522 NATURE MEDICINE • VOLUME 8 • NUMBER 5 • MAY 2002

ARTICLES

HIV-associated nephropathy is a clinicopathologic entity thatincludes proteinuria, focal segmental glomerulosclerosis oftenof the collapsing variant, and microcystic tubulointerstitial dis-ease1–4. Increasing evidence supports a role for HIV-1 infectionof renal epithelium in the pathogenesis of HIV-associatednephropathy5–8. Using in situ hybridization, we previouslydemonstrated HIV-1 gag and nef mRNA in renal epithelial cellsof patients with HIV-associated nephropathy9. Here, to investi-gate whether renal epithelial cells were productively infectedby HIV-1, we examined renal tissue for the presence of HIV-1DNA and mRNA by in situ hybridization and PCR, and we mole-cularly characterized the HIV-1 quasispecies in the renal com-partment. Infected renal epithelial cells were removed bylaser-capture microdissection from biopsies of two patients,DNA was extracted, and HIV-1 V3-loop or gp120-envelope se-quences were amplified from individually dissected cells bynested PCR. Phylogenetic analysis of kidney-derived sequencesas well as corresponding sequences from peripheral bloodmononuclear cells of the same patients revealed evidence oftissue-specific viral evolution. In phylogenetic trees constructedfrom V3 and gp120 sequences, kidney-derived sequencesformed tissue-specific subclusters within the radiation of bloodmononuclear cell–derived viral sequences from both patients.These data, along with the detection of HIV-1-specific proviralDNA and mRNA in tubular epithelium cells, argue strongly forlocalized replication of HIV-1 in the kidney and the existence ofa renal viral reservoir.

Renal biopsies from two patients with typical pathological fea-tures of HIV-associated nephropathy (HIVAN) were used to in-vestigate the renal HIV-1 populations. At the time of the biopsy(3 April 1998), patient RB23 had a CD4 count of 290 cells per ml,plasma viral RNA level of 400 copies per ml and a serum creati-nine of 3.9 mg/dl. Patient RB5 (11 June 1997) had a CD4 countof 640 and a plasma viral RNA level of 1.0 × 103 copies per ml. Asis typical with HIVAN, both patients were African Americans10–12.

The presence of HIV-1-specific mRNA and proviral DNA inrenal tissue was first confirmed by in situ hybridization and insitu PCR. HIV-1 gag mRNA was detected in tubular epithelial cellsby in situ hybridization (Fig. 1a; sense-negative control is shownin 1b) and HIV-1 env DNA was detected by in situ PCR (Fig. 1c).

Note that the pattern of cells positive for in situ PCR (Fig. 1c) isvirtually identical to the pattern of cells positive for mRNA (Fig.1a). At higher magnification, numerous infected tubular cells (asdetermined by in situ PCR) were readily apparent (Fig. 1e).Glomerular podocytes were also found positive for HIV-1 DNA(Fig. 1f). As a positive control, in situ PCR was performed usingprimers specific for human β-globin gene. All nuclei were posi-tive in both tubular and glomerular epithelium (Fig. 1g and h).As a negative control, HIV-1 primers that were used to detectHIV-1 proviral DNA (Fig. 1c, e and f) were used to probe a kidneybiopsy from an HIV-1-uninfected patient (Fig. 1d). In summary,both in situ PCR and in situ hybridization confirmed the pres-ence of HIV-1 nucleic acids in the renal epithelial cells in both ofthe patients, with focal areas of infection throughout the renalepithelium.

To genetically characterize the HIV-1 quasispecies in the kid-ney, we used laser capture microdissection (LCM) to remove in-fected tubular cells without inadvertent admixture of adjacentinterstitial cells, including HIV-1-infected tissue macrophages.Due to the focal nature of the infection, serial sections were firstanalyzed by in situ PCR to identify infected tubules. Individualtubule dissections were placed into different tubes, high-molecu-lar-weight DNA was extracted, and PCR was performed such thateach amplification product represented viral-envelope se-quences from a single dissected tubule. Two different sets of en-velope primers were used, one amplifying a 200-basepair (bp)fragment encompassing the V3-loop region and a second ampli-fying a 1,700-bp fragment encompassing the entire gp120 do-main. The same primers were also used to amplify envelopesequences from peripheral blood mononuclear cell (PBMC) DNAobtained from the same patients at the time of the renal biopsy.

As an additional precaution, PCR analysis was performed toevaluate the presence of T-cell receptor (TCR) rearrangements.We reasoned that in microdissections that included interstitialinfiltrating cells, T cells and their signature TCR rearrangementscould be identified. In contrast, microdissections of pure epithe-lial cells should not exhibit evidence of TCR rearrangements.Thus, the absence of TCR rearrangements in combination with apositive interstitial control served to exclude nonepithelial conta-mination. PCR was performed to amplify rearranged TCR-γ vari-able joining (VJ) junctions, followed by heteroduplex and PAGE

Replication and compartmentalization of HIV-1 in kidneyepithelium of patients with HIV-associated nephropathy

DANIELE MARRAS1, LESLIE A. BRUGGEMAN1, FENG GAO6, NOZOMU TANJI4, MAHESH M. MANSUKHANI4, ANDREA CARA7, MICHAEL D. ROSS1, G. LUCA GUSELLA1,

GARY BENSON3, VIVETTE D. D’AGATI4, BEATRICE H. HAHN5, MARY E. KLOTMAN2

& PAUL E. KLOTMAN1

Divisions of 1Nephrology, 2Infectious Diseases, Department of Medicine, and 3Department of BiomathematicalSciences, Mount Sinai School of Medicine, New York, New York, USA

4Department of Pathology, Columbia Presbyterian Medical Center, New York, New York, USA5Departments of Medicine and Microbiology, University of Alabama at Birmingham, Birmingham, Alabama, USA

6Department of Medicine, Duke University Medical Center, Durham, North Carolina, USA7Istituto Superiore di Sanita, Rome, Italy

Correspondence should be addressed to D.M.; email: daniele.marras@mssm.edu

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://m

edic

ine.

nat

ure

.co

m

NATURE MEDICINE • VOLUME 8 • NUMBER 5 • MAY 2002 523

ARTICLES

analysis. The results are shown for patient RB5 in Fig. 2. DNAfrom a bulk tissue section (Fig. 2a) was analyzed in parallel withDNA extracted from microdissected renal tubular epithelial cellsfrom the same tissue sample (Fig. 2b). The specific amplicons pre-sent in lane 2 show the presence of a polyclonal product of the VJjunction of the first Vγ family. In contrast, no such rearrange-ments are detected in microdissected epithelium (Fig. 2b). As apositive control for the PCR, a non-rearranged gene, IFNAR2 (Fig.2, lane 5), was readily amplified. These results demonstrated thatthe microdissected tubular epithelial cells were not contaminatedwith T cells from the interstitial compartment.

To determine whether there was evidence for kidney-specificviral evolution in patients with HIVAN, PCR-derived V3 andgp120 amplification products were cloned and one representa-tive clone sequenced. All kidney-derived sequences representedsingle amplification products from individual microdissectedtubules. RB23 sequences were derived from tubules collectedfrom 15 different locations within the renal biopsy, thus provid-ing a broad representation of infected tubules within the speci-

men. However, not all microdissected tubules yielded sufficientamounts of DNA for subsequent PCR amplification. Thus, fewerkidney-derived HIV-1 V3 or gp120 sequences were obtained foreach patient than corresponding sequences from the blood.

To determine the evolutionary relationships of the kidney-de-rived sequences to those from the blood of the same patient andto reference sequences from the database, phylogenetic treeswere constructed from V3 nucleotide (Fig. 3a) and gp120 amino-acid (aa) (Fig. 3b) sequences. As expected, sequences from bothpatients grouped with HIV-1 (group M) subtype B reference se-quences, the predominant subtype in the United States.Moreover, sequences from each patient clustered together in ahighly statistically significant manner (supported by 100% bootstrap samples). However, within the quasispecies of each patient, kidney and PBMC-derived sequences were not inter-spersed. Instead, the viral sequences from the kidney formed tis-sue-specific subclusters within the radiation of viral sequencesfrom the PBMCs1. This was most obvious in the gp120 treewhere RB5 and RB23 kidney sequences clustered together with97% and 95% bootstrap values, respectively (Fig. 3b). Tissue-spe-cific grouping was also apparent in the V3 tree, which containedadditional kidney-derived sequences, although the bootstrap

a b c

e f g h

d

Fig. 1 Detection of viral nucleic acids in renal biopsies. a and b, Detectionof viral mRNA in HIVAN biopsies by in situ hybridization. Antisense riboprobehybridization demonstrates positive tubular epithelial cells (a). A sense ribo-probe was used as a negative hybridization control in serial section (b). c–f, Detection of proviral DNA in renal biopsies by in situ DNA PCR. In a ser-ial section to a, HIV-1 proviral DNA is shown (c; magification, ×60). Higher

magnification (×120) demonstrated proviral DNA in renal tubular (e) andglomerular (f) epithelial cells. Arrow in e indicates an infected interstitial cell,likely an infiltrating lymphocyte. The same conditions were used in an in situPCR for proviral DNA on a renal specimen from an HIV-seronegative subjectand no amplification of HIV-1 DNA was observed at ×60 magnification (d).In situ DNA PCR for β-globin is shown as a positive control (g and h; ×100).

a

b

Fig. 2 Analysis of the T-cell receptor rearrangement in microdissected tissue.a, Analysis of the whole kidney section from patient RB5. Lane 1, molecularweight markers; lanes 2 and 3, amplification with Vγ1 primers; lanes 4 and 5,Vγ9 primers; lanes 6 and 7, Vγ10/11 primers. In each case, reverse primersfor J1/2, JP1/P2 and JP were used. PCR products were electrophoresed bothdirectly (lanes 2, 4 and 6) and following heteroduplex analysis (lanes 3, 5and 7). The results revealed a polyclonal T-cell population with the Vγ1primer (lane 2; product size range, 271 and 281; arrow). Lane 8 is a positivecontrol with primers specific for IFNAR2 (product size range, 275–280; ar-rowhead). The visible product was necessary to verify that sufficient DNAwas isolated for each dissection to obtain a signal with amplification. b, Analysis of microdissected tubules from patient RB5. The entire PCR prod-uct was run without heteroduplexing to maximize sensitivity. Lanes 2, 3 and4 are amplifications using the TCR-γ reactions using primer pairs described ina (lanes 2, 4, 6, respectively); lane 5 is the IFNAR2 PCR control amplification.Only 10% of extracted DNA was used in this last reaction, whereas the re-maining 90% was equally divided for the three TCR-γ PCRs.

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://m

edic

ine.

nat

ure

.co

m

524 NATURE MEDICINE • VOLUME 8 • NUMBER 5 • MAY 2002

ARTICLES

values supporting the kidney clusters were not significant due tothe short sequence length (172 bp). Only one kidney-derived V3sequence (RB5-D.V3K) grouped outside the main kidney cluster.

Here we demonstrate the presence of HIV-1 proviral DNA andmRNA nucleic acids in kidney epithelial cells in patients withHIVAN. Moreover, we provide phylogenetic evidence for HIV-1compartmentalization in kidney tissue. These data provide com-pelling evidence for active replication of HIV-1 in kidney epithe-lium and indicate that HIV-1 strains residing in the kidneymicroenvironment are not in a state of unrestricted bidirectionalmovement with viruses circulating in the blood. This, in turn,suggests that the kidney may serve as a viral reservoir potentiallyharboring HIV-1 strains that have evolved under tissue-specificselection pressures.

The finding of kidney-specific HIV-1 variants is not a priori anindicator of tissue-specific selection. The kidney-specific cluster-ing of HIV-1 envelope sequences observed in Fig. 3 is best ex-plained by a stochastic event representing a founder effectfollowing the initial seeding of kidney epithelium by a blood-de-rived variant. Our data suggest that this happened only once inthe two patients analyzed. However, this conclusion is based ononly a limited number of sequences and only single kidney biop-sies. It is possible that analyses of additional tubules from se-quential biopsies from RB5 or RB23 would have yielded HIV-1sequences that would have formed clusters outside the kidney-specific lineages shown in Fig. 3. Indeed, the one divergent V3-

loop sequence from patient RB5 (RB5-D.V3K in Fig. 3b) may berepresentative of such a second, independent seeding event.However, even if this were the case, this would not argue againsta productive infection of tubular epithelial cells. The extent ofgenetic diversity observed among the kidney-derived HIV-1 se-quences in both RB5 and RB23 necessitates local replication.Moreover, the sampling of distant tubules from different parts ofthe biopsy specimen from patient RB23 indicates that viral repli-cation is not restricted to the immediate vicinity of any one in-fected tubular epithelial cell.

The mechanism for viral entry into renal epithelium remainsunknown. In addition to the CXCR4/CCR5 chemokine-recep-tors13, at least nine potential coreceptors for HIV-1 have beenidentified14–17. In renal epithelium, expression of CXCR4 andCCR5 is generally not detected18. Moreover, it remains unclearwhether CD4 is expressed on normal renal epithelial cells, al-though the presence of CD4 mRNA has been reported in cul-tured renal epithelial cells by RT-PCR (ref. 19). Passive transfer ofreceptors from one cell to another represents still another theo-retical possibility to confer susceptibility to infection in vitro20.Notably, CCR5+ intestinal epithelial cells have been found to se-lect and transfer R5 viruses to CCR5+ indicator cells in vitro, sup-porting a model for viral diffusion through the uppergastrointestinal tract21. Moreover, it has been hypothesized thatthe infection of the trophoblastic layer through fusion of mater-nal HIV-1-infected PBMCs and the subsequent viral transcytosis

a b

Fig. 3 Quasispecies complexity of kidney and PBMC-derived from 2 pa-tients with HIVAN. a, V3-loop sequence tree constructed from V3 nucleotidesequences (consensus length, 172 bp) using the neighbor-joining methodand rooted with HIV-1/NDK. b, gp120 sequence tree constructed from de-duced aa sequences (consensus length 414 aa) using the neighbor-joiningmethod and rooted with HIV-1/NDK. In both panels, horizontal branch

lengths are drawn to scale, with scale bars indicating 10% sequence diver-sity (vertical separation is for clarity only). Numbers at nodes indicate thepercentage of bootstrap values with which the adjacent cluster is supported(only values of 80% or greater are shown). Newly derived kidney (high-lighted) and PBMC sequences from RB5 and RB23 are shown, along with 20subtype B reference sequences from the Los Alamos Sequence Database.

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://m

edic

ine.

nat

ure

.co

m

NATURE MEDICINE • VOLUME 8 • NUMBER 5 • MAY 2002 525

ARTICLES

may have a role in vertical HIV-1 transmission in utero22. This lat-ter case may also represent a model for the initial infection ofrenal tubules in HIVAN, where leukocyte infiltration of the kid-ney may lead to a yet unidentified fusion event. However, trans-cytosis cannot explain cell-to-cell spread within the kidney,leading to the fully infected tubules frequently seen by in situPCR and in situ hybridization. The presence of closely related yetgenetically distinct HIV-1 variants in individual microdissectedtubular epithelial cells necessitate complete rounds of viral repli-cation, including fusion, uncoating and reverse transcription.Given that CD4 and CCR5 are not readily detectable, it is possi-ble that the HIV-1 strains that replicate in the kidney have ac-quired the ability to use alternate mechanisms to gain entry intoepithelial cells. Analysis of the receptor and coreceptor require-ments of HIV-1 chimeras with kidney-derived envelope genesmay shed new light on this question.

Finally, the success of combination antiretroviral therapy forHIV-1 infection has generated enormous interest in the mecha-nisms by which HIV-1 can persist in certain anatomic sites de-spite the presence of drugs that effectively inhibit critical stepsof the virus life cycle23,24. In this context, it is important to notethat we have recently demonstrated the persistence of HIV-1mRNA in the renal epithelium in a patient on HAART regimenwith no detectable plasma viral RNA (ref. 8). This raises the ques-tion to what extent kidney represents a previously unappreci-ated viral reservoir. It will be interesting to determine whetherdrug-resistant mutants are present in infected kidney epithelialcells and how their frequency and evolution compare to drug-re-sistant mutants circulating in the blood. It will also be importantto determine to what extent the kidney epithelium is susceptibleto available drugs and whether the relatively slow turnover ofepithelial cells affects drug efficacy3. The new approach of usingin situ PCR-guided laser microdissection of individual HIV-1-in-fected tubular cells described here should greatly facilitate theseimportant investigations.

MethodsRenal biopsies. Renal biopsies were obtained from patients RB5 and RB23who had been diagnosed with HIVAN by 2 independent pathologists. Ineach case, the patient’s consent was obtained and the studies were per-formed under an Institutional Review Board (IRB) approved protocol. Tissuesamples from kidney biopsies were formalin-fixed paraffin embedded, andcut sections (5 µM) were stained with H&E. To avoid cross-contamination,a new microtome blade was used each time a new case was sectioned. As acontrol, a renal biopsy from an HIV-seronegative African-American femalewas used. A single blood sample was also obtained at the time of the biopsyand PBMCs were isolated by Ficoll-hypaque centrifugation.

In situ hybridization. The in situ hybridization technique was performedunder non-denaturing conditions, as described, using a digoxigenin-anti-digoxigenin technique6. The gag probe comprised a 359-bp PCR fragmentfrom HXB2 (nucleotides 1031–1390) subcloned into pGEM-T Easy(Promega, Madison, Wisconsin). Digoxigenin riboprobes were generatedas described using T7 and SP6 polymerases from NdeI and NcoI linearizedplasmids respectively9.

In situ PCR. Using the same tissue blocks, in situ PCR was used to localizeHIV DNA by a 3-step method as described9,25.

Laser-capture microdissection. The PixCell LCM System (ArcturusEngineering, Mountain View, California) was used for LCM of paraffin-em-bedded tissue26. Following the manufacturer’s protocol, we dissected singletubules, transferred them to a polymer film that was activated by laserpulses under direct microscopic visualization, and added the cells directlyto lysis buffers.

T-cell receptor rearrangement. DNA was amplified in 4 separate reac-tions, including 3 semi-multiplexed reactions for TCR-γ gene rearrange-ment, and one for interferon-α receptor 2 (IFNAR2)27. The forward primersfor the 3 TCR-γ reactions Vγ2–5, Vγ7 and Vγ8; Vγ9 Vγ10 and Vγ11 and the 3J primers specific for J1 and J2, JP1 and JP2, and for the JP region have beendescribed27. The entire amount of DNA extracted was used in the 4 reac-tions. Following a denaturation step of 94 °C for 5 min, PCR was performedfor 40 cycles, at 94 °C for 30 s, 55 °C for 45 s and 72 °C for 45 s. The prod-uct was run on a 4–20% gradient polyacrylamide minigel. A positiveIFNAR2 product of 230 bp indicated successful amplification of DNA. Thepresence of a polyclonal smear around the same size indicated the presenceof a polyclonal T-cell population. Note: most αβ T-cells have 1 or 2 re-arranged, non-functional TCR-γ alleles, allowing the use of this test to de-tect T cell DNA.

PCR amplification of HIV-1 envelope sequences. DNA from microdis-sected tubular epithelial cells was extracted following the protocol of theQIAamp tissue kit (QIAGEN, Valencia, California) and lyophilized. DNAfrom the donors’ PBMCs was extracted using the DNeasy Tissue Kit (QIA-GEN, Valencia, California). For the amplification of the V3 fragment, single-round PCR was performed in 50 µl of distilled water with 2 mM dNTPs, 2.5U of AmpliTaq and 1× PCR Buffer (Perkin Elmer, Branchburg, New Jersey)and 5 ng/µl of the following primers: 5′-TGTCCAAAGGTATCCTTTGAGC-CAATTCC-3′ and 5′-AGTAGAA-AAATTCCCCTCCACAATTAA-3′. 35 cycles ofPCR were performed as follows: denaturation at 94 °C for 60 s, annealing at68 °C for 30 s, and extension at 72 °C for 30 s. For amplification of full-length gp120 fragments, a nested PCR approach was used employing thefollowing primers: 5′-GACTAATAGAAA-GAGCAGAAGACAGTGGCA-3′ and5′-AGTGCTTCCTGCTGCTCCCAAGAACCC-3′ in the first round; 5′-ATGA-GAGTGAAGGAGAAATATCAGCAC-3′ and 5′-GAACAAAGCTCCTATTCC-CACTGCTCT-3′ in the nested round. The PCR was performed in a finalvolume of 50 µl, 2 mM dNTPs, 1.5 mM MgCl2 and 5 ng/µl of each primerand 2 µl of Elongase (Invitrogen, Carlsbad, California). 40 cycles of PCRwere performed as follows: 92 °C for 60 s, 55 °C for 3 min, and 72 °C for 3min. Amplification products were gel-purified, cloned into PGEM-T(Promega, Madison, Wisconsin) and sequenced using an automatic DNASequenator (model ABI-PRISM-377).

Phylogenetic analysis. The phylogenetic relationships of HIV-1 infectingkidney and PBMCs of patients RB5 and RB23 were estimated from com-parisons of V3 nucleotide and gp120 deduced aa sequences with the cor-responding sequences of previously reported subtype B referencesequences from the Los Alamos sequence database (http://hiv-web.lanl.gov/HTML/alignments.html). For RB5, 12 PBMC-derived gp120sequences were available as well as 3 kidney-derived gp120 and 4 kidney-derived V3-loop sequences. For RB23, 4 PBMC-derived gp120 sequencesand 2 PBMC-derived V3-loop sequences were available as well as 2 kid-ney-derived gp120 and 5 kidney-derived V3-loop sequences. Sequenceswere aligned using CLUSTAL W (ref. 28) and adjusted manually usingMASE (ref. 29). Sites with a gap in any of the sequences as well as areas ofuncertain alignment were excluded from further analyses. Evolutionarydistances were estimated using the Kimura two-parameter mode30.Phylogenetic trees were constructed using the neighbor-joining method31

and the reliability of topologies estimated by performing bootstrap analy-ses with 1,000 replicates. Bootstrap values ≥ 80% were considered signif-icant.

Nucleotide sequence accession numbers. All newly generated sequenceshave been deposited in GenBank: accession numbers AF401325–AF401331and AF401335–AF401338 (V3 loop); AF482965–AF482979, AF401048,AF401050, AF401051, AF401053–AF401055 (gp120).

AcknowledgmentsWe thank I. Gelman for critical review of the manuscript and W. Abbott forartwork preparation. This work was supported by grants from the NationalInstitute of Health P01 DK50795, P01 DK56492, N01 AI85338 andP20AI27767, and the National Science Foundation (to G.B.) CCR-0073081and DBI-0090789.

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://m

edic

ine.

nat

ure

.co

m

526 NATURE MEDICINE • VOLUME 8 • NUMBER 5 • MAY 2002

ARTICLES

Competing interests statementThe authors declare that they have no competing financial interests.

RECEIVED 12 FEBRUARY; ACCEPTED 25 MARCH 2002

1. Rao, T.K., Friedman, E.A. & Nicastri, A.D. The types of renal disease in the ac-quired immunodeficiency syndrome. N. Engl. J. Med. 316, 1062–1068 (1987).

2. Strauss, J. et al. Renal disease in children with the acquired immunodeficiencysyndrome. N. Engl. J. Med. 321, 625–630 (1989).

3. Winston, J.A., Burns, G.C. & Klotman, P.E. Treatment of HIV-associatednephropathy. Semin. Nephrol. 20, 293–298 (2000).

4. Humphreys, M.H. Human immunodeficiency virus-associated glomerulosclero-sis. Kidney Int. 48, 311–320 (1995).

5. Dickie, P. et al. HIV-associated nephropathy in transgenic mice expressing HIV-1genes. Virology 185, 109–119 (1991).

6. Bruggeman, L.A. et al. Nephropathy in human immunodeficiency virus-1 trans-genic mice is due to renal transgene expression. J. Clin. Invest. 100, 84–92(1997).

7. Cohen, A.H., Sun, N.C., Shapshak, P. & Imagawa, D.T. Demonstration of humanimmunodeficiency virus in renal epithelium in HIV-associated nephropathy. Mod.Pathol. 2, 125–128 (1989).

8. Winston, J.A. et al. Nephropathy and establishment of a renal reservoir of HIVtype 1 during primary infection. N. Engl. J. Med. 344, 1979–1984 (2001).

9. Bruggeman, L.A. et al. Renal Epithelium is a previously unrecognized site of HIV-1 infection. J. Am. Soc. Nephrol (2000).

10. Cantor, E.S., Kimmel, P.L. & Bosch, J.P. Effect of race on expression of acquiredimmunodeficiency syndrome-associated nephropathy. Arch. Intern. Med. 151,125–128 (1991).

11. US Renal Data System. (National Institute Of Health, NIDDK, Bethesda,Maryland, 1997).

12. Bourgoignie, J.J., Ortiz-Interian, C., Green, D.F. & Roth, D. Race, a cofactor inHIV-1-associated nephropathy. Transplant Proc. 21, 3899–3901 (1989).

13. Deng, H. et al. Identification of a major co-receptor for primary isolates of HIV-1.Nature 381, 661–666 (1996).

14. Hoffman, T.L., Stephens, E.B., Narayan, O. & Doms, R.W. HIV type I envelope de-terminants for use of the CCR2b, CCR3, STRL33, and APJ coreceptors. Proc. Natl.Acad. Sci. USA 95, 11360–1135 (1998).

15. Choe, H. et al. The orphan seven-transmembrane receptor apj supports the entryof primary T-cell-line-tropic and dualtropic human immunodeficiency virus type

1. J. Virol. 72, 6113–6118 (1998).16. Owman, C. et al. The leukotriene B4 receptor functions as a novel type of core-

ceptor mediating entry of primary HIV-1 isolates into CD4-positive cells. Proc.Natl. Acad. Sci. USA 95, 9530–9534 (1998).

17. Moore, J.P. Coreceptors: Implications for HIV pathogenesis and therapy. Science276, 51–52 (1997).

18. Eitner, F. et al. Chemokine receptor CCR5 and CXCR4 expression in HIV-associ-ated kidney disease. J. Am. Soc. Nephrol. 11, 856–867 (2000).

19. Conaldi, P.G. et al. HIV-1 kills renal tubular epithelial cells in vitro by triggering anapoptotic pathway involving caspase activation and Fas upregulation. J. Clin.Invest. 102, 2041–2049 (1998).

20. Mack, M. et al. Transfer of the chemokine receptor CCR5 between cells by mem-brane-derived microparticles: A mechanism for cellular human immunodefi-ciency virus 1 infection. Nature Med. 6, 769–775 (2000).

21. Meng, G. et al. Primary intestinal epithelial cells selectively transfer R5 HIV-1 toCCR5+ cells. Nature Med. 8, 150–156 (2002).

22. Lagaye, S. et al. Cell-to-cell contact results in a selective translocation of maternalhuman immunodeficiency virus type 1 quasispecies across a trophoblastic barrierby both transcytosis and infection. J. Virol. 75, 4780–4791 (2001).

23. Chun, T.W. et al. Presence of an inducible HIV-1 latent reservoir during highly ac-tive antiretroviral therapy. Proc. Natl. Acad. Sci. USA 94, 13193–13197 (1997).

24. Bukrinsky, M.I., Stanwick, T.L., Dempsey, M.P. & Stevenson, M. Quiescent T lym-phocytes as an inducible virus reservoir in HIV-1 infection. Science 254, 423–427(1991).

25. Tanji, N. Effect of tissue processing on the ability to recover nucleic acid fromspecific renal tissue compartment. Exp. Nephrol. (2000).

26. Emmert-Buck, M.R. et al. Laser capture microdissection. Science 274, 998–1001(1996).

27. Bottaro, M., Berti, E., Biondi, A., Migone, N. & Crosti, L. Heteroduplex analysis ofT-cell receptor γ gene rearrangements for diagnosis and monitoring of cutaneousT-cell lymphomas. Blood 83, 3271–3278 (1994).

28. Higgins, D.G., Thompson, J. D. & Gibson, T.J. Using CLUSTAL for multiple se-quence alignments. Meth. Enzymol. 266, 383–402 (1996).

29. Faulkner, D.V. & Jurka, J. Multiple aligned sequence editor (MASE). TrendsBiochem. Sci. 13, 321–322 (1988).

30. Kimura, M. A simple method for estimating evolutionary rates of base substitu-tions through comparative studies of nucleotide sequences. J. Mol. Evol. 16,111–120 (1980).

31. Saitou, N. & Nei, M. The neighbor-joining method: A new method for recon-structing phylogenetic trees. Mol. Biol. Evol. 4, 406–425 (1987).

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://m

edic

ine.

nat

ure

.co

m