Top, Saliva kills human mononuclear leukocytes as determined by trypan blue uptake in 3 separate experiments; bottom, morphologic disruption in a representative experiment. Asterisks indicate P<.05 by regression analysis and nonparametric testing (see the "Statistical Analysis" section in the text for details).
Disruption of human mononuclear leukocytes incubated at 34°C for 15 minutes with saliva compared with blood plasma.
Saliva interrupts the multiplication of, top, human immunodeficiency virus (HIV) in infected human mononuclear leukocytes; bottom, vesicular stomatitis virus (VSV) in the CEM human lymphocyte cell line. Asterisks indicate P<.05 (see the "Statistical Analysis" section of the text for details).
Reconstitution of isotonicity with concentrated nutrient medium reverses saliva's inhibition of virus production. Asterisks indicate P<.05 (see the "Statistical Analysis" section of the text for details).
Baron S, Poast J, Cloyd MW. Why Is HIV Rarely Transmitted by Oral Secretions?Saliva Can Disrupt Orally Shed, Infected Leukocytes. Arch Intern Med. 1999;159(3):303-310. doi:10.1001/archinte.159.3.303
Copyright 1999 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1999
Oral transmission of human immunodeficiency virus (HIV) by the millions of HIV-infected individuals is a rare event, even when infected blood and exudate is present. Saliva of viremic individuals usually contains only noninfectious components of HIV indicating virus breakdown.
To determine whether unknown HIV inhibitory mechanisms may explain the almost complete absence of infectious HIV in the saliva.
Since most of the infectious HIV that is shed mucosally by asymptomatic individuals is found in, produced by, and transmitted by infected mononuclear leukocytes, we determined whether saliva, which is hypotonic, may disrupt these infected cells, thereby preventing virus multiplication and cell-to-cell transmission of HIV. Specifically, we measured (1) whether mononuclear leukocytes were lysed by saliva and (2) whether the lysis by saliva inhibits the multiplication of HIV and other viruses in infected leukocytes and other cells.
Saliva rapidly disrupted 90% or more of blood mononuclear leukocytes and other cultured cells. Concomitantly, there was a 10,000-fold or higher inhibition of the multiplication of HIV and surrogate viruses. Further experiments indicated that the cell disruption is due to the hypotonicity of saliva.
Hypotonic disruption may be a major mechanism by which saliva kills infected mononuclear leukocytes and prevents their attachment to mucosal epithelial cells and production of infectious HIV, thereby preventing transmission. Implications for the known oral HIV transmission by milk and seminal fluid, as well as potential oral transmission to contacts and health care workers, are considered. This effective salivary defense may be applicable medically to interdict vaginal, rectal, and oral transmission of HIV by infected cells in seminal fluid or milk by the use of anticellular substances.
HUMAN immunodeficiency virus (HIV) transmission from oral secretions of the millions of HIV-viremic individuals during kissing, dental treatment, biting, and aerosolization is a rare event, even when infectious HIV is shed into the oral cavity by infected blood or exudate.1- 8 Despite this shedding of infected blood or exudate, saliva of infected individuals usually contains only noninfectious components of HIV, indicating that there may be a breakdown or inactivation of infectious HIV, and may contain fragments or the entire noninfectious genome.9- 11 Only 1% to 5% of patients' saliva contains infectious HIV although they all carry virus in their blood.2,12,13 At other mucosal surfaces (vaginal and seminal fluids), the percentage of patients shedding infectious HIV is much higher (approximately 20%).14- 16 Also, most of the HIV shed at any mucosal surface during the asymptomatic infection originates from infected mononuclear leukocytes17- 25 because most of the cell-free virus in body fluids is neutralized by antibody produced after the initial acute infection.24- 27 In addition, cell-free infectious HIV or simian immunodeficiency virus (SIV) has low infectivity for the CD4-negative epithelial cells on all mucosal surfaces.16,28,29 However, at the vaginal mucosal surface, shed-infected mononuclear leukocytes deposited with seminal fluid are infectious14- 16; they survive and can attach to and infect the HIV-resistant CD4-negative epithelial cells,16,30- 33 or penetrate the epithelial layer, and infect susceptible CD4-positive subepithelial mononuclear leukocytes.34- 37
In saliva, inhibition of HIV may be partly due to several inhibitors of viruses that are present in the saliva.38- 48 For example, absence of nonspecific inhibitors in the saliva of a few patients with the acquired immunodeficiency syndrome correlates with the presence of infectious HIV in their saliva.12 Free secretory antibody also is present in saliva but may not be effective due to its low concentration.49- 51 However, considering the limited in vitro inhibition of HIV by salivary inhibitors (2- to 5-fold)38- 44,46,47,52- 54 but the almost complete absence of infectious HIV in saliva, even after shedding of infected blood, additional mechanisms may inhibit infectious HIV shed orally. Since most of the infectious virus that is shed orally during the asymptomatic phase of infection is in, or produced by, infected leukocytes, and since the CD4-negative mucosal epithelial cells resist infection by cell-free HIV, we hypothesized that salivas (that have only one seventh the tonicity of normal interstitial fluids55,56) may disrupt these crucial-infected cells and render them incapable of supporting virus multiplication and cell-to-cell transmission of HIV. Experimental support for this hypothesis is presented.
The multiplication of HIV was assayed as the yield of infectious HIV from 2 × 106 phytohemagglutinin-stimulated normal peripheral blood mononuclear cells that had been cultured with phytohemagglutinin for 3 days before treatment with 20 U/mL of interleukin 2 for an additional day and then infected with 106 50% tissue culture infectious doses of HIV, strain AC-1.57 The infected cells were then incubated for 3 days, washed 4 times before incubation with undiluted saliva or culture medium for 30 minutes. These leukocytes were subsequently washed and incubated for HIV production in RPMI 1640 tissue culture medium, plus 15% fetal bovine serum and 20 U/mL of interleukin 2 for 24 hours before harvesting the cell-free medium for assay of HIV yield as described later. In other experiments, easily studied surrogate viruses were substituted for HIV to identify mechanisms of action before confirmatory experiments with HIV. Vesicular stomatitis virus (VSV), sindbis virus, and mengovirus multiplication was measured as the yield of infectious virus in the culture medium from the human CEM–transformed lymphocyte cell line, murine L cells, or monkey Vero cells. This was done over an 8-hour single growth cycle, following infection with 1000 to 10,000 viral plaque-forming units. Undiluted saliva was added to the infected CEM lymphocytes or L cells at the times noted for each experiment.
The final yield of HIV was assayed as infectious focus-forming units on HIV-susceptible human colon carcinoma, SW480-CD4-tat, clone 3 indicator cells. The assay on these cells is the same as other established assays on other adherent indicator cells (eg, HeLa) transfected with the HIV receptor, CD4.58 The dilution response curve with clone 3 has been established to be linear, as it is for other reported adherent indicator cells and clone 3 has the same high sensitivity to infection by HIV.59,60 The clone 3 cells were developed from plastic-adherent, human colon carcinoma cells (ATCC SW480) that were transfected with the pMV7-T4 retroviral vector to stably express the human HIV receptor, CD4, and also HIV gene product tat (Enrico Duru, PhD, and M.W.C., written communication, 1998). For assays, serial 10-fold dilutions of HIV are added in duplicate to cultures of clone 3 cells in 24 well-culture plates and incubated at 37°C, usually for 3 days, which allows HIV multiplication as foci. The cultures are then fixed with 80% cold acetone, air dried, and a 1:100 dilution of pooled patient antiserum (titer of 10,000 by Western blot) to HIV is added for 1 hour at room temperature. The cultures are then rinsed 3 times and 0.2 mL of a 1:100 dilution of goat anti-human immunoglobulin (Sigma, St Louis, Mo), conjugated with alkaline phosphatase is added for 1 hour at room temperature. The cultures are rinsed again and the substrate, BCIP/NBT, liquid substrate system for alkaline phosphatase (Sigma), is added to stain the antibody-coated foci, which are then read as foci under a dissecting microscope. The yield of VSV was determined as the virus plaque titer on mouse L cells. The yields of sindbis virus and mengovirus were determined as the virus plaque titer on monkey Vero cells.
Four healthy individuals donated unstimulated whole saliva and each sample was studied separately. Variability between donor saliva sample was within experimental error. The donors were 2 women aged 43 and 49 years and 2 men aged 64 and 69 years. The saliva samples were clarified by centrifugation at 2000 rpm for 30 minutes, and were used unchanged, pH adjusted to 7.0 with sodium bicarbonate, or filtered for sterility through a 0.22-µm filter. These treatments of saliva did not alter cell lysis or inhibition of virus multiplication. A mock saliva was prepared as a hypotonic salt solution containing concentrations of sodium chloride and potassium chloride equivalent to saliva.55 A 10-fold concentrate of Dulbecco balanced salt solution or Eagle minimum essential medium was used to reconstitute isotonicity of saliva and hypotonic salt solutions. Healthy human leukocytes for the HIV multiplication experiments were from a male donor aged 69 years and the results confirmed with 2 additional donors as well as the CEM lymphocyte cell line. Institutional internal review board approval was obtained for all samples.
Undiluted saliva was mixed with mononuclear leukocytes and incubated for the times shown. Measurements of trypan blue dye uptake by cells was performed by mixing equal volumes of trypan blue (4 mg/mL) in a balanced salt solution with the Ficoll-hypaque–purified mononuclear leukocytes and counting the stained and unstained cells at various times over 60 minutes. Morphologic cell disruption was quantified by Wright staining of microscopic slide smears of Ficoll-hypaque–purified human monocytes or whole blood, and microscopic counting of total and disrupted leukocytes. The morphologic criteria for leukocyte disruption were swollen and burst cells with fragmented cell or nuclear membranes and leakage of nuclear or cytoplasmic contents (Figure 1 and Figure 2). A minimum of 100 cells were counted for each point for both the trypan blue and morphologic assay. All experiments were replicated a total of 2 or 3 times and either all or appropriately representative experiments are presented. All measurements were done in a masked fashion.
In addition to basic descriptive statistics in the figure legends, statistical methods such as regression analysis and nonparametric methods were used, with no adjustments for multiple testings. In every instance the pertinent trends and differences observed were so large, all related significance levels (P values) were small, even with sample sizes used.
Saliva rapidly disrupted (≥90%) mononuclear leukocytes and the other cultured cells. The extreme swelling followed by disruption began within minutes, as measured by trypan blue dye uptake (50% in Figure 1, top) and by morphologic criteria (≥90% in Figure 1, bottom) using Wright stain. Trypan blue dye uptake (Figure 1, top), which requires an intact cell membrane to retain the dye, often underestimates cell death when the cells and their membranes are severely disrupted61,62 and, therefore, may not be a definitive test of cell viability in this case. Consistent with this, Wright stain for morphologic disruption (Figure 1, bottom) showed virtually complete cell disruption, with broken membranes, within a few minutes. Figure 2 shows the severe morphologic disruption of mononuclear leukocytes incubated for 15 minutes in saliva compared with plasma. The single, apparently intact nucleus, lacks cytoplasm and was likely extruded from a disrupted cell. This profound cell killing by saliva was confirmed by the complete inhibition of virus multiplication in saliva-treated cells, indicating that few, if any, cells escape killing, as presented later.
To study the effect of salivary-induced cell lysis on virus-infected cells, we infected healthy human mononuclear leukocytes with HIV or we infected transformed human lymphocytes (CEM cells) or mouse L cells with the more easily studied surrogate virus, VSV. Undiluted saliva was applied to the infected cells for 30 minutes to simulate the high ratio of saliva to infected blood or exudate in the mouth. As shown in Figure 3, concomitant with cell killing (Figure 1 and Figure 2), there was interruption of HIV and VSV multiplication. Human immunodeficiency virus production over 24 hours was inhibited 10,000-fold or higher (Figure 3, top) compared with the 2- to 5-fold inhibition reported for salivary inhibitors.38- 44,46,47 Vesicular stomatitis virus yield in a single cycle of multiplication was inhibited 1000-fold by saliva (Figure 3, bottom). This same degree of inhibition should occur even with 1 minute of incubation of HIV-infected cells and saliva, since virtually all leukocytes are lysed at 1 minute (Figure 1, bottom). Also, since the saliva samples were sterile filtered, the class of large saliva inhibitors would have been removed40 but the strong inhibition of HIV still occurred as reported previously.47
The cause of cell disruption may largely be due to the hypotonicity of saliva,55,63,64 since cell disruption and the resulting interruption of virus multiplication occurred with solutions containing the same low-salt concentration as saliva (Figure 3, bottom). Also, saliva-induced cell disruption and inhibition of surrogate virus multiplication was reversed by reconstituting the saliva to isotonicity by adding concentrated salts in medium (Figure 3, bottom, and Figure 4). Consistent findings come from the reported killing of leukocytes with interruption of HIV multiplication by water.65
A sufficient volume of isotonic fluid (eg, blood, exudate, and milk) would prevent hypotonic lysis of leukocytes by diluting saliva with the isotonic fluid. To determine the ratio of saliva to an isotonic solution that would prevent cell lysis, we serially diluted saliva in 15% increments in isotonic medium and treated VSV-infected human CEM lymphocytes or murine L cells with each mixture. Complete inhibition of virus yield (>1000-fold) occurred with mixtures containing between 100% and 75% saliva, implying that all infected cells were killed by saliva and, therefore, did not produce virus. Diminishing inhibition of virus yield occurred with the mixtures containing 75% to 25% saliva, implying partial killing of cells by saliva. No inhibition of virus occurred below 25% saliva, implying complete protection of cells from salivary lysis. Since the residual volume of saliva in the mouth averages 0.75 mL,66 we may estimate that the volume of shed-infected blood required to partially protect its infected leukocytes against salivary lysis is between 0.5 mL (containing about 0.25 mL of isotonic plasma) and 4.5 mL (containing 2.25 mL of isotonic plasma). Complete protection of shed leukocytes would be expected to occur with shed blood volumes greater than 4.5 mL. The salivary flow rate of 0.3 mL of hypotonic saliva per minute56 would double the requirement for blood at 2 to 3 minutes. These estimates are being verified using HIV-infected mononuclear leukocytes, including those obtained from HIV-infected patients.
During the asymptomatic infection, infectious HIV shed from blood or exudate onto mucosal surfaces is contained mainly in mononuclear leukocytes.17- 25 This appears to result from inactivation of most of the extracellular virus by antibody in body compartments, leaving productively infected or latently infected mononuclear leukocytes available for HIV shedding.24- 27 In addition, mucosal epithelial cells resist cell-free HIV infection because they lack the CD4 receptor.16,28,29 Successful transmission of HIV to epithelial cells at mucosal surfaces may require that the viable infected mononuclear leukocytes remain viable, to attach to the epithelial cells and produce HIV locally, for successful infection or transposition to subepithelial CD4-positive cells.16,30- 32 Also, viability of infected leukocytes is required for their migration to infect subepithelial CD4-positive cells.34- 37 Thus, if HIV were to be transmitted mucosally (where the surface cells are noninfectable by cell-free HIV), it is critical that the shed-infected leukocytes remain viable. The demonstrated killing of HIV-infected leukocytes by the hypotonicity of saliva would prevent all these routes of infection, ie, attachment to epithelial cells, penetration of the epithelium by infected mononuclear leukocytes, and virus production.
This disruption of infected leukocytes by saliva is consistent with the previous findings that oral shedding of infected blood during dental treatment usually does not result in infectious HIV in saliva2 although infected leukocytes are present in the blood. Instead, fragments of HIV nucleic acids can be recovered from saliva,9,10 which is consistent with disruption of shed-infected cells and with the presence of antibody-neutralized cell-free virus. Another report40 consistent with hypotonic lysis of infected cells is that saliva, diluted in a hypotonic solution, did not lose HIV inhibitory activity even at the highest dilution of saliva, thereby implying that the hypotonic diluent may have lysed the assay cells to give an artificially high inhibitor titer. Other studies of salivary inhibitors using isotonic diluent detected some inhibition of HIV up to a 1:10 dilution, but that inhibition was small (2- to 5-fold) compared with the salivary inhibition of 10,000-fold or higher.
Consistent with the resistance of CD4-negative mucosal epithelial cells to cell-free HIV are the large doses of cell-free SIV required to infect mucosal surfaces of monkeys on which the epithelial cells are virus-receptor negative. Orally, 830 times more cell-free SIV was needed to achieve infection compared with intravenous infection.29 About 500 times more SIV was required to achieve rectal infection compared with intravenous infection.67,68 Approximately 7500 times more virus was required to achieve vaginal or male urethral infection compared with intravenous infection.69- 73 In addition, 95% to 99% of saliva samples from HIV-infected individuals do not contain any infectious HIV, leaving low levels of cell-free infectious HIV in the saliva samples of only a few individuals.2,12,13 Thus, infection of resistant CD4-negative epithelial cells by any low dose of cell-free HIV that may occur in a few saliva samples seems highly improbable.
Hypotonic lysis of cells is well studied at the biophysical and molecular levels. Moderate hypotonicity affects cellular ions, proteins, and regulatory mechanisms,74- 78 but extreme hypotonicity, as with saliva, causes cell death by osmotic swelling and bursting of cells and their plasma membranes.79
The resistance of mucosal epithelial cells to cell-free SIV implies that in humans vaginal HIV infection via seminal fluid may be due to its infected mononuclear leukocytes rather than to its low level of cell-free HIV. The evidence is that, first, too little cell-free HIV is present in seminal fluid,14- 16,80,81 ie, the titers of cell-free virus in seminal fluid are mostly undetectable or far below the 500- to 7500-SIV dose required to infect CD4-negative epithelial surfaces. Second, enough HIV-infected leukocytes are present in seminal fluid14- 16,80,81 to infect epithelial cells or to penetrate subepithelially to infect CD4-positive cells. Only a few of these HIV-infected mononuclear leukocytes are required to infect CD4-negative epithelial cells16,30- 32 or to penetrate the epithelial layer and infect susceptible CD4-positive cells.34- 37
Although most orally shed infectious HIV in asymptomatic patients is expected to occur as infected leukocytes, there is concomitant shedding of both a large amount of antibody-neutralized, cell-free HIV and, in 1% to 5% of patients, a small amount of infectious cell-free HIV.13,65 The effect of hypotonicity on any shed infectious cell-free HIV may be inferred from studies using water without salts.65,82,83 Inactivation of cell-free infectious HIV in water is reported to occur at rates of 3-fold per day,83 10-fold over 6 to 12 hours,82 and 10-fold over 12 to 1 hour.65 Thus, inactivation of cell-free infectious HIV by hypotonic solutions such as water, and perhaps saliva, may occur over hours, compared with the immediate lysis of infected leukocytes. Nevertheless, the slower but significant inactivation of any cell-free infectious HIV in hypotonic saliva could eventually contribute to the absence of infectious HIV and the presence of noninfectious HIV nucleic acids in saliva.9,10
Other viruses, such as herpes simplex virus type 1, are shed as cell-free virus and their transmissibility by saliva indicates sufficient stability in saliva. However, cell-associated transmission by other viruses (eg, human T-lymphotropic virus type 1 [HTLV-1]) probably should be studied.
Although infectious HIV is normally not present in the mouth of most patients, the present findings suggest circumstances under which infectious HIV may occur and thereby put contacts and health care workers at risk. The hypotonic lysis of shed-infected leukocytes by saliva would be expected to be overcome by an unusually large volume of isotonic blood in the mouth. We showed this possibility experimentally in vitro, since a 25% to 75% dilution of saliva in isotonic culture medium (equivalent to 0.5 to 4.5 mL of shed blood) partially inhibited cell lysis by saliva and more than 75% dilution (>4.5-mL blood) fully inhibited the cell lysis. The partly and fully protected leukocytes would be expected to permit virus transmission. A possible example may be the case report of oral transmission of HIV by the bite of a patient who was experiencing heavy bleeding in the mouth.84
Another instance relates to mononuclear leukocyte shedding into crevicular fluid.85 The volume of crevicular fluid, if it mixes with saliva, would be much too small to overcome the hypotonic effect of the 0.75 mL of saliva in the mouth and may explain why it has not been possible to isolate HIV from crevicular fluid that is mixed with saliva.86 In addition, during dental treatment, infected cells in shed blood, exudate, and crevicular fluid would be expected to be lysed not only by saliva but also by any water used for irrigation and rinsing.2,65 However, it would be potentially hazardous to irrigate the mouth with isotonic solutions because this might allow the survival of HIV-infected leukocytes. Deposition of any viable HIV-infected leukocytes (eg, aerosol during saline irrigation) into the mouth of a recipient would not be expected to transmit infection, because the recipient's saliva would kill the infected cells. However, deposition into the eye or nasal cavity might permit transmission because the isotonic tears87 and nasal secretions88,89 would not kill infected cells.
Another condition that may lead to oral shedding of infectious HIV is the high titer of HIV during both the initial stage of infection24,25 and the late-stage disease.23,52,90- 93 These studies report a strongly increased titer of both cell-free and cell-associated infectious HIV in shed blood, seminal fluid, and exudate during the acute and late infection.
Still another condition of oral transmission is breast-feeding, where 15% transmission from HIV-infected mothers to infants is tentatively reported.94- 98 This may be due to the known viability of HIV-infected or normal leukocytes in milk99- 102 that would be expected to resist salivary lysis during breast-feeding, because, as estimated earlier, volumes of more than 0.25 to 2.25 mL of milk would partially or fully reconstitute the tonicity of the baby's hypotonic saliva. Analogously, HTLV-1 leukemia virus is transmitted by milk leukocytes which, like HIV-infected mononuclear leukocytes, can attach to and infect epithelial cells.16,30- 32,103 Freeze-thawing of HTLV-1 milk from HTLV-1–infected mothers would be expected to kill infected cells104 and prevent transmission. Such freeze-thawing is reported to have prevented transmission to 13 infants over a 12-month observation period.105 Therefore, freezing and heating of HIV-infected human milk have been suggested as procedures to reduce HIV transmission to infants.95,106
Another condition under which HIV may be transmitted orally is via seminal fluid during oral sex.107- 112 Infected mononuclear leukocytes are the major infectious component of donor's seminal fluid.14- 16,80,81 These leukocytes are reported to survive in the isotonic seminal fluid113- 115 deposited in the recipient's vagina. As noted earlier, the infected mononuclear leukocytes in the seminal fluid can penetrate the vaginal epithelium and transmit infection to subepithelial leukocytes, or attach to and infect CD4-negative epithelial cells.16,30- 37 Speculatively, in the mouth, pharynx, and esophagus, oral sex may permit transmission if the volume of donor seminal fluid is sufficient to protect infected leukocytes from hypotonic lysis by restoring tonicity to recipient's saliva. Such protection of leukocytes may actually occur during oral sex, since the volume of donor seminal fluid (2.6-3.8 mL)116,117 deposited in the recipient's mouth would dilute the residual volume of 0.75 mL of saliva66 sufficiently to protect the leukocytes from lysis. In addition, viscous seminal fluid may protect infected leukocytes by physically excluding saliva. As noted earlier, there is growing epidemiological evidence to support oral transmission by oral sex.107- 112
The effective salivary defense may be medically applicable. Successful transmission of HIV by infected cells in seminal fluid and milk appears to be dependent on their protection by the seminal fluid and milk, against hypotonic lysis by saliva and against lysis by fecal material (unpublished data). This protection of HIV-infected cells may be overcome by medical application of anticellular substances such as hypotonic water, surfactants, and detergents (unpublished data).
Taken together, the present findings indicate that the rarity of transmission of HIV from the oral cavity may be due not only to the reported, nonspecific salivary inhibitors but also largely due to hypotonic saliva-induced disruption of infected cells. This conclusion is based on the limited inhibition of HIV by the reported salivary inhibitors (2- to 5-fold) compared with the 10,000-fold or higher inhibition by salivary lysis of infected leukocytes. The findings also indicate that this oral protection by hypotonicity may be reduced in situations where the hypotonicity is overcome by ingested isotonic solutions, such as milk and seminal fluid, as well as severe oral bleeding. Furthermore, oral transmission to contacts and health care workers may be increased under special circumstances such as (1) acute HIV infection where cell-free HIV is high, (2) advanced acquired immunodeficiency syndrome where cell-associated HIV is high, (3) oral sex where the seminal fluid protects the infected cells against saliva, and (4) saline irrigation of the mouth that may overcome the hypotonicity of saliva and create aerosols.
Accepted for publication May 18, 1998.
We acknowledge the suggestions made by our colleagues Bruce Baum, DDS, Ferdinando Dianzani, MD, Phillip Fox, DDS, Armond Goldman, MD, Abner Notkins, MD, Bellur Prabhakar, PhD, Luis Reuss, MD, and Sharon Wahl, PhD, who read the manuscript. We also acknowledge the valuable statistical analyses by Elbert Whorton, PhD.
Reprints: Samuel Baron, MD, Department of Microbiology and Immunology, University of Texas Medical Branch, 301 University Blvd, 499M 4.142D, MRB, Galveston, TX 77555-1019.