The human P^k histo-blood group antigen provides protection against HIV-1 infection

HIV-1 infection and development of AIDS vary greatly among persons and populations and are probably, at least in part, dependent on genetic factors.¹  Indeed, the first natural resistance factor reported for HIV infection was a polymorphism within the CCR5 HIV-1 coreceptor gene, termed CCR5-Δ32.^1,2 

However, no genetic factors thus far have been able to adequately explain the variability in both in vitro and in vivo susceptibility to HIV-1 infection.^2,3 

There is a longstanding association between pathogens and histo-blood groups, both in protection conferred by a specific blood type and in pathogen interactions with blood group antigens.⁴  The P/P1/P^k blood group antigens are of particular interest, with many defined pathogen interactions,^4,,–7  and an expression profile not limited to erythrocytes. Galabiose (Galα1-4Gal) is the terminal structure of P1 and P^k, also known as globotriaosylceramide (Gb₃) and a marker for germinal center B lymphocytes (CD77).⁸  P^k is the precursor for the P antigen, also known as globotetraosylceramide (globoside, Gb₄), which terminates with β1-3GalNAc.⁹  P₁ and P₂ are the 2 common P/P1/P^k-related blood group phenotypes. P₁ persons (∼ 80% of whites but only ∼ 20% of Asians)^6,10  express P and P1 but normally express moderate to low amounts of P^k on their cell surfaces. P₂ persons (∼ 20% of whites and ∼ 80% of Asians)^6,10  express only P and low amounts of P^k. However, rare phenotypes exist, having anomalies in one or more of the P/P1/P^k blood group antigens. Individuals deficient in P antigen have mutations in the B3GALNT1 gene causing lack of functional P (Gb₄) synthase (β3GalNAc transferase)^9,11  and consequently express increased levels of precursor, P^k. These persons may express P₁ antigen (P₁^k phenotype) or not (P₂^k), but the molecular basis for this is still unclear.¹⁰  Persons without any P/P1/P^k antigens have mutations in the A4GALT gene (α4Gal transferase or P^k (Gb₃) synthase), causing lack of P^k synthesis, and the rare p blood group phenotype.^11,,–14  (Table 1).

The P and P^k antigens are glycosphingolipids (GSLs), and GSLs play an important role in HIV-host cell interactions.^15,,–18  HIV envelope glycoprotein gp120 targets CD4 and CCR5 or CXCR4 chemokine coreceptors on monocytes and T cells, as the major HIV-host cell interaction,^19,–21  but HIV gp120 also binds to several GSLs in vitro, including P^k.^15,–17,22  GSL interactions are mediated by a sphingolipid recognition motif on the gp120-V3 loop, thought to facilitate post-CD4 binding and membrane fusion.^18,22  Inhibition of GSL biosynthesis can prevent HIV-host cell membrane fusion and infection.^23,24  This can be overcome by reintroduction of purified GSLs, or overexpression of CD4 and CXCR4, suggesting that GSLs have a facilitative role.²⁴  P^k, and to a lesser extent GM3, has appeared to be primarily implicated in augmenting HIV-membrane fusion, at least in in vitro reconstitution models.²⁴ 

In contrast, our recent work suggested P^k, when accumulated because of a lack of activity of α-galactosidase A in Fabry disease,²⁵  is protective against R5 HIV-1.²⁶  In addition, a soluble analog of P^k inhibits HIV infection in vitro.²⁷  Moreover, HIV-infected peripheral blood–derived mononuclear cells (PBMCs) have increased GSL expression, including P^k, indicating a potential cellular response to HIV-1.²⁸  Recently, pharmacologic modulation of P^k expression in vitro in HIV-1–infectable P^k-expressing non-T cells has further implicated an important role for P^k in HIV infection.²⁹ 

In light of these findings, we have now assessed HIV-1 susceptibility of PBMCs that are naturally high in P^k (P₁^k phenotype) or naturally devoid of P^k (p phenotype). In addition, we have genetically and biochemically manipulated P^k expression in HIV-1–infectable CD4⁺ HeLa cells and Jurkat T cells. Our findings show significant differences that reveal P^k status to be an important factor for susceptibility to HIV-1 infection.

Waste buffy coat material from anonymous regular blood donors was from the Lund University Hospital Blood Center (Lund, Sweden). This provision complies with current national regulation regarding the use of superfluous material from blood donations where the donor origin cannot be traced. Consent was obtained at the time of donation. Waste buffy coat material was provided from various centers with informed consent according to the Declaration of Helsinki from the donors of P₁^k and p phenotype blood and made anonymous for this study. The protocol was reviewed and approved by the Canadian Blood Services Institutional Review Board Committee. The regular donor controls were matched for ABO group and date of donation. PBMCs and lymphoblastic cell lines were cultured in “complete medium” consisting of RPMI 1640 medium (Invitrogen, Burlington, ON) plus 10% fetal bovine serum, 2 mM l-glutamine, and 10 μM gentamicin antibiotics. The human T-cell line, Jurkat FHCRC (Jurkat C), was a gift from Dr Gordon Mills (M. D. Anderson Cancer Center, Houston, TX), and Jurkat E6.1 and MT-4 cells were obtained from the National Institutes of Health (NIH) AIDS Research and Reference Reagent Program (Rockville, MD). The human cervical cancer cell line, HeLa (clones 6C [HT4-6C] and 1022), stably transfected to express CD4 (CD4⁺ HeLa) and sorted for high expression for infection with HIV-1, was obtained from the NIH AIDS Research and Reference Reagent Program or from Dr Alan Cochrane (University of Toronto, Toronto, ON) and cultured in RPMI plus 10% fetal bovine serum, 0.01% gentamicin antibiotics, and 0.1% amphotericin B. Test and control blood was collected into acid-citrate-dextrose anticoagulant on the same day and transported together to Canada for analyses. On arrival, PBMCs were isolated and activated using phytohemagglutinin (PHA)/interleukin-2 (IL-2) or PHA alone as described.²⁶  D-threo-1-phenyl-2-palmitoylamino-3-pyrrolidino-1-propanol (P4) was purchased from Matreya (Pleasant Gap, PA).

Blood samples acquired for this study were characterized extensively for categorization as control (P₁ or P₂), p, or P₁^k. Standard serologic techniques determined the erythrocyte phenotype and antibody specificities of blood samples. DNA was isolated from whole blood with the Qiagen QIAmp Blood Extraction kit (QIAGEN, Hilden, Germany). Genotypic characterization of samples was performed as reported.^9,14  (Tables 2, 3).

X4 HIV-1_IIIB and R5 HIV-1_Ba-L were from the National Institutes of Health AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, NIH: HTLV_IIIB = HIV-1_IIIB from Dr Robert Gallo, HIV-1_Ba-L from Dr Suzanne Gartner, HIV-1_JR-FL from Dr Irvin Chen, HIV-1_Ada-M from Dr Howard Gendelman, and HIV-1_{NL4-3gp41(36G) V38E, N42S} from Trimeris (Durham, NC). HIV-1_IIIB viral stocks were grown in Jurkat C cells, and multiplicity of infection (MOI) was determined as described using MT-4 cells.³⁰  All other viral stocks were grown in PBMCs, and infectious dose calculated from total p24^gag levels^31,32  measured by enzyme-linked immunosorbant assay (ELISA; Beckman Coulter, Fullerton, CA or ZeptoMetrix, Buffalo, NY). Briefly, cells were incubated with HIV-1 for 1 hour at 37°C, the cells washed extensively with phosphate-buffered saline (PBS), and cultured in complete medium. Culture supernatant aliquots were taken 2 hours after initial viral infection and subsequent time points. To determine viral production, ELISA was used to measure p24^gag antigen levels.

Fluorescence-activated cell sorter (FACS) analysis was performed as previously described using PHA- and PHA/IL-2–activated PBMCs and 1.5 μg monoclonal mouse anti-GM3 or anti-P^k (both from Seikagaku, Tokyo, Japan).²⁶  Alternatively, 12.5 μg/mL monoclonal mouse anti-CCR5 (clone 45549.111; NIH AIDS Research and Reference Reagent Program) was used. Secondary antibodies were either 5 μg/mL allophycocyanin (APC)–labeled goat antimouse IgG (Invitrogen) or 1 μL fluorescein isothiocyanate (FITC)–labeled goat antimouse IgG (Sigma-Aldrich, St Louis, MO). For anti-GM3–labeled samples, 10 μg/mL APC-labeled goat anti–mouse IgM was used (Cedarlane Laboratories, Burlington, ON). FACS analysis for P^k expression of small interfering (si)RNA-transfected CD4⁺ HeLa cells was carried out using 5 μg/mL VT1B-Alexa₄₅₈ (produced in the Lingwood laboratory). For tricolor FACS analysis, an additional incubation with 20 μL 10% mouse serum in FACS buffer for 10 minutes at 4°C in the dark was carried out before incubation with 10 μL mouse anti–CD4-peridinin chlorophyll protein (PerCP) Cy5.5 (BD Biosciences, San Jose, CA) and/or 5 μL mouse anti–CXCR4-phycoerythrin (PE; Serotec, Oxford, United Kingdom). Data were collected with a calibrated (BD CaliBRITE; BD Biosciences) Becton Dickinson FACSCalibur cell cytometer and analyzed using CellQuest software.

Extraction and thin layer chromatography (TLC) separation of GSLs, including ganglioside GM3, was as previously described.³³  GSL species were detected either by orcinol spray (Sigma-Aldrich) at 110°C for 10 minutes, or P^k selectively detected by verotoxin-1 (VT1) TLC overlay.³⁴  For VT1 overlay, the plate was blocked with 1% bovine gelatin; and after incubation at 37°C, the plate was washed with 50 mM Tris-buffered saline (TBS), pH 7.4. The plate was incubated for 45 minutes at room temperature with purified VT1, 1 μg per 10 mL in TBS. After washing, the plate was incubated for 45 minutes with a monoclonal antiverotoxin B subunit³⁴  (diluted 1/2000), washed and incubated with horseradish peroxidase-conjugated goat antimouse IgG (diluted 1/2000; Bio-Rad, Hercules, CA). For GM3 immunostaining, monoclonal anti-GM3 was substituted for VT1, followed by incubation with horseradish peroxidase–conjugated goat antimouse IgG. The plate was developed for 1 to 10 minutes with a 3 mg/mL solution of 4-chloro-1-naphthol in methanol freshly mixed with 5 volumes of TBS and 1/2000 dilution of 30% H₂O₂. Where band intensity was calculated from TLC blots, ImageJ software (NIH) was used to quantify relative intensities of bands visualized on TLC, where background signals were subtracted.

Jurkat E6.1 cells do not express P^k and are highly infectable with HIV-1. P^k or P liposomes were prepared by drying 400 μg P^k or P (Sigma-Aldrich) with 200 μg phosphatidylethanolamine and 200 μg phosphatidylserine in chloroform/methanol (2:1) under nitrogen. Alternatively, control phospholipid (PL) liposomes were prepared by drying 200 μg phosphatidylethanolamine with 200 μg phosphatidylserine. Liposomes were generated by vortexing well in 400 μL PBS and sonicating for 30 minutes. Liposomes or equivalent volumes of PBS were incubated with 16 × 10⁶ Jurkat E6.1 cells (4 × 10⁶ cells/mL) in serum-free RPMI 1640 for 1 hour at 37°C on a shaker (100 rpm). After incubation, cells were washed twice with PBS and cultured 18 to 24 hours at 37°C before infection with HIV-1_IIIB.

Ad5/F35 vectors were generated as previously described³⁵  by in vivo recombination in Escherichia coli BJ5183 cells between pAdenoVator transfer plasmids and pAdEasy-1/F35 adenoviral genome (a generous gift from Dr X. Fan, Lund University, Lund, Sweden)³⁶  using the AdenoVator Vector system (Qbiogene, Irvine, CA). Transfer plasmids containing the cytomegalovirus (CMV) promoter/enhancer with a β-globin/IgG chimeric intron (CMVi) were purchased from Qbiogene. For enhanced yellow fluorescent protein (EYFP) control Ad5/F35 vectors, EYFP from pIRES-EYFP (Clontech, Mountain View, CA) was cloned into the transfer plasmids. For P^k expression, Ad5/F35 vectors containing an expression cassette encoding EYFP under the control of the mouse PGK promoter³⁵  was first cloned into the CMVi transfer plasmid, and the full-length human P^k synthase (P^k-S) cDNA, cloned from CaCo-2 cells using primers for reverse-transcribed polymerase chain reaction that were designed based on the published sequence (GenBank database accession no. AB037883),³⁷  was then cloned into the CMVi expression cassette.

Recombinant Ad5/F35 vectors were transfected into QBI-293A cells using a standard calcium phosphate transfection procedure, and recombinant viruses were plaque-purified. Viruses were than amplified by transduction of large HEK293 cell cultures. Viruses were extracted by 3 consecutive freeze-thaw cycles and purified by a discontinuous CsCl gradient followed by a continuous CsCl gradient to completely separate infectious from defective viral particles. The viral preparations were dialysed against Tris 20 mM, pH 8, 2.5% glycerol, and 25 mM NaCl, concentrated using Amicon Ultra-4 MWCO 30 000 concentrators (Millipore, Ville St Laurent, QC) and sterilized by filtration through 0.22 μm Millex-GV filters (Millipore). The viral titers were determined by the tissue-culture infectious dose (TCID₅₀) method following the manufacturer's instructions (Qbiogene). Two adenoviral vectors were made: a control vector (pAd5/F35-CMVi-EYFP) and a test vector (Ad5/F35-CMVi-P^k-S-EYFP). The titers of the viral stocks were between 3 × 10⁸ and 1 × 10⁹ infectious units/μL.

Approximately 5 × 10⁵ CD4⁺ HeLa cells were plated in triplicate in 6-well plates and incubated overnight. Cells were then incubated for 1 hour at 37°C and 5% CO₂ in 250 μL Iscove modified Dulbecco medium (IMDM) containing the control (pAd5/F35-CMVi-EYFP) or test (Ad5/F35-CMVi-P^k-S-EYFP) vector at a MOI of 25. Untransduced control cells received 250 μL of IMDM only. After incubation, IMDM media was added to a volume of 2 mL. Plates were incubated at 37°C and 10% CO₂ for 48 hours; and after transduction, cells were recovered and subjected to FACS analysis and infection with X4 HIV-1_IIIB (MOI, 0.3) where productive infection was monitored over time.

The glucosyl ceramide synthase inhibitor, P4,³⁸  was used at 2 μM to pretreat CD4⁺ HeLa cells for 5 days before HIV infection. FACS analysis indicated that this treatment was sufficient to greatly reduce cell-surface expression of P^k without cell toxicity.²⁹  After 5 days of P4 treatment, cells were washed and then infected with X4 HIV-1_IIIB (MOI, 0.3) and productive infection monitored over time.

Approximately 2 × 10⁵ CD4⁺ HeLa cells were plated in triplicate in 6-well plates and incubated overnight in antibiotic-free RPMI media. Media was then replaced with serum-free, antibiotic-free RPMI. Depletion of P^k-S by siRNA was carried out as per the manufacturer's instructions (Thermo Electron, Waltham, MA) with modifications. Briefly, 2 μL Dharmafect-1 was added to 198 μL serum-free RPMI (tube 1). At the same time, 100 μL siRNA (2μM) was added to 100 μL serum-free RPMI (tube 2). Both tubes were mixed separately by pipetting and incubated for 5 minutes. Tube 1 was then added to tube 2, mixed carefully, and incubated for 20 minutes at room temperature. This mixture was added to cells, which were subsequently cultured for 24 hours. This procedure was repeated after 24 and 48 hours. The siRNA depletion of P^k-S was monitored by FACS analysis of P^k expression. CD4⁺ HeLa cells, with demonstrated reduction in surface levels of P^k (> 70%), were infected with X4 HIV-1_IIIB (MOI, 0.3) and aliquots of culture supernatant taken over time to monitor p24^gag production by ELISA.

A 2-sample Student t test, assuming unequal variance with 2-tailed distribution, was used to determine significance. The means of the data points for blood group phenotype were compared with their respective matched controls and represented plus or minus SEM, where n = 4. The means of the data points for treated/manipulated cells were compared with their respective control-treated or unmanipulated cells and represented plus or minus SEM, where n = 3 or 4. Data were considered statistically significant if P was less than .05 or highly significant if P was less than .002.

We first assessed the susceptibility to HIV-1 infection of PBMCs from P₁^k persons. Given the rarity of these samples (Table 1),⁶  P₁^k PBMCs from one donor (P₁^k-a) were used to assess R5 HIV-1 infection and a second donor (P₁^k-b) to assess X4 HIV-1 infection (Tables 2 and 3 for test and control designations, respectively). Infection of PHA-activated P₁^k-a with R5 HIV-1_Ba-L showed significantly lower productive HIV-1_Ba-L infection compared with its draw-date- and ABO-matched control (Figure 1A). PHA/IL-2–activated P₁^k-b were similarly protected against productive X4 HIV-1_IIIB infection (Figure 1B) compared with the respective control. Based on comparison with draw-date–matched controls, infection levels for P₁^k PBMCs for both HIV-1_Ba-L and HIV-1_IIIB were less than 12% (data not shown).

To determine whether expression levels of HIV receptors may have influenced the reduced infection levels, cell-surface CD4, CCR5, and CXCR4 levels on the same cell populations used for infection studies were analyzed by flow cytometry. PHA-activated P₁^k-a showed approximately 10% less CD4-expressing cells than the matched control; however, CD4 expression levels (mean fluorescence intensity [MFI]) were approximately 1.5-fold higher (Figure 1C,D left panels). There were also approximately 11% more CCR5-expressing cells in P₁^k-a and slightly higher CCR5 expression (MFI ∼ 1.2-fold difference; Figure 1C,D). The percentage of R5 HIV-1–susceptible target PBMCs, expressing both CD4 and CCR5, was also slightly higher in P₁^k-a (Figure 1C).

PHA/IL-2–activated P₁^k-b demonstrated approximately 7% more CD4-expressing cells compared with control and approximately 3.5-fold higher CD4 expression levels (MFI) (Figure 1C,D right panel). In addition, there were 27% more CXCR4-expressing cells than control, and approximately 3.5-fold higher CXCR4 expression in P₁^k-b (Figure 1C,D). Indeed, even the percentage of X4 HIV-1–susceptible target PBMCs, expressing both CD4 and CXCR4, was higher in P₁^k-b (Figure 1C). The percentage of cells expressing cell-surface P^k on PHA- and PHA/IL-2-activated P₁^k PBMCs was approximately 1.5- to 2-fold higher than that of controls (Figure 1E), with up to 4.5-fold higher density of cell-surface P^k expression as measured by MFI (Figure 1E). Cells expressing both CD4 and P^k, which encompass HIV-1–susceptible target cells, were also found to be twice as frequent in P₁^k PBMCs compared with their respective controls (Figure 1F).

We also assessed susceptibility of PBMCs from 3 P^k-deficient p persons (Table 2) to R5 and X4 HIV-1 infection. HIV-1_Ba-L infection of PHA-activated p PBMCs (denoted p1, p2, and p3; Table 2) resulted in much higher levels of productive infection compared with their draw-date- and ABO-matched control (Figure 2A). Following infection over time depicts exponential kinetics in HIV-1 production for p PBMCs. The difference in infection levels between p PBMCs and control showed a change of approximately 5-fold higher for p1, 12-fold higher for p2, and approximately 3000-fold higher for p3 (data not shown). This increased infection was consistent for R5 HIV-1 infection in general, as 2 other R5 strains, HIV-1_Ada-M and HIV-1_JR-FL, also showed much higher productive HIV-1 infection in the p PBMCs compared with control (Figure 2C).

As with R5 infection, HIV-1_IIIB infection of PHA/IL-2-activated p PBMCs from 2 persons (p1 and p3) also showed much higher levels of productive infection compared with their matched controls (Figure 2B,D). However, the p2 sample showed a 2-fold lower infection level (Figure 2B center panel), but overall infection in this experiment (C-p2 and p2) was much less than for the other p PBMC experiments. For the last p1 and p3 samples analyzed, the difference in infection levels between p PBMCs and their respective controls showed a change of 3-fold higher for p1 and approximately 600- to 1000-fold higher for p3 (Figure 2B). One other X4 strain, HIV-1_{NL4-3 gp41}, used to infect PHA/IL-2-activated p3 also showed more than 1000-fold higher productive HIV-1 infection compared with control (Figure 2D), consistent with X4 HIV-1_IIIB results.

To determine whether expression levels of HIV receptors influenced the observed susceptibility to infection, the same cell population used for infection was subjected to flow cytometry to determine cell-surface CD4, CCR5, and CXCR4. In general, PHA-activated p PBMCs (p1, p3) presented more CD4-expressing cells than their controls (an increase of ∼ 14%-40%), which also translated into higher CD4 expression levels (MFI, 1.3- to 3-fold higher; Figure 3A,C). This was most evident for p3, which showed the highest susceptibility to R5 HIV-1. There were also more CCR5-expressing cells in p PBMCs (p1, p2; an increase of ∼ 12%-19%), and 3-fold higher CCR5 expression levels (MFI; Figure 3A,C). Furthermore, the percentage of R5 HIV-1–susceptible target PBMCs, expressing both CD4 and CCR5, was greater in p PBMCs (p1, p2, and slightly in p3; Figure 3A). However, p3 PBMCs showed reduced CCR5 levels (Figure 3A,C).

PHA/IL-2–activated p PBMCs (p1, p3) demonstrated more CD4-expressing cells compared with their controls (an increase of ∼ 21%-42%), and up to 3-fold higher CD4 expression levels (MFI; Figure 3B,D). These differences were once again most evident in p3, which demonstrated the highest increase in X4 HIV-1 infection. The p3 sample showed more CXCR4-expressing cells than control (> 10%), and overall there was 1.3- to 2.3-fold higher CXCR4 expression in p-PBMCs from both p1 and p3 samples (Figure 3B,D). The percentage of X4 HIV-1–susceptible target PBMCs, expressing both CD4 and CXCR4, was noticeably higher in p PBMCs from p3 (Figure 3B). In contrast, p2 showed a somewhat opposite expression profile, exhibiting approximately 15% less CD4 and approximately 5% less CXCR4 expressing cells compared with the matched control (Figure 3B center panel), as well as lower receptor expression levels (MFI; Figure 3D center panel).

Increased HIV-1–induced T-cell fusion has been reported in p-CD4⁺ T cells, ascribed to higher total levels of GM3.³⁹  Although GM3 is reported less fusogenic than P^k,24, we investigated the possibility that GM3 levels influenced p-PBMC (p3) susceptibility to infection in our system. Total GSLs isolated from PHA-activated PBMCs revealed loss of GM3 in control compared with p-PBMCs (Figure 4A), calculated according to band intensity on the TLC plate to be approximately 3-fold different (Figure 4C). Resting or PHA/IL-2–activated PBMCs, however, showed minimal differences in total p-PBMC GM3 levels compared with the respective control (Figure 4A-C). Higher total GM3 expression in PHA-activated p PBMCs did not translate to higher percentage of cells or cell-surface GM3 expression as measured by FACS analysis (Figure 4D center panel; Figure 4E). Although there was a slightly higher percentage of GM3-expressing p PBMCs in the PHA/IL-2–activated population, only subtle differences were seen in cell-surface GM3 expression (Figure 4D right panel; Figure 4E).

Exogenous P^k was introduced into P^k-deficient Jurkat T-cell membranes by P^k-liposome fusion. After fusion, approximately 35% of the Jurkat cell population expressed surface P^k at high levels (MFI; Figure 5A,B). Within the CD4⁺ target population, approximately 32% expressed both CXCR4 and P^k (Figure 5C right panel). P^k-liposome–treated cells showed no differences in CD4 or CXCR4 expression compared with PBS or PL-liposome controls (Figure 5B,C). Increased P^k expression after P^k-liposome transfer was confirmed by TLC (Figure 5E). A significant reduction in X4 HIV-1_IIIB infection was observed in the P^k supplemented Jurkat cells, being only 43% of the HIV-1_IIIB infection levels of PBS, P-, or PL-liposome controls (Figure 5F).

To confirm that P^k expression levels influence HIV-1 infection levels, we tested whether modulating the expression of P^k in CD4⁺ HeLa cells, which express P^k and are infectable, would correlate with subsequent HIV-1 infection (Figure 6). Cells transduced with adenoviral vector expressing P^k synthase (P^k-S) resulted in increased levels of total and cell surface P^k compared with nontransduced cells or cells transduced with a control adenoviral vector (Figure 6A,B). Compared with untreated cells or control adenoviral vector transduced cells, HIV-1 infection was significantly lower in the increased P^k-expressing CD4⁺ HeLa cells transduced with the adenoviral vector P^k-S (Figure 6C).

P4 was used to inhibit glucosylceramide-based GSL synthesis, thus blocking the biosynthetic pathway to P^k.³⁸  P4 treatment of CD4⁺ HeLa cells resulted in a substantial decrease in cell populations expressing P^k (Figure 6D). A decrease in P^k expression was also shown in the total GSL profile (Figure 6E). P4-treated cells further demonstrated significantly increased HIV-1 infection levels (Figure 6F).

To demonstrate that specific reduction of P^k influences the level of HIV-1 infection, siRNA was used to transiently silence the P^k synthase gene, encoding the enzyme responsible for the addition of the terminal galactose to the precursor for P^k.^5,10  Transfection of P^k synthase-specific siRNAs into CD4⁺ HeLa cells resulted in a substantial decrease in cells expressing P^k compared with siRNA controls (Figure 7A,B). The decrease in total P^k was selective without significant decrease in P^k precursors GlcCer or LacCer, monitored by TLC and VT1 overlay (Figure 7B,C). Cells transiently transfected with siRNA to P^k synthase demonstrated significantly increased HIV-1 infection levels (Figure 7D).

Our findings indicate a new phenomenon of P^k-mediated reduced susceptibility to HIV-1 infection. P₁^k PBMCs, which highly express P^k on their cell surface, demonstrate lower levels of productive R5 and X4 HIV-1 infection. In contrast, p PBMCs, which do not express P^k, show a higher susceptibility to R5 and X4 HIV-1 infection. Accordingly, P^k-liposomal transfer or P^k-synthase gene transduction facilitated a reduction in HIV-1 infection, whereas GlcCer-based GSL (P^k) depletion or P^k-synthase gene silencing resulted in an increase in HIV-1 infection. Thus, higher expression of P^k in vivo and in vitro correlates with decreased HIV-1 infection, whereas a lower expression or lack of P^k expression results in increased HIV-1 infection.

Our studies indicate that susceptibility to HIV-1 infection in p PBMCs might be influenced both by the lack of P^k antigen and by increased receptor and coreceptor expression; however, this is not the case with the P₁^k phenotype. P₁^k PBMCs demonstrated reduced susceptibility to R5 and X4 HIV-1 infection despite having increased expression of HIV-1 receptors. Thus, both rare p and P₁^k PBMCs showed increased patterns of HIV receptor and coreceptor expression, but this resulted in higher susceptibility to HIV infection only in the p PBMCs. Thus, P^k expression was a better indicator of susceptibility to HIV-1 infection than CD4 or chemokine coreceptor expression.

Although the presence (or absence) of P^k is important in blood group classification and transfusion medicine, P^k is not restricted to erythrocytes. P^k is expressed on monocyte populations, which encompass R5 HIV-1-susceptible target cells.^40,41  T-lymphoblasts mostly represent X4 HIV-1-susceptible target cell populations and have been reported to express little or no P^k40 ; thus, T cells are similar to the p phenotype in their lack of P^k expression, which may promote susceptibility to HIV-1 infection. Furthermore, variations in P^k expression occur in the general population,⁴²  which could explain differences in susceptibility to HIV-1 infection seen in vitro and in vivo.

Differences in P^k expression could influence lipid raft composition of target cell membranes and affect CD4 and/or coreceptor localization. Lipid rafts are central to HIV infection,⁴³  and CD4 and CCR5 are known to be associated with lipid rafts, whereas CXCR4 is not.⁴⁴  However, even CD4-HIVgp120-CXCR4 associations have been demonstrated within rafts and are required for membrane fusion.⁴⁵  If P^k levels were able to influence appropriate localization of CD4 and/or coreceptors in lipid rafts, because of changes in the membrane milieu, this could affect target cell susceptibility to HIV-1.

Importantly, heightened susceptibility of cells without P^k, and reduced susceptibility of cells that express increased P^k, to both X4 and R5 HIV-1 infection would argue against current models, suggesting that P^k is important in post-CD4-binding.²²  Increased GM3 has been proposed to promote membrane fusion in p-CD4⁺ T cells.³⁹  However, cell-surface expression and total GM3 do not correlate with enhanced PHA- or PHA/IL-2-activated PBMC HIV-1 infection in our study, although purified target cells remain to be assessed. It is clear, however, that P^k is not an absolute requirement for membrane fusion and infection. HIV-gp120 binds P^k via the V3 loop.^17,18,22  This loop also mediates chemokine receptor binding.^46,47  Thus, P^k (or a soluble mimic²⁷ ) binding to gp120 may interfere with post-CD4 recognition of chemokine coreceptor binding to prevent fusion and infection. Indeed, the binding motif, XXXGPGRAFXXX,⁴⁸  within the V3 loop for P^k binding overlaps with the consensus binding motif, S/GXXXGPGXXXXXXXE/D,⁴⁶  for chemokine coreceptors. It has also been shown that CD4 enhances gp120-P^k interaction,⁴⁹  probably by a similar mechanism that allows for the interaction of chemokine coreceptor with gp120 after CD4 binding.⁵⁰  Perhaps, under conditions of chemokine receptor deficiency (or the absence of CD4), P^k may thus (less efficiently) mediate viral internalization. However, when receptor levels are normal, and P^k is expressed at higher levels, P^k has the potential to interfere with the appropriate interactions between gp120 and chemokine coreceptors, thus inhibiting viral internalization (see Figure 7E for a working model).

The lack of P in P₁^k cells could suggest that P can facilitate, rather than P^k inhibit, infection. However, the high susceptibility of the p phenotype, which lack both P and P^k, makes this unlikely. In addition, gp120 binds P^k but not P.¹⁵  Furthermore, the introduction of P^k by liposome transfer into a cell line deficient in P^k expression (providing a close representation to the p phenotype), confirmed the decrease in susceptibility to HIV-1 on increased P^k levels. The fact that introduction of P into this cell line does not affect HIV infection would argue against any ability to facilitate infection. Only the levels of P^k closely correlate to HIV susceptibility. This is further supported by use of a cell line, HeLa, which does not express P (Figure 7B), whereby after the introduction of the P^k synthase gene (α4Gal transferase), which increased the cell-surface expression levels of P^k was able to reduce HIV-1 infection. In addition, specific gene silencing using siRNAs to the P^k synthase gene resulted in increased HIV-1 infection.

In our previous study of Fabry patient samples,²⁶  which present intracellular P^k accumulation because of the lack of α-galactosidase A activity, we demonstrated a reduced susceptibility to HIV-1 infection. However, because we could only detect low levels of cell-surface expressed P^k, the mechanism of reduced HIV infection was unclear. This could have involved aspects of the abnormal pathology as a result of Fabry disease and/or abnormal trafficking of necessary coreceptors for HIV-1 infection.²⁶  Indeed, Fabry PBMCs only demonstrated a reduction in R5 HIV-1 infection, and CCR5 coreceptor was greatly decreased on the cell-surface of these patient samples. In contrast, in the current study, we show that HIV-1 infection directly correlates to increased or decreased cell-surface expression of P^k, and this is largely independent of CXCR4 or CCR5 coreceptor expression. When P^k is highly expressed on the cell surface, as is the case in P₁^k persons' PBMCs, infection with HIV-1 × 4 and R5 viruses is largely reduced. However, when there is no P^k cell-surface expression, such as in p persons' PBMCs, HIV-1 infection is potentially several logs greater than in cells having normal P^k cell-surface expression.

Although natural resistance factors to HIV infection have been actively sought, there have been no reports as yet of a cell-surface receptor that can provide a natural barrier to HIV infection.^1,,–4  The Δ32 polymorphism in the CCR5 chemokine cell-surface receptor that provides natural resistance to HIV infection is the result of a mutation that prevents the transport of this receptor to the cell surface. Thus, persons with this polymorphism do not express the receptor for R5 viruses on their cell surface.²  We now provide the first evidence of a possible role for a naturally expressed cell-surface factor, the P^k GSL, as potentially providing some protection to both R5 and X4 strains of HIV-1. Although studies examining the incidence of the p and P₁^k phenotype in cohorts of HIV-infected, HIV-exposed but uninfected, HIV progressers and nonprogressers would be desirable, the frequency of these extremely rare phenotypes, estimated for p to be 5.8 per million, and with P₁^k much less frequent (∼ 1 per million),^5,6  precludes these studies. Significantly, genetic studies identified chromosome 22q12-13 to be associated with HIV resistance,⁵¹  and this region contains the P^k synthase gene¹³  and HIV transgenic mice showed increased P^k synthesis.⁵²  To determine whether P^k cell-surface expression may indeed represent a natural resistance factor for HIV infection, population studies are required using normal cohorts with common P₁/P₂ phenotypes known to have differential P^k expression⁴²  to assess HIV-1 susceptibility in vitro. Furthermore, analyses of HIV-1-infected and HIV-1-resistant cohorts, using genetic and serologic/flow cytometric techniques are necessary. Nonetheless, based on our findings, P^k alone provides some protection to infection with HIV-1 and studies of modulation of P^k expression, by pharmacologic²⁹  or other intervention, may prove to be important for future HIV/AIDS treatment modalities.

The authors thank those donors whose PBMCs were used in this study, Mathieu Drouin for preparation of the P^k synthase–containing adenovector, and Dr Xiaolong Fan from Lund University for providing the Ad5/F35 adenovirus backbone.

This work was supported by the Canadian Blood Services through a graduate fellowship award (N.L.), the Canadian Institutes for Health Research via a Doctoral Research Award (N.L.) and operating grants, Canadian Institutes for Health Research (C.A.L., D.R.B.), the Ontario HIV Treatment Network (C.A.L., D.R.B.), the Canadian Association for HIV Research (C.A.L., D.R.B.), the Swedish Research Council (project no. K2005/2008-14251), the Medical Faculty of Lund University, governmental grants for clinical research to Lund University Hospital, and Region Skåne, Sweden (ÅH., M.L.O.).

Contribution: N.L. performed experiments, analyzed data, and contributed to the writing of the manuscript; M.L.O. provided essential samples, analyzed data, and contributed to design of experiments and to the writing of the manuscript; Å.H. provided and characterized essential samples; S.R., D.S., and B.B. performed experiments; V.Y. and C.L. provided essential samples; X.-Z.M. and D.J. provided essential reagents and contributed to the writing of the manuscript; and C.A.L. and D.R.B. contributed to the design of experiments, analysis of the data, and the writing of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Donald R. Branch, Canadian Blood Services, Toronto General Research Institute, 67 College Street, Toronto, ON M5G 2M1 Canada; e-mail: don.branch@utoronto.ca.