Free circulating ICAM-1 in serum and cerebrospinal fluid of HIV-1 infected patients correlate with TNF-α and blood-brain barrier damage

The mechanism for the initiation of blood-brain barrier damage and intrathecal inflammation in patients infected with the human immunodeficiency virus (HIV) is poorly understood. We have recently reported that tumour necrosis factor-α (TNF-α) mediates active neural inflammation and blood-brain barrier damage in HIV-1 infection. Stimulation of endothelial cells by TNF-α induces the expression of intercellular adhesion molecule-1 (ICAM-1), which is an important early marker of immune activation and response. We report herein for the first time the detection of high levels of free circulating ICAM-1 in serum and cerebrospinal fluid of patients with HIV-1 infection. Free circulating ICAM-1 in these patients correlated with TNF-α concentrations and with the degree of blood-brain barrier damage and were detected predominantly in patients with neurologic involvement. These findings have important implications for the understanding and investigation of the intrathecal inflammatory response in HIV-1 infection.


Introduction
Neurologic involvement is a frequent feature of human immunodeficiency virus (HIV) infection and may be the initial presentation in about 10% of patients. 2 There is increasing evidence that indirect mechanisms, such as the release of cytokines, play an important role in mediating brain inflammation in HIV infection. Indeed it is now widely acknowledged that tumour necrosis factor-(TNF-z), a central mediator of inflammation, plays a crucial role in the development of acquired immunodeficiency syndrome (AIDS). 3 TNF-z is implicated in the pathogenesis of most clinical and pathologic features of AIDS 3 and selectively kills HIV-infected cells, probably through a direct cytotoxic effect. Furthermore, TNFenhances the replication of HIV and induces the expression of a wide array of inflammatory cytokines.
Damage to cerebral endothelial cells and blood-brain barriers is another important pathologic feature that contributes to brain damage in HIV infection. Neuropathologic studies have clearly demonstrated that cerebral vasculitis is an early and frequent pathologic feature in HIV seropositive subjects. 4 Such vascular inflammation seems to be immune mediated 4's and could be connected to the release of inflammatory mediators.
Evidence has been presented that TNF-z induces inflammatory changes on human cerebral endothelial cells. 6'7 Moreover, we have recently shown in a different series of patients that TNF-o mediates blood-brain barrier damage in HIV infected patients. 8 However, the precise mechanism of the TNF<z mediated endothelial damage and the mechanism by which circulating immune cells recognize the HIV infected cells, including cerebral endothelium, are not well understood.
Adhesion of inflammatory cells to vascular endothelium is essential for. their migration into inflamed tissues. Endothelial cells lining the postcapillary venules and microcirculation elaborate several adhesion molecules both constitutively and in response to a wide range of inflammatory mediators. 9'1 Detection of adhesion molecules expression in vivo at sites of acute immunologic inflammation 1 has been used to infer that functional activation of the endothelium may be occurring at these sites. Intracellular adhesion molecule-1 (1CAM-l, CD54), a molecule bound to the ceil surface membrane, is an important early marker of immune activation and response 12'13 including the production of inflammatory vascular injury in io. TM Although the expression of ICAM-1, which may confer adhesivity for lymphocytes is most common in haematopoietic tissues, it is also detected in several organs, including the central nervous system (CNS). 1 The presence of free circulating ICAM-1 (clCAM-1) has been recently documented in human sera 16'17 and the expression of ICAM-1 is upregulated by TNFand other cytokines. 18 There is also increasing evidence that adhesion receptors can play a significant role in the pathogenesis of AIDS (reviewed by Koopman and Palsl9).
In this study, serum and CSF samples from HIV infected patients have been examined for the presence of free clCAM-1, as an in vivo marker of endothelial cell activation. It is believed this is the first time that circulating ICAM-1 has been shown to be present in serum and CSF from HIV infected patients and that it correlates with both TNFlevels and the degree of blood-brain barrier damage.

Patients and Methods
Patients: Paired CSF and serum samples were obtained from 37 HIV type-1 (HIV-1) seropositive patients (28 males, 9 females; median age 27.6 years, age range 19-49 years). Patients were classified according to the guidelines of the Centres for Disease control 2 and their clinical features are presented in Table 1.
Cerebrospinal fluid was obtained by a lumbar puncture and cells were separated by cytocentrifugation then all samples were filtered through a 0.45 #m disposable sterile filter (Millipore, Harrow, UK) to remove contaminating particulate materials. The CSF was concentrated by means of Minicon CS15 concentrators (Amicon, Upper Mill, UK). Samples were frozen in aliquots at --70C and thawed just before use; repeated Controls: Control CSF and serum samples were obtained from 30 HIV-1 seronegative patients with various non-inflammatory neurologic diseases in whom signs of blood-brain barrier damage were detected at presentation. They included eight patients with meningioma, four with craniopharyngioma, five with intracranial arteriovenous malformation, six with cerebrovascular diseases, three with benign intracranial hypertension, and four with obstructive hydrocephalus. Paired samples were also obtained from 18 normal subjects (median age 32.5 years, range 16-54 years) to determine reference ranges. These subjects presented with nonspecific complaints such as headache or blurring of vision and neurologic examination as well as detailed investigations had excluded a specific cause of their symptoms.
Detection of cICAM-I: Measurement of circulating immunoreactive ICAM-1 in serum and CSF samples was performed by a sensitive dot blot analysis 16 with minor modifications. In brief, 1:50 diluted serum and CSF samples concentrated to the same albumin level were spotted onto polyvinyl difluoride membrane (Immobilon, Millipore, Harrow). After nonspecific blocking and washing, the blots were incubated with a monoclonal antibody to ICAM-121 and subsequently incubated with a peroxidase-conjugated F(ab')2 fragment (Sigma, St Louis, MO, USA). The blots were then developed with a chromogen solution containing 0-phenylenediamine dihydrochloride and 100#1 H202 in 100 ml of 0.02 M acetate buffer. As controls, blots were incubated without the first antibody and with a Mab (anti-CD18) against the common beta chain of LFA-1 and Mac-1.
Quantitative analysis of the blots was performed by densitometric evaluation as recently described = using a J oyce Loebl Chromoscan 3 densitometric scanner. The intensity of the colour reaction of blots is dependent on the ICAM-1 concentration in the test samples. The colour intensity was measured as reflected light and the maximum absorption was determined at 492 nm. The lower limit of the detection of clCAM-1 is 8 ng/ml. Other assays: Levels of TNF-0 in the test samples were determined by a sandwich-type enzyme-linked immunoassay (ELISA) described previously. 23  tegrity of the blood-brain barrier was assessed by calculating CSF to serum albumin quotient (Qalb) 24 and the degree of barrier damage was graduated according to Qalb as already described, as

Results
Distribution of cICAM-A variable range of intensities of free clCAM-1 was detected in serum from the control healthy individuals ( Figure 1); however, no cICAM-1 reactivity was seen in their CSF. The mean 4-2 SD of maximum absorbance of clCAM-1 in serum from healthy controls (19.7 + 14.6%) was calculated as the cut-off value for determining abnormally high amounts of clCAM-1 in the study population.
High clCAM-1 levels (above the cut-off value in normal controls) were detected in serum of 25 HIV-1 seropositive patients and 13 seronegative controls, whereas cICAM-1 in CSF was seen predominantly in HIV-1 serpositive pttients (Table 2). Serum cICAM-1 in HIV-1 seropositive patients correlated with corresponding CSF amounts (r 0.61, p < 0.001). Free clCAM-1 was detected in CSF of all HIV-1 seropositive patients with neurologic involvement but was not detected in CDC groups II or Ili; however, cICAM-1 was also detected in CSF of group IV-A patients who showed no clinical evidence of neurologic disease.
Correlation of cICAM-I with TNFlevels: As shown in Table 2, high TNF-0 levels were seen in serum of 20 and in CSF of 22 HIV-1 seropositive patients. There was a strong association between TNF-0 and clCAM-1 in the test samples. Of the 22 seropositive patients who had high CSF TNFlevels, 17 (77%) had high CSF cICAM-1 levels, whereas only two (13%) of 15 patients with no detectable TNFin CSF had high CSF cICAM-1 levels. Similarly, 18 (90%) of the 20 patients with high serum levels of TNF-0 had high levels of serum clCAM-1, and only two (10%) patients with high serum TNF-0 levels had normal serum clCAM-1 levels.
Correlation of dCAM-1 with blood-brain barrier damage: Twenty-four HIV-1 seropositive patients had high Qalb values suggestive of blood-brain barrier damage. All HIV-1 seropositive patients who had  (Figure 4). In contrast, the degree of blood-brain barrier damage in seronegative controls did not correlate with either serum or CSF levels of cICAM-1.

Discussion
In this study of HIV-1 seropositive patients who have relatively high levels of clCAM-1 in serum and CSF, a correlation with both serum and CSF TNFlevels and the degree of blood-brain barrier damage was observed. It is thought that this represents the CSF TNF-a (Unit/ml) The presence of high levels of free circulating ICAM-1 in patients with HIV-1 infections also raises the question of whether this molecule carries out biological functions similar to its cell surface bound counterpart, such as the regulation of the intercellular adhesion process. Recent evidence 17 suggests that cICAM-1 retains almost all extracellular domains of membrane ICAM-1 and most structural features necessary for binding to the adhesion receptor lymphocyte function associated antigen 1. Thus, further characterization of cICAM-1 may provide insight into the pathophysiology of inflammatory CNS involvement during HIV-1 infection, particularly the impairment of the blood-brain barrier.
The blood-brain barrier, which is formed by specialized endothelial cells, regulates the interaction between the immune and the central nervous systems. In the normal brain there is very limited lymphocyte traffic, but lymphocyte infiltration is critical for the pathogenesis of several brain diseases. 26'27 A crucial early step in mounting an effective inflammatory or immune response is the promotion of leucocyte adhesion to the vascular endothelium before they can migrate chemotactically to the appropriate microenvironment. 9 1CAM-1 is highly inducible on various cell types, particularly endothelial cells, during inflammation and in response to proinflammatory cytokines, which suggests that 1CAM-1 upregulation and cell surface expression are important in the regulation of 28 immune responses.
It has been reported recently that the expression of 1CAM-1 is important in leucocyte homing and adhesion to the blood-brain barrier during active stages of experimental autoimmune demyelination. 29 Similarly, the expression of adhesion molecules at the blood-brain barrier has been demonstrated recently in humans. 3'3. Since neurologic involvement in HIV-1 infection is commonly associated with cerebral endothelial and blood-brain barrier damage, 32-34 which can be mediated by TNF-a, 8 it is reasonable to suggest that 1CAM-1 may act as a homing signal in HIV-1 related CNS inflammation. In addition, the adhesive interactions between inflammatory cells and the functional adhesion molecules expressed on endothelial cells may result in the activation of inflammatory cells prior to their migration into the intrathecal compartment. The lack of correlation between clCAM-1 and non-inflammatory blood-brain barrier damage observed in the seronegative controls further underlies the potential importance of 1CAM-1 in the regulation of CNS inflammation. Whether the release of cICAM-1 preceded or occurred during damage to the blood-brain barrier is not yet clear but will be the subject of further studies.
Free cICAM-1 was detected in CSF of all HIV-1 seropositive patients with neurologic involvement but was absent from CSF of patients with early HIV-1 infection (i.e. CDC groups II and III) who had no evidence of neurologic disease. In this regard, cICAM-1 could be a useful marker of neurologic involvement in patients with HIV-1 infection. The presence of clCAM-1 in CSF of patients with constitutional disease (CDC group IV-A), however, suggests that the intrathecal release of this molecule may precede the onset of clinically manifested neurologic disease.
The correlation between cICAM-1 and levels of TNFpresented here extends earlier in vitro observations, which reported an increased expression of 1CAM-1 on brain microvascular endothelial cells after activation of the cell with TNFor other proinflammatory cytokines in a dose-dependent manner. 3s Further evidence was presented recently that the TNFmediated increase in vascular permeability and oedema are instituted by 1CAM-Idependent mechanisms36--a fact that may explain the significant correlation of cICAM-1 with both TNFlevels and blood-brain barrier damage in our patients.
The importance of our results is also suggested by recent in vivo studies, which detected abundant expression of ICAM-1 on brain endothelial cells in inflammatory/infectious CNS diseases, such as herpes simplex encephalitis and in active multiple sclerosis plaques. 5 Interestingly, ICAM-1 is poorly expressed by CNS endothelial cells in animals without an intrathecal inflammatory process, 37 but ICAM-1 expression is upregulated during the initial phase of chronic relapsing allergic encephalomyelitis in association with inflammatory cell invasion. 9 In conclusion, these results suggest that cICAM-1 may be important in the pathogenesis of inflammatory changes within the intrathecal compartment of HIV-1 infected patients. The correlation of cICAM-1 with both TNFconcentrations and the degree of blood-brain barrier damage supports a functional role. It has to be noted, however, that the molecular events regulating cellular migration to the intrathecal compartment are multifactorial and involve several adhesion molecules in addition to ICAM-1. Analysing the dynamics of these molecules is crucial to evaluate the inflammatory cell invasion of CNS in HIV-1 infection.