Surprisingly, VCAM-1 was not observed on vascular endothelial cells in these biopsies, even in the presence of a marked infiltrate; this contrasts with other tissues (e.g. skin and synovium). The presence of VCAM-1 on tubular cells in the inflamed kidney indicates the potential for these cells to interact with mononuclear cells, either as accessory cells or as cytotoxic targets. The unexpected absence of VCAM-1 in renal vascular endothelial cells suggests local differences in the endothelial cells of this organ.

These data indicate that lipofection is an efficient method to transfer therapeutic genes to the corneal epithelium, and that it can be used to transfer constructs that utilize an inducible promoter controlled by TNF. As TNF is present in the aqueous humor during allograft rejection, and this is in contact with the corneal endothelium, this has the potential to restrict expression of a therapeutic gene to rejection episodes in the cornea.

SUMMARY Fixed torque devices were used to measure joint mobility at three sites in 364 adolescents including 39 families with at least two siblings. Increasing age and being male were associated with reduced laxity, and a strong effect of family was observed in the 39 sibling sets studied. The population included 47 Asians whose mobility was similar to that of the non-Asians, though the female/male difference was apparently greater in the former group. None of the above conclusions apply to index finger hyperextension, and it is apparent that genetic and constitutional factors only affect mobility at some sites. Such observations could lead to a review of current scoring systems for clinical hypermobility. classes at a local school and two polytechnics. The polytechnic classes were added to extend the age range of the sample. The school had recently become coeducational and this combined with the differential self selection of individuals to study biology resulted in there being only 97 (26-6%) female compared with 263 (72-3%) male students (in four cases the sex was not recorded). Ninety five per cent of all subjects were aged between 11 and 20, though three were above 25 years. There were 47 children whose parents were either of Indian or Pakistani origin in the population studied; the remainder (excluding one with no record of ethnic origin) were European. In 39 instances there was more than one individual from the same family-37 sibling pairs and two trios.Joint mobility was measured for the following movements: index finger hyperextension, forearm rotation, and lower limb rotation. The last two represent composite movements and thus reflect mobility at more than one joint. Measurements were undertaken using fixed torque measuring devices first described by Jobbins et al.9 The particular devices used in this study have low levels of both within and between observer variability as previously described,10 making them appropriate for use in epidemiological studies. The method of measurement in this population has beerr reported elsewhere.2 In each class three trained observers were used, one for each measurement. During the course of the study five' individual observers were used. The constraint of timetabling resulted in some individuals not being tested for all three movements. 209

Patients and methodsThe study population consisted of 364 individuals gathered from a local mixed secondary school and from the biological sciences departments of two polytechnics. There were 266 males and 98 females with an age range of 10 to 24 years. The inclusion of individuals was based solely on their attendance at designated biology classes.The range of the following joint movements was measured: index finger hyperextension, forearm rotation (total range-pronation plus supination), and lower limb rotation (total range-external plus internal rotation). The techniques used for the fixed torque measurement of these movements and the torques used have been described elsewhere.8 All measurements were taken from the subjects' right side to maintain standard conditions. It has previously been described that dominance probably does not affect mobility range.9 A single observer was used where possible for each of the three movements measured, though the observers used were not identical for each of the school classes of students measured. In all five observers were used.The interobserver variation for these devices is low,8and thus for analysis the results were pooled from all observers. The constraints of time tabling resulted in some individual subjects not being tested for all three movements. Thus index finger hyperextension was measured in 346, forearm rotation in 260, and lower limb rotation in 277 individuals; with 217 of 27 ResultsThe frequency distributions for the three movements measured were Gaussian or near Gaussian, and the main features of the distributions are summarised in Table 1. All three movements are characterised by a wide variation in this normal population. Only two individuals of the 217 who had measurements at each of the three sites were in the top 10% in all of the three distributions, but this surprisingly small number was higher than the 0*24 (p

Using an endothelial/epithelial hybrid cell line, three different non-HLA antibody types have been identified by flow cytometry in patients who have rapidly rejected multiple renal allografts. These antibodies may be classified as anti-endothelial-monocyte, anti-activated endothelial cell, or anti-epithelial cell.