Identification of a novel surface protein on activated CD4+ T cells that induces contact-dependent B cell differentiation (help).

CD4+ T lymphocytes provide contact-dependent stimuli to B cells that are critical for the generation of specific antibody responses in a process termed T helper function. The surface structures on activated CD4+ T cells that mediate this function are not fully known. We previously reported the isolation of a functionally unique subclone of the Jurkat leukemic T cell line (D1.1) that constitutively expressed contact-dependent helper effector function. To identify T cell surface molecules that mediate contact-dependent T helper function, a monoclonal antibody (mAb), designated 5c8, was generated that inhibits D1.1-mediated B cell activation and immunoprecipitates a novel 30-kD protein structure from surface-iodinated D1.1 cells. Normal CD4+ T cells express 5c8 antigen (Ag) transiently 5-6 h after activation by phorbol myristate acetate and phytohemagglutinin with maximal expression 5-6 h after activation and absence of expression by 24 h. In contrast, neither resting nor activated CD8+ T cells express 5c8 Ag. In functional studies, mAb 5c8 inhibits the ability of fixed, activated CD4+ T cells to induce B cell surface CD23 expression. In addition, mAb 5c8 inhibits the ability of CD4+ T cells to direct terminal B cell differentiation driven by pokeweed mitogen. Taken together, these data suggest that 5c8 Ag is a novel, activation-induced surface T cell protein that is involved in mediating a contact-dependent element of the helper effector function of CD4+ T lymphocytes.

Involvement of CRAF1, a relative of TRAF, in CD40 signaling.

CD40 is a receptor on the surface of B lymphocytes, the activation of which leads to B cell survival, growth, and differentiation. A yeast two-hybrid screen identified a gene, CRAF1, encoding a protein that interacts directly with the CD40 cytoplasmic tail through a region of similarity to the tumor necrosis factor-alpha (TNF-alpha) receptor-associated factors. Overexpression of a truncated CRAF1 gene inhibited CD40-mediated up-regulation of CD23. A region of CRAF1 was similar to the TNF-alpha receptor-associated factors TRAF1 and TRAF2 and so defined a shared TRAF-C domain that was necessary and sufficient for CD40 binding and homodimerization. The CRAF1 sequence also predicted a long amphipathic helix, a pattern of five zinc fingers, and a zinc ring finger. It is likely that other members of the TNF receptor superfamily use CRAF-related proteins in their signal transduction processes.

Non-antigen signals for B-cell growth and differentiation to antibody secretion.

Recently, significant progress had been made in understanding the T-B lymphocyte interactions that control humoral immunity. This review highlights experiments that demonstrate a central role for interactions between T-cell-B-cell-activating molecule (CD40 ligand) expressed on T cells and CD40 on B cells in B-cell activation and immunoglobulin isotype switching, both in vitro and in vivo.

2 A crystal structure of an extracellular fragment of human CD40 ligand.

BACKGROUND: The CD40 ligand (CD40L) is a member of the tumor necrosis factor (TNF) family of proteins and is transiently expressed on the surface of activated T cells. The binding of CD40L to CD40, which is expressed on the surface of B cells, provides a critical and unique pathway of cellular activation resulting in antibody isotype switching, regulation of apoptosis, and B cell proliferation and differentiation. Naturally occurring mutations of CD40L result in the clinical hyper-IgM syndrome, characterized by an inability to produce immunoglobulins of the IgG, IgA and IgE isotypes. RESULTS: We have determined the crystal structure of a soluble extracellular fragment of human CD40L to 2 A resolution and with an R factor of 21.8%. Although the molecule forms a trimer similar to that found for other members of the TNF family, such as TNF alpha and lymphotoxin-alpha, and exhibits a similar overall fold, there are considerable differences in several loops including those predicted to be involved in CD40 binding. CONCLUSIONS: The structure suggests that most of the hyper-IgM syndrome mutations affect the folding and stability of the molecule rather than the CD40-binding site directly. Despite the fact that the hyper-IgM syndrome mutations are dispersed in the primary sequence, a large fraction of them are clustered in space in the vicinity of a surface loop, close to the predicted CD40-binding site.

Metabolic scaling and twin placentas.

OBJECTIVE: Do monochorionic (MC) and/or dichorionic (DC) twins show allometric scaling between placental and birth weight (PW, BW)? METHODS: We extracted BW, PW, gestational age (GA) and cord insertion type from 52 MC to 310 DC twins to calculate ß. DC twins were analyzed as summed and as individuals if placentas were separate. RESULTS: Mean ß for MC (0.78 ± 0.02), DC summed (0.78 ± 0.02), and DC with separate placentas (0.77 ± 0.03 and 0.76 ± 0.04) all non-significant. GA, summed BWs, total PW, BW discordance, and cord insertion sites did not differ between twin types or with ß. CONCLUSION: MC and DC twins show allometric scaling similar to singletons.

CD20-induced B cell death can bypass mitochondria and caspase activation.

The apoptotic pathway activated by chimeric anti-CD20 monoclonal antibodies (rituximab, IDEC.C2B8) was analyzed using the Burkitt lymphoma cell line Ramos. Crosslinking of CD20 (CD20XL) induced apoptosis in Ramos cells, which involved loss of mitochondrial membrane potential (Deltapsi(m)), the release of cytochrome-c (cyt-c), and activation of caspases-9 and -3. Nevertheless, several lines of evidence showed that the apoptotic outcome did not depend on these events. First, under circumstances where Ramos cells display resistance to either CD95- or B cell receptor (BCR)-induced apoptosis, CD20XL-induced apoptosis was not affected, pointing to a distinct pathway. Second, the broad-spectrum caspase inhibitor zVAD-fmk prevented processing of caspase-9, -3 and PARP as well as DNA fragmentation, but did not block apoptosis as measured by annexin V staining, cell size and membrane integrity. Lastly, Bcl-2 overexpression blocked cyt-c release and the decrease in Deltapsi(m), and completely prevented CD95- or BCR-mediated apoptosis; however, it did not affect CD20XL-induced cell death. We conclude that although CD20XL can initiate the mitochondrial apoptosis pathway, CD20-induced apoptosis does not necessarily require active caspases and cannot be blocked by Bcl-2. Since most chemotherapeutic drugs require the activation of caspases to exert their cytotoxicity, these findings provide an important rationale for the use of CD20 mAbs in chemoresistant malignancies.

Fas ligand is constitutively secreted by prostate cancer cells in vitro.

LNCaP, DU145, and PC3 prostate carcinoma cells secrete the 27-kDa soluble Fas ligand (sFasL) into their local environment. sFasL arises from the 40-kDa membrane-bound form (mFasL), which can be found on the cell surface in the LNCaP line, as demonstrated by monoclonal antibody staining. mFasL was also found in extracts of all three cell lines, as demonstrated by Western blotting. FasL mRNA was detected not only in the cell lines, but in the normal prostate as well. sFasL protein could also be detected immunohistochemically in prostate secretions and in human semen. Cleavage of mFasL to sFasL could be inhibited by several matrix metalloprotease inhibitors without a change in the cellular levels of FasL. Prostate-derived sFasL is biologically active, as demonstrated by its induction of apoptosis in Fas-positive Ramos cells, which was detected by terminal deoxynucleotidyl transferase-mediated nick end labeling assay. Mitoxantrone induces cellular apoptosis in all three prostate cancer cell lines. Mitoxantrone treatment and doxorubicin treatment also cause up-regulation of Fas, the cell surface receptor for FasL, in LNCaP cells, but not in DU145 or PC3 cells. Furthermore, the up-regulation of Fas expression by mitoxantrone at a high concentration was potentiated by hydrocortisone. When FasL interacts with its Fas, the Fas-bearing cell undergoes apoptosis. When LNCaP cells were treated with mitoxantrone and incubated with an anti-FasL monoclonal antibody, apoptosis was partially blocked. This not only further suggests that the sFasL is biologically active, but that the up-regulation of Fas in the presence of sFasL accounts, in part, for the cytotoxicity of mitoxantrone.

Ligation of CD40 on fibroblasts induces CD54 (ICAM-1) and CD106 (VCAM-1) up-regulation and IL-6 production and proliferation.

CD40 was originally described as a functionally significant B cell surface molecule. However, CD40 is also expressed on monocytes, dendritic cells, epithelial cells, and basophils. We now report that synovial membrane (SM) or dermal fibroblasts also express cell surface CD40 in vitro. Fibroblast CD40 expression declines with increasing time in culture and recombinant interferon-gamma (rINF-gamma) induces fibroblast CD40 up-regulation. This effect of rINF-gamma is augmented by recombinant interleukin-1 alpha or recombinant tumor necrosis factor-alpha. CD40 expression on fibroblasts is functionally significant because CD40L-CD40 interactions induce SM fibroblast CD54 (intercellular adhesion molecule-1) and CD106 (vascular cell adhesion molecule-1) up-regulation. Moreover, ligation of CD40 augments IL-6 production by SM fibroblasts and induces fibroblasts to proliferate. In addition, rINF-gamma enhances the effect of CD40L-CD40 interactions on fibroblast proliferation. Taken together, these studies show that fibroblasts can express CD40, cytokines can regulate fibroblast CD40 expression, and CD40 ligation induces fibroblast activation and proliferation.

TRAF-3 interacts with p62 nucleoporin, a component of the nuclear pore central plug that binds classical NLS-containing import complexes.

The TRAF-3 gene encodes a number of splice-variant isoforms that function as adapter molecules in NF-kappaB signaling, in part by associating with the cytoplasmic tails of CD40 or other TNF-receptor (TNF-R) family members. To identify downstream molecules in TRAF-3 signaling, a yeast two-hybrid library was screened with a full-length TRAF-3 construct. Nine independent TRAF-3 interacting clones encoded fragments of p62 Nucleoporin (p62), a 522 amino acid (aa) component of the nuclear pore central plug, that is known to bind karyopherin-beta/classical-NLS import factor complexes. The interaction of p62 with TRAF-3 was specific, since p62 failed to interact with TRAF-2, -4, -5, or -6. Deletional analysis in yeast revealed that the p62:TRAF-3 interaction is mediated by a p62 carboxy (C)-terminal coiled-coil domain and TRAF-3's fifth zinc (Zn) finger and coiled-coil domain. In human 293 T cells, recombinant TRAF-3 or p62 specifically co-immunoprecipitates the other species. In addition, endogenous p62 co-precipitates over-expressed TRAF-3. The functional effects of over-expressing a TRAF-3 binding fragment, p62(aa 336-522) were studied on NF-kappaB-dependent, or control STAT1-dependent reporter activity in 293 T cells, either resting or after stimulation by CD40 or IFN-gamma, respectively. Over-expression of p62(aa 336-522) induces NF-kappaB activation in resting cells and augments CD40-induced NF-kappaB activation, but has no effect on control STAT1 reporter activity, either at baseline or after IFN-gamma induction. The finding that TRAF-3 binds p62, suggests that TRAF-3 may serve as an adapter molecule at the nuclear membrane, in addition to its known adapter function at the plasma membrane.

TRAF-3 mRNA splice-deletion variants encode isoforms that induce NF-kappaB activation.

Although TRAF-3 gene products are required for signaling in T-B cell collaboration, full-length TRAF-3 appears to lack signaling function in transient transfection assays that measure NF-kappaB activation. However, the TRAF-3 gene also encodes at least three mRNA splice-deletion variants that predict protein isoforms (delta25aa, delta52aa and delta56aa) with altered zinc (Zn) finger domains and unknown functional capacities. To determine whether TRAF-3 splice-deletion variants may transmit activating receptor signals to the nucleus, cDNAs for five additional splice-variant isoforms (delta27aa, delta83aa, delta103aa, delta130aa and delta221aa) were cloned from a TRAF-3+ lymphoma and the expression and function of each of the eight TRAF-3 splice-deletion variants was analyzed. Among the splice-deletion variants, TRAF-3 delta130 mRNA is expressed by tonsillar B cells and by each of a panel of B and T cell lines. TRAF-3 delta221 protein is expressed by tonsillar B cells and by each of the lymphocytic lines. The functional effect of over-expressing each TRAF-3 splice-deletion variant on NF-kappaB activation was studied in 293 T cells. Seven of the TRAF-3 splice-deletion variants, such as TRAF-3 delta130, induce substantial NF-kappaB-driven luciferase activity (80-500 fold). In contrast, TRAF-3 delta221 (in which the complete Zn finger domain is absent) fails to induce NF-kappaB activation. Although full-length TRAF-3 alone is inactive, it augments the functional effects of the seven activating TRAF-3 splice-deletion variants (1.4-5 fold). These data indicate that alterations of the Zn finger domains render the TRAF-3 splice-deletion variants capable of inducing NF-kappaB activation and that full-length TRAF-3 augments their signaling.

A single amino acid substitution in a common African allele of the CD4 molecule ablates binding of the monoclonal antibody, OKT4.

The CD4 molecule is a relatively non-polymorphic 55 kDa glycoprotein expressed on a subset of T lymphocytes. A common African allele of CD4 has been identified by non-reactivity with the monoclonal antibody, OKT4. The genetic basis for the OKT4- polymorphism of CD4 is unknown. In the present paper, the structure of the CD4 molecule from an homozygous CD4OKT4- individual was characterized at the molecular level. The size of the CD4OKT4- protein and mRNA were indistinguishable from those of the OKT4+ allele. The polymerase chain reaction (PCR) was used to map the structure of CD4OKT4- cDNAs by amplifying overlapping DNA segments and to obtain partial nucleotide sequence after asymmetric amplification. PCR was then used to clone CD4OKT4- cDNAs spanning the coding region of the entire, mature CD4 protein by amplification of two overlapping segments followed by PCR recombination. The nucleotide sequence of CD4OKT4- cDNA clones revealed a G----A transition at bp 867 encoding an arginine----tryptophan substitution at amino acid 240 relative to CD4OKT4+. Expression of a CD4OKT4- cDNA containing only this transition, confirmed that the arginine----tryptophan substitution at amino acid 240 ablates the binding of the mAb OKT4. A positively charged amino acid residue at this position is found in chimpanzee, rhesus macaque, mouse and rat CD4 suggesting that this mutation may confer unique functional properties to the CD4OKT4- protein.

Isolation of cDNAs encoding T-BAM, a surface glycoprotein on CD4+ T cells mediating contact-dependent helper function for B cells: identity with the CD40-ligand.

"T-cell B-cell Activating Molecule" (T-BAM) is an activation-induced surface protein on CD4+ T cells that mediates a contact-dependent signal for B cell differentiation and immunoglobulin (Ig) secretion. The T-BAM protein on a helper clone of Jurkat (D1.1) was affinity purified using the anti-T-BAM mAb, 5c8. The NH2-terminal amino acid sequence of purified T-BAM was determined and found to be highly homologous to the predicted NH2-terminal sequence of a T cell ligand to the B cell CD40 molecule (CD40-L). From a D1.1 cDNA library, a clone was isolated that encodes CD40-L by sequence and drives expression of T-BAM protein on transfected cells, demonstrating that the T-BAM and CD40-L genes and proteins are identical. Moreover, transfection of T-BAM was shown to confer to non-lymphoid cells, the ability to induce B cells to upregulate the expression of surface CD23 molecules. In previous studies we showed that T-BAM was expressed predominantly on activated CD4+ and on few if any CD8+ cells. Although the current work confirms that T-BAM is largely restricted to activated CD4+ T cells, we now provide definitive evidence that T-BAM can be expressed by a small population of CD8+ T cells after activation. Importantly, a subset of CD8+ T cells do not express T-BAM after activation and this T-BAM- phenotype is maintained on certain CD8+ T cell clones. Taken together, these data unify the biology and structure of T-BAM and CD40-L and this synthesis has implications for understanding the T cell regulation of the humoral immune response.

A single gene for human TRAF-3 at chromosome 14q32.3 encodes a variety of mRNA species by alternative polyadenylation, mRNA splicing and transcription initiation.

Human TRAF-3 is a signaling molecule that interacts with the cytoplasmic tails of CD40 and other TNF-receptor family members. TRAF-3 mRNA is expressed as two major classes of approximately 2 and 8 kb and a number of TRAF-3 encoding cDNA clones differ in discrete gene segments. Because this variety of mRNA species could result from mRNA processing events and/or multiple genes, the structure and localization of TRAF-3 encoding gene elements were determined. FISH and radiation hybrid mapping demonstrated that TRAF-3 is located at chromosome 14q32.3, approximately 1 Mb centromeric to the Ig heavy chain gene complex. Physical mapping of four overlapping genomic PAC clones established that TRAF-3 transcripts are encoded by a single gene, comprised of 13 exons and spanning 130 kb. Alternative polyadenylation in the mRNA segment encoded by exon 12 accounts for the difference between the 2 kb and the 8 kb classes of transcripts. Alternative mRNA splicing in the coding region (encoded by exons 3-12) generates transcripts which delete exons 8 (75 nt), 7+8 (156 nt) or 8+9 (168 nt) and that encode distinct protein isoforms (delta25, delta52 and delta56 aa, respectively). Alternative splicing of exon 2 (139 nt) and alternative transcriptional initiation result in mRNA species with distinct 5'UTRs. Together, these data indicate that a single TRAF-3 gene encodes a variety of mRNA species by a combination of alternative polyadenylation, alternative mRNA splicing and/or alternative initiation.

Variability of water content of lean tissue of pregnant and nonpregnant rats and its effect on body fat estimation.

In humans, body fat is often estimated from body water. Measurements of the body water and estimates of the percent water in the lean tissue (WLT) are used to calculate the lean tissue which is then subtracted from body weight to estimate the quantity of body fat. The reliability of this method and the effect that pregnancy, obesity, fasting, and food restriction have on the body fat estimate derived from the mean WLT were assessed. Pregnant and nonpregnant rat carcasses were analyzed gravimetrically for water and for hexane extractable fat and the WLT was calculated. Mean WLT values differed significantly in pregnant and nonpregnant groups. In pregnant animals, prior food restriction or fasting significantly lowered the mean WLT. Estimates of individual fat content derived from the appropriate mean WLT varied from 63 to 332% of assay values. Thus body fat estimates derived from the mean WLT may be very inaccurate, particularly in fasted individuals.

The effects of exercise on milk yield, milk composition, and offspring growth in rats.

This study examined effects of exercise training throughout pregnancy and lactation upon selected indicators of lactational performance in Wistar rats. During the 7 wk prior to mating, experimental female rats were gradually trained to swim for 2 h/day, 5 days/wk with a 3% tail weight; they continued swimming until the 19th day of pregnancy. Control animals remained sedentary. Swimming resumed during days 2-14 of lactation. Food intake was greater in exercised rats during the lactation period. Body weights of the two groups were comparable. Mild yield (day 10), energy content (day 15), and protein and fat concentrations did not differ, but milk of exercised rats had a lower lactose concentration. The exercise regimen had no statistically significant effect on litter size or on offspring weight to day 15. This study indicates that moderate-intensity, aerobic exercise does not markedly affect the lactational performance of ad libitum-fed rats.

Effects of protein and carbohydrate supplements on fetal and maternal weight and on body composition in food-restricted rats.

The effects of equal amounts of carbohydrate or protein supplements given from day 15 of gestation after a 50% food restriction for the previous 10 days were studied in the rat. Maternal body weight and composition and fetal and placental weights of supplemented rats were compared with controls, fed ad libitum, and with rats maintained on a 50% food restriction until term. The control group had the highest gain in net maternal body weight followed by the protein supplemented and the carbohydrate supplemented rats. In contrast, the restricted rats had a small loss in body weight. Increments in net body weight in the protein supplemented rats were considerably higher than in carbohydrate-supplemented rats. Similarly, the amount of water and lean dry tissue was also higher in the protein-supplemented rats although both values were still significantly lower than controls. Both types of supplements increased fetal weight equally compared with the restricted group but not to the control levels. Placental weight was higher in the protein supplemented group compared with the carbohydrate supplemented group but placental weight did not reach the control level.

Anthropometric equations for studying body fat in pregnant women.

Anthropometric data from 200 pregnant women were used to estimate body fat at gestation weeks 14 and 37 and changes in body fat from week 14 to week 37 with four formulas from the literature. The resulting estimates were evaluated against the estimation of fat by a four-compartment model that determined fat from weight, total body water, bone mineral mass, and body density. The estimates of fat by existing anthropometric models were statistically different from those by the four-compartment model in both early and late pregnancy. Most importantly, the change in body fat estimated by the anthropometric models (all > 4 kg) was considerably higher than that estimated by the four-compartment model (3.3 kg). Two new anthropometric equations were developed, both of which used the four-compartment model as the reference method. The equation for predicting change in fat mass from week 14 to 37 of pregnancy was as follows: 0.77 (change in weight, kg)+ 0.07 (change in thigh skinfold thickness, mm)-6.13 (r2 = 0.73). The equation for determining fat (kg) at term was as follows: 0.40 (weight at week 37, kg)+ 0.16 (biceps skinfold thickness at week 37, mm) + 0.15 (thigh skinfold thickness at week 37, mm)-0.09 (wrist circumference at week 37. mm)+ 0.10 (prepregnancy weight.kg)-6.56 (r2 = 0.89). Both equations were derived on a randomly selected half of the total sample and validated on the remaining half. Both equations were found to be valid for use in studying pregnant women with different prepregnancy body mass indexes, different gestational weight gains, different ethnicities, and different socioeconomic status.

Human immunodeficiency virus type-1 infection of homosexual men is accompanied by a decrease in circulating B cells.

As part of the multidisciplinary effort to characterize the natural history of human immunodeficiency virus type 1 (HIV-1) infection, the cell-surface phenotypes of lymphocytes from a cohort of homosexual men were analyzed in detail and related to clinical and laboratory parameters associated with HIV-1 infection. The present study represents a cross-sectional analysis of coded specimens from 153 homosexual men, of whom 74 were seronegative and 79 seropositive for HIV-1. Fewer circulating B lymphocytes (CD19+) were found in HIV-1-seropositive subjects relative to a seronegative reference group. HIV seropositivity was not associated with decreased numbers of CD8+ T cells or activated T cells, which suggests that the number of circulating B cells specifically decreased. In addition to CD19, B cells were measured by CD20 and CD21 in a subset of subjects, and decreases in circulating CD20+ and CD21+ B cells were also apparent in HIV-1-seropositive subjects. The decrease in B-cell numbers was present at the earliest stages of HIV-1 infection (asymptomatic, clinically silent) and became more pronounced at more advanced stages of HIV-1 infection. The absolute B-cell numbers correlated with absolute CD4+ cell numbers (r = 0.59, p less than 0.001). These data suggest that HIV-1 infection is associated with progressive, selective decreases in the numbers of circulating CD4+ T cells and B cells.

Induction of xenoreactive CD4+ T-cell anergy by suppressor CD8+CD28- T cells.

BACKGROUND: The underlying mechanism of immune suppression mediated by regulatory T cells is not completely understood. In previous studies we have shown that antigen-specific human T suppressor cells (Ts) can be generated in vitro by multiple rounds of stimulation with allogeneic, xenogeneic, or antigen-pulsed autologous antigen-presenting cells (APC). Human Ts express the CD8+CD28- phenotype and require specific recognition of MHC class I/peptide complexes on the surface of APC to block proliferation of T helper cells (Th). The aim of the present study was to explore the activation requirements of Ts as well as the nature of Th unresponsiveness to xenogeneic (swine) antigens induced by Ts. METHODS AND RESULTS: We investigated whether specific antigenic stimulation of Ts is required for their ability to inhibit early activation of xenoreactive Th (up-regulation of CD40 ligand). Flow cytometry studies indicated that Ts function required specific recognition of MHC class I on the surface of the stimulating APC. However, neither proliferation nor protein synthesis was required for the ability of Ts to inhibit Th. Ts drastically reduced the capacity of xenoreactive Th cells to produce interleukin (IL)-2 in response to the specific APC, without affecting their surface expression of IL-2 receptor. The suppressor effect that Ts exerted on Th proliferation could not be circumvented by CD40 ligation on the surface of the APC but could be reversed by the addition of exogenous IL-2. CONCLUSION: These data indicate that Ts induce anergy of xenoreactive human Th cells upon specific recognition of MHC class I antigens. Hence, Ts may prevent the activation of T cell-mediated immune responses against xenogeneic transplants.

Estimating infant mortality from human immunodeficiency virus and other causes in breast-feeding and bottle-feeding populations.

This paper discusses the literature dealing with breast-feeding in women with human immunodeficiency virus (HIV) infection. This review is used to develop a background for a quantitative assessment of factors determining whether breast-feeding or bottle-feeding will result in a lower overall mortality in areas of different HIV prevalence. An algebraic formula is presented that enables calculation of the mortality that would result in any population if the following variables are known or capable of estimation: portion of childbearing women infected with HIV, portion of newborns infected with HIV at birth, relative mortality of breast-fed and bottle-fed infants that are not HIV infected, and transmissibility of HIV during breast-feeding. The information available for estimating these variables is incomplete. Nevertheless, boundaries can be set for each variable, based on information obtained locally or from the literature, to aid in objective evaluation of the risks of promoting bottle-feeding or breast-feeding in populations with different risk characteristics. Consideration of these factors indicates that the benefits of breast-feeding over bottle-feeding can substantially outweigh any putative risk of HIV transmission during breast-feeding unless the prevalence of HIV infection is quite high or the difference in mortality of breast-fed and bottle-fed infants is very low.