1. Vaccine Research Center, NIAID, NIH, Maryland 20892, USA
2. Laboratory of Immunoregulation, NIAID, NIH, Maryland 20892, USA
3. Department of Experimental Transplantation and Immunology, Medicine Branch, NCI, NIH, Maryland 20892, USA
4. Department of Medicine, University of Texas Southwestern Medical Center, Dallas, Texas 75390, USA
5. Department of Infectious Diseases, Northwestern University Medical School, Chicago, Illinois 60611, USA
6. Laboratory of Immunology, NIAID, NIH, Maryland, USA, and Department of Physiology and Pharmacology, Tel Aviv University, Tel Aviv 69978, Israel
7. Nuffield Department of Clinical Medicine, John Radcliffe Hospital, Oxford OX3 9DU, UK
8. These authors contributed equally to this work

Correspondence to: Daniel C. Douek^1,2,3 Correspondence and requests for materials should be addressed to D.C.D. (e-mail: Email: ddouek@mail.nih.gov).

HIV infection is associated with the progressive loss of CD4⁺ T cells through their destruction or decreased production^{1, 2}. A central, yet unresolved issue of HIV disease is the mechanism for this loss, and in particular whether HIV-specific CD4⁺ T cells are preferentially affected^{3, 4, 5}. Here we show that HIV-specific memory CD4⁺ T cells in infected individuals contain more HIV viral DNA than other memory CD4⁺ T cells, at all stages of HIV disease. Additionally, following viral rebound during interruption of antiretroviral therapy, the frequency of HIV viral DNA in the HIV-specific pool of memory CD4⁺ T cells increases to a greater extent than in memory CD4⁺ T cells of other specificities. These findings show that HIV-specific CD4⁺ T cells are preferentially infected by HIV in vivo. This provides a potential mechanism to explain the loss of HIV-specific CD4⁺ T-cell responses, and consequently the loss of immunological control of HIV replication⁶. Furthermore, the phenomenon of HIV specifically infecting the very cells that respond to it adds a cautionary note to the practice of structured therapy interruption.