First evidence of HIV infection of CD8 lymphocytes expressing CD4 during primary HIV-1 infection

Activated CD8 lymphocytes express CD4 generating a CD8brightCD4dim phenotype that is susceptible to HIV infection both in vitro[1–3] and in vivo[4]. Here we demonstrate for the first time that CD8 lymphocytes in a patient with primary HIV infection express CD4 and are infected with HIV. This finding has implications for the immune control of HIV during primary infection, and for the generation of a long-lived viral reservoir. The patient presented with a typical HIV seroconversion illness, consisting of fever, rash and lymphadenopathy approximately 4 weeks after HIV exposure through unprotected heterosexual intercourse. Primary HIV infection was confirmed by two weakly reactive HIV antibody tests with a negative Western blot (fourth generation HIV combo assay, Abbott Diagnostics, Abbott Park, Illinois, USA; fourth generation Vidas Duo assay, BioMerieux, Boulogne, France; and New LAV Blot 1, BioRad, Hercules, California, USA). HIV p24 antigen was positive (Vidas p24 antigen assay, BioMerieux) and could be neutralized with specific antiserum. The initial HIV viral load was 3.5 × 106 RNA copies/ml serum (Roche Cobas Monitor HIV-RNA assay, Roche Diagnostics, Mannheim, Germany) rising to 6.3 × 106 a week later, then dropping to 2 × 104 after one month. Blood was drawn for this study during the phase of rising viraemia, 3 weeks after the onset of symptoms. The prevalence of CD8brightCD4dim lymphocytes circulating in peripheral blood was determined on a FACSVantage flow cytometer (Becton Dickenson, Franklin Lake, New Jersey, USA) using fluorescent conjugated monoclonal antibodies: CD4-FITC (Molecular Probes, Paisley, UK), CD8beta-PE (Coulter Immunotech, Beckman Coulter, Inc. Miami, Florida) and CD3-cychrome (BD Biosciences). Of the total T lymphocytes (defined as CD3 expressing cells with forward and side scatter properties typical of lymphocytes) 26% were CD4 lymphocytes, 67% were CD8 lymphocytes, and 0.4% were CD8brightCD4dim lymphocytes (Fig. 1a). This frequency of CD8brightCD4dim lymphocytes was similar to levels found previously both in the general population [5] and in individuals with chronic HIV infection [6]. This single subject thus demonstrated CD4 cell expression on CD8 lymphocytes, but does not suggest that primary HIV infection leads to the exceptionally high levels of expression described for primary Epstein–Barr virus and cytomegalovirus infection [7].Fig. 1: The extent of CD4 expression on CD8 lymphocytes and the level of HIV infection of these cells are shown for the study subject with primary HIV infection. (a) Expression of CD4 and CD8β on T lymphocytes (gated on light scatter characteristics and CD3 expression from peripheral blood mononuclear cells). The proportions of CD4 lymphocytes (P1), CD8 lympocytes not expressing CD4 (P2) and CD8brightCD4dim lymphocytes (P3) are shown. (b) HIV DNA load attributable to CD4 lymphocytes, CD8brightCD4dim lymphocytes and CD8+CD4− lymphocytes. Filled bars represent detected attributable HIV DNA loads, the unfilled bar represents undetectable viral DNA, and half the lower limit of detection is given. Error bars show standard errors estimated using the program ‘Quality’ [9].The level of HIV infection of CD8brightCD4dim and CD8+CD4− negative lymphocytes was determined using a method previously validated in chronic HIV infection [4]. Briefly, highly purified populations of CD4 lymphocytes, CD8 lymphocytes expressing CD4 (CD8βbrightCD4dim) and CD8 lymphocytes not expressing CD4 (CD8βbrightCD4−) were isolated. The HIV-DNA copies per million cells in each of the isolated populations were determined by the extraction of DNA followed by quantitative polymerase chain reaction for HIV long terminal repeat (LTR) repeat DNA. The HIV DNA load in CD4 lymphocytes was 697 (SE 225) HIV DNA copies per million CD4 lymphocytes, a frequency in keeping with results from a previous series of 16 subjects with primary HIV infection [8]. Interestingly, the level of infection of CD8brightCD4dim lymphocytes was similar to that of CD4 lymphocytes at 523 (SE 160) HIV DNA copies per million CD8brightCD4dim lymphocytes, with only two HIV DNA copies per million cells being attributable to CD4 lymphocyte contamination. No HIV DNA was found in the CD8 lymphocytes not expressing CD4 (Fig. 1b) [9]. The frequency of infection of CD4 lymphocytes circulating in the blood during primary HIV infection substantially underrepresents the level of HIV infection occurring in lymph nodes and mucosal lymphoid tissue [10]. The same is likely to be true of CD8 lymphocytes, particularly as the susceptible (CD8brightCD4dim) phenotype is generated on antigen-specific activation in lymph tissue. Therefore, the proportion of circulating CD8 lymphocytes infected with HIV observed in this study may reflect much greater frequencies of infection in lymph tissue. Events during primary HIV infection have a major influence on the course of disease, with poor viraemic control being prognostic of rapid progression [11]. CD8 lymphocytes are vital in the control of initial viraemia [12,13], and thus any compromise of CD8 lymphocyte function at this stage could have a profound effect on prognosis. HIV infection of CD8brightCD4dim lymphocytes leads to the downregulation of CD4 expression, the deregulation of IFN-γ and FAS ligand expression, and decreased effector function [14]. The infection of CD8brightCD4dim lymphocytes demonstrated here could thus significantly compromise the primary anti-HIV CD8 lymphocyte response. In addition, the infection of CD8 lymphocytes has implications for the generation of a long-lived HIV reservoir. Unlike CD4 lymphocytes where activation is rapidly followed by apoptosis, a significant proportion of activated CD8 lymphocytes differentiate into memory cells [15]. While resting, these memory cells are invisible to immune surveillance, can form a long-lived refuge for HIV [16], and therefore could contribute to the failure of antiretroviral therapy to achieve complete HIV clearance, even when treatment is started soon after seroconversion [17]. Acknowledgements The authors would like to thank S. Cameron (Specialist Virology Centre, Gartnavel General Hospital, Glasgow, UK) for laboratory confirmation of primary HIV infection. Sponsorship: This work was funded by the Medical Research Council, UK, and the North Glasgow University Hospitals Trust.