Comparison of human lung cancer/SCID mouse tumor xenografts and cell culture growth with patient clinical outcomes.

Leukemic cell growth in SCID mice has been reported as a predictor of disease relapse. However, there is a paucity of literature regarding xenograft growth and clinical outcomes in non-small cell lung cancer (NSCLC). Seventy-nine specimens from patients with NSCLC were either subcutaneously implanted into SCID mice and/or placed in tissue culture. Retrospective chart review was correlated with stage, histology, necrosis, disease-free interval, and survival. Tumor xenografts were successfully established with 17 of 37 (46%) tumor biopsy tissues. Thirteen of 59 (22%) specimens grew in cell culture. Patients whose tumors grew in SCID mice had no difference in survival compared to those with no growth ( n=20, p=0.42). Median survival was 36 months in 13 patients whose tumors grew in cell culture compared to 39 months in 46 patients without growth. Eight of 12 (67%) patients with metastasis showed SCID/human xenograft growth, whereas nine of 25 (36%) without metastases did so ( p=0.08). Growth of tumor cells in vitro occurred in 11 of 31 (35%) adenocarcinomas, one of 25 (4%) squamous cell carcinomas, and one of three (33%) large cell carcinomas ( p=0.02). Well or moderately differentiated tumors grew in cell culture in only two of 22 (9%), whereas poorly or undifferentiated tumors grew in 11 of 32 (34%) cases ( p=0.03). We conclude that neither the ability of a tumor to engraft and grow in SCID mice nor its ability to grow in vitro in cell culture is a reliable predictor of disease outcome or survival in patients with NSCLC. The ability to propagate tumors in vitro appears to be more dependent upon the histological type of tumor and its degree of differentiation.

Antitumor efficacy of a human interleukin-12 expression plasmid demonstrated in a human peripheral blood leukocyte/human lung tumor xenograft SCID mouse model.

Genes encoding the p35 and p40 subunits of human interleukin-12 (IL-12) and the bacterial aminoglycoside phosphotransferase were cloned into a mammalian expression plasmid. The resultant plasmid, pCMVIL-12neo, was used to transfect human lung tumor cell lines in vitro. Stably transfected subclones were generated and found to secrete human IL-12 for at least 10 days following a lethal dose of gamma-radiation. The ability of the IL-12--producing tumor cells to promote an antitumor response in vivo was evaluated in SCID mice co-engrafted subcutaneously with human peripheral blood lymphocytes (PBLs) and viable human lung tumor cells (SCID-Winn assay). Using this model system, it was established that IL-12 released locally into tumors by irradiated IL-12--transfected cells activated the human PBL and promoted their ability to suppress tumor development in a dose-dependent fashion. PBL subset depletion studies revealed that the antitumor effect promoted by the IL-12--modified cells was dependent on the presence of human CD8(+) T cells and, to a lesser extent, human CD56(+) natural killer cells within the xenograft. We conclude that (a) irradiated human lung tumor cells genetically modified with pCMVIL-12neo secrete bioactive human IL-12 at concentrations sufficient to promote a human lymphocyte-mediated antitumor response in the microenvironment of the xenograft, and (b) that the SCID-Winn assay provides a useful model for the preclinical evaluation of cytokine-based human immunotherapy protocols.

Human-SCID mouse chimeric models for the evaluation of anti-cancer therapies.

The ability to engraft human tumors and human immunocompetent cells successfully in severe combined immunodeficient (SCID) mice has spawned the development and use of human-mouse chimeric models to evaluate anti-cancer therapies. The lack of standardization and many other potential pitfalls have contributed to the current controversy surrounding the reliability of these different models. Five frequently used SCID mouse models and their specific applications are summarized with the specific aim of providing an objective discussion of the strengths and limitations of each model, together with suggestions for overcoming some of the variabilities and for improving the design and use of future models.

Cytokines delivered by biodegradable microspheres promote effective suppression of human tumors by human peripheral blood lymphocytes in the SCID-Winn model.

A new technology for the local and sustained delivery of immunostimulatory molecules to the tumor environment for cancer immunotherapy was evaluated. The ability of cytokines delivered by biodegradable microspheres to promote the antitumor activity of human peripheral blood lymphocytes (PBL) was tested in a human PBL, human tumor, and SCID mouse (SCID-Winn) model. Co-engraftment of human recombinant IL-12-loaded microspheres with human PBL and tumors in SCID mice promoted complete tumor suppression in as many as 100% of the mice, whereas microspheres loaded with polyethyleneglycol-interleukin-2 suppressed but did not eliminate the growth of tumor xenografts. Control microspheres (loaded with bovine serum albumin) in the presence of human PBL or cytokine-loaded microspheres in the absence of human PBL had no tumor-suppressive effect. Coincident with the enhancement of the human PBL-mediated antitumor activity in mice treated with IL-12-loaded microspheres was the production and release of human IFN-gamma indicating that IL-12 released from the microspheres results in the activation of the engrafted human PBL. The results establish that biodegradable microspheres represent an effective tool for the local and sustained delivery of cytokines to the tumor environment for cancer immunotherapy.

2-Methyl-6-(phenylethynyl)-pyridine (MPEP), a potent, selective and systemically active mGlu5 receptor antagonist.

In the present paper we describe 2-methyl-6-(phenylethynyl)-pyridine (MPEP) as a potent, selective and systemically active antagonist for the metabotropic glutamate receptor subtype 5 (mGlu5). At the human mGlu5a receptor expressed in recombinant cells, MPEP completely inhibited quisqualate-stimulated phosphoinositide (PI) hydrolysis with an IC50 value of 36 nM while having no agonist or antagonist activities at cells expressing the human mGlu1b receptor at concentrations up to 30 microM. When tested at group II and III receptors, MPEP did not show agonist or antagonist activity at 100 microM on human mGlu2, -3, -4a, -7b, and -8a receptors nor at 10 microM on the human mGlu6 receptor. Electrophysiological recordings in Xenopus laevis oocytes demonstrated no significant effect at 100 microM on human NMDA (NMDA1A/2A), rat AMPA (Glu3-(flop)) and human kainate (Glu6-(IYQ)) receptor subtypes nor at 10 microM on the human NMDA1A/2B receptor. In rat neonatal brain slices, MPEP inhibited DHPG-stimulated PI hydrolysis with a potency and selectivity similar to that observed on human mGlu receptors. Furthermore, in extracellular recordings in the CA1 area of the hippocampus in anesthetized rats, the microiontophoretic application of DHPG induced neuronal firing that was blocked when MPEP was administered by iontophoretic or intravenous routes. Excitations induced by microiontophoretic application of AMPA were not affected.

SIB-1757 and SIB-1893: selective, noncompetitive antagonists of metabotropic glutamate receptor type 5.

Cell lines expressing the human metabotropic glutamate receptor subtype 5a (hmGluR5a) and hmGluR1b were used as targets in an automated high-throughput screening (HTS) system that measures changes in intracellular Ca2+ ([Ca2+]i) using fluorescence detection. This functional screen was used to identify the mGluR5-selective antagonist, SIB-1757 [6-methyl-2-(phenylazo)-3-pyridinol], which inhibited the glutamate-induced [Ca2+]i responses at hmGluR5 with an IC50 of 0.37 microM compared with an IC50 of >100 microM at hmGluR1. Schild analysis demonstrated a noncompetitive mechanism of inhibition. Pharmacophore mapping was used to identify an additional compound, SIB-1893 [(E)-2-methyl-6-(2-phenylethenyl)pyridine], which was also shown to block glutamate-induced increases in [Ca2+]i at hmGluR5 with an IC50 of 0.29 microM compared with an IC50 of >100 microM at hmGluR1. SIB-1757 and SIB-1893 showed little or no activity when tested for agonist and antagonist activity at the other recombinant human mGluR subtypes, alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid, kainate, and N-methyl-D-aspartate receptors. In rat neonatal brain slices, SIB-1757 and SIB-1893 inhibited (S)-3,5-dihydroxyphenylglycine (DHPG)-evoked inositol phosphate accumulation in hippocampus and striatum by 60% to 80%, with a potency similar to that observed on recombinant mGluR5. However, in the cerebellum, a brain region with low mGluR5 expression, SIB-1757 failed to inhibit DHPG-evoked inositol phosphate accumulation. In cultured rat cortical neurons, SIB-1757 and SIB-1893 largely inhibited DHPG-evoked [Ca2+]i signals, revealing a population of neurons that were less sensitive to SIB-1757 and SIB-1893. This is the first description of highly selective, noncompetitive mGluR5 antagonists. These compounds will be useful tools in evaluating the role of mGluR5 in normal physiology and in animal models of disease.

(R,S)-4-phosphonophenylglycine, a potent and selective group III metabotropic glutamate receptor agonist, is anticonvulsive and neuroprotective in vivo.

Group III metabotropic glutamate receptors (mGluRs) are thought to modulate neurotoxicity of excitatory amino acids, via mechanisms of presynaptic inhibition, such as regulation of neurotransmitter release. Here, we describe (R,S)-4-phosphonophenylglycine (PPG) as a novel, potent, and selective agonist for group III mGluRs. In recombinant cell lines expressing the human receptors hmGluR4a, hmGluR6, hmGluR7b, or hmGluR8a, EC50 values for (R,S)-PPG of 5.2 +/- 0.7 microM, 4.7 +/- 0.9 microM, 185 +/- 42 microM, and 0.2 +/- 0.1 microM, respectively, were measured. The compound showed EC50 and IC50 values of >/=200 microM at group I and II hmGluRs and was inactive at cloned human N-methyl-D-aspartate, alpha-amino-3-hydroxy-5-methyl-isoxazole-4-propionate, and kainate receptors (>300 microM). On the other hand, it showed micromolar affinity for a Ca2+/Cl--dependent L-glutamate binding site in rat brain, similar to other phosphono-substituted amino acids like L-2-amino-4-phosphonobutyrate. In cultured cortical neurons, (R, S)-PPG provided protection against a toxic pulse of N-methyl-D-aspartate (EC50 = 12 microM), which was reversed by the group III mGluR antagonist (R,S)-alpha-methylserine-O-phosphate but not by the group II antagonist (2S)-alpha-ethylglutamate. Moreover, (R,S)-PPG protected against N-methyl-D-aspartate- and quinolinic acid-induced striatal lesions in rats and was anticonvulsive in the maximal electroshock model in mice. In contrast to the group III mGluR agonists L-2-amino-4-phosphonobutyrate and L-serine-O-phosphate, (R,S)-PPG showed no proconvulsive effects (2200 nmol i.c.v.). These data provide novel in vivo evidence for group III mGluRs as attractive targets for neuroprotective and anticonvulsive therapy. Also, (R,S)-PPG represents an attractive tool to analyze the roles of group III mGluRs in nervous system physiology and pathology.

The human N-methyl-D-aspartate receptor 2C subunit: genomic analysis, distribution in human brain, and functional expression.

cDNAs encoding four isoforms of the human NMDA receptor (NMDAR) NMDAR2C (hNR2C-1, -2, -3, and -4) have been isolated and characterized. The overall identity of the deduced amino acid sequences of human and rat NR2C-1 is 89.0%. The sequences of the rat and human carboxyl termini (Gly925-Val1,236) are encoded by different exons and are only 71.5% homologous. In situ hybridization in human brain revealed the expression of the NR2C mRNA in the pontine reticular formation and lack of expression in substantia nigra pars compacta in contrast to the distribution pattern observed previously in rodent brain. The pharmacological properties of hNR1A/2C were determined by measuring agonist-induced inward currents in Xenopus oocytes and compared with those of other human NMDAR subtypes. Glycine, glutamate, and NMDA each discriminated between hNR1A/2C-1 and at least one of hNR1A/2A, hNR1A/2B, or hNR1A/2D subtypes. Among the antagonists tested, CGS 19755 did not significantly discriminate between any of the four subtypes, whereas 5,7-dichlorokynurenic acid distinguished between hNR1A/2C and hNR1A/2D. Immunoblot analysis of membranes isolated from HEK293 cells transiently transfected with cDNAs encoding hNR1A and each of the four NR2C isoforms indicated the formation of heteromeric complexes between hNR1A and all four hNR2C isoforms. HEK293 cells expressing hNR1A/ 2C-3 or hNR1A/2C-4 did not display agonist responses. In contrast, we observed an agonist-induced elevation of intracellular free calcium and whole-cell currents in cells expressing hNR1A/2C-1 or hNR1A/2C-2. There were no detectable differences in the macroscopic biophysical properties of hNR1A/2C-1 or hNR1A/2C-2.

Pharmacological characterization of the human ionotropic glutamate receptor subtype GluR3 stably expressed in mammalian cells.

We have cloned the human ionotropic alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor GluR3 flip splice variant (hGluR3i) and developed a stable cell line expressing this receptor in HEK293 cells. Electrophysiological recordings demonstrated that glutamate-evoked currents desensitize rapidly, with a mean desensitization time constant of 5.4 ms. Robust glutamate-evoked increases in intracellular Ca++ ([Ca++]i) were observed in the presence of cyclothiazide, which attenuated receptor desensitization. [Ca++]i measurements were used to perform a detailed pharmacological characterization of hGluR3i with reference agonists and antagonists. The results of these studies showed that kainate and domoate were not fully efficacious agonists relative to glutamate. The binding affinities of agonists and competitive antagonists were determined in a [3H]AMPA competition binding assay. There was a good correlation between the functional data and the binding affinities obtained for competitive antagonists. However, the binding affinities of the agonists did not correlate with their functional EC50 values from [Ca++]i data, possibly because the binding assay predominantly measures the desensitized high-affinity state of the receptor. [3H]AMPA binding also was performed on membranes prepared from rat forebrain, and comparison of the data from HEK293 cells expressing hGluR3i and rat forebrain suggest that nearly all of the reference compounds show similar binding activities between the two membrane preparations, with the exception of fluoro-willardiine, kainate and 6-nitro-7-sulfamoylbenzo(f)quinoxaline-2-3-dione (NBQX). These data suggest that cells stably expressing recombinant hGluR3i represent pharmacologically valid experimental systems to study human AMPA receptors.

Functional characterization of human N-methyl-D-aspartate subtype 1A/2D receptors.

The human NMDAR2D subunit was cloned, and the pharmacological properties of receptors resulting from injection of transcripts encoding human NMDAR1A and NMDAR2D subunits in Xenopus oocytes were characterized by profiling NMDA receptor agonists and antagonists. We found that glutamate, NMDA, glycine, and D-serine were significantly more potent on hNMDAR1A/2D than on hNMDAR1A/2A or hNMDAR1A/2B. Also, the potencies of NMDA and glycine were higher for hNMDAR1A/2D than for hNMDAR1A/2C. Ifenprodil was more potent at hNMDAR1A/2B than at hNMDAR1A/2D, whereas 5,7-dichlorokynurenate was more potent at hNMDAR1A/2A than at hNMDAR1A/2D. As measured in transiently transfected human embryonic kidney 293 cells, the maximal inward current in the presence of external Mg2 occurred at -40 mV, and full block was not observed at negative potentials. Kinetic measurements revealed that the higher affinity of hNMDAR1A/2D for both glutamate and glycine relative to hNMDAR1A/2A and hNMDA1A/2B can be explained by slower dissociation of each agonist from hNMDAR1A/2D. The hNMDAR1A/2D combination represents a pharmacologically and functionally distinct receptor subtype and may constitute a potentially important target for therapeutic agents active in the human CNS.

Stable expression and characterization of recombinant human heteromeric N-methyl-D-aspartate receptor subtypes NMDAR1A/2A and NMDAR1A/2B in mammalian cells.

The electrophysiological and pharmacological properties of two mammalian cell lines stably transfected with cDNAs encoding recombinant human N-methyl-D-aspartate (NMDA) receptor subtypes NMDAR1A/2A and NMDAR1A/2B are described. In whole-cell electrophysiological recordings, application of NMDA/glycine elicited inward currents at negative holding potentials in human NMDAR1A/2A (hNMDAR1A/2A)- and hNMDAR1A/2B-expressing cells. The current-voltage relationships determined in both cell lines in the presence and absence of external Mg++ were similar to those observed with recombinant rat NMDA receptors. Power spectra calculated from NMDA/glycine-induced currents for both NMDA receptor-expressing cell lines suggested a kinetically homogeneous population of channels. Immunoprecipitation with an anti-NMDAR1A antibody coprecipitated the corresponding NMDAR2 subunit with the NMDAR1A, suggesting that heteromeric complexes are formed in these stable cell lines. Stimulation of NMDA receptors evoked an increase in intracellular Ca++, which was used to characterize their pharmacological properties. NMDA displayed less intrinsic activity than did glutamate in both NMDA receptor-expressing cell lines and was a 4-fold more potent agonist at hNMDAR1A/2B than hNMDAR1A/2A. NMDA/glycine-evoked increases in Ca++ levels were inhibited by CGS 19755, (+/-)-3-(2-carboxypiperazin-4-yl)propyl-1-phosphonate, MK-801, ketamine and ifenprodil. (+/-)-3-(2-Carboxypiperazin-4-yl)propyl-1-phosphonate was a 3-fold more potent antagonist at hNMDAR1A/2A than hNMDAR1A/2B, whereas ifenprodil was markedly more selective toward hNMDAR1A/2B, being 250-fold more potent than against hNMDAR1A/2A. These data suggest that cells stably expressing recombinant heteromeric hNMDAR1A/2A and hNMDAR1A/2B represent pharmacologically valid experimental systems to study human NMDA receptors.

Cloning and functional characterization of human heteromeric N-methyl-D-aspartate receptors.

Human cDNAs encoding N-methyl-D-aspartate receptor type (NMDAR)1A, NMDAR2A and NMDAR2B subunits were cloned and receptors encoded by these cDNAs were functionally expressed by injection of the respective mRNAs in Xenopus oocytes. The pharmacological properties of recombinant human N-methyl-D-aspartate (NMDA) receptors were characterized by profiling two agonists and four antagonists at both the NMDA and glycine sites in voltage-clamped oocytes. NMDA, glycine and D-serine were significantly more potent at human NMDAR (hNMDAR)1A/2B receptors than at nNMDAR1A/2A, whereas there was no detectable subtype-dependent difference in the potency of glutamate. Of the NMDA-site antagonists tested, CGP 43487 and 3-(2-carboxypiperazin-4-yl) propyl-1-phosphonate exhibited 5.8- and 3.9-fold greater potency, respectively, at hNMDAR1A/2A receptors than at hNMDAR1A/2B. Of the four glycine-site competitive antagonists tested, L-689,560 displayed 5-fold greater potency at hNMDAR1A/2A, whereas 5,7-dichlorokynurenic acid, HA-966 and CGP 58411 did not discriminate between hNMDAR1A/2A and hNMDAR1A/2B. Receptors resulting from injection of hNMDAR1A, hNMDAR2A and hNMDAR2B transcripts in a 1:1:1 ratio were indistinguishable from hNMDAR1A/2B receptors in terms of their sensitivity to NMDA, glycine, D-serine, CGS 19755 and CGP 40116. Ifenprodil was approximately 350-fold more potent at hNMDAR1A/2B than at hNMDAR1A/2A receptors. Ifenprodil sensitivities of receptors formed in oocytes injected with a constant amount of hNMDAR1A mRNA but varying ratios of hNMDAR2A or hNMDAR2B mRNAs were compared. The receptors expressed at a 10:1 ratio of 2A:2B transcripts displayed an ifenprodil sensitivity that would be predicted for a population in which 51% was represented by hNMDAR(1A)2(2A)3 complexes. Our results underscore the need for subtype-selective compounds acting at novel sites to sufficiently probe the pharmacological differences between NMDA receptor subtypes formed by different subunit combinations.

Molecular and functional characterization of recombinant human metabotropic glutamate receptor subtype 5.

We have isolated and characterized overlapping cDNAs that encode two isoforms of the human metabotropic glutamate receptor subtype 5 (hmGluR5). The deduced amino acid sequences of human and rat mGluR5a are 94.5% identical. However, a region in the putative cytoplasmic domain (SER926-ALA1121) displays significant sequence divergence. Genomic analysis of this region showed that the sequence divergence results from species-specific differences in the genomic sequences, not from alternative splicing. The distribution of mGluR5 mRNA in human brain was most strongly detected throughout the hippocampus, with moderate levels in the caudate-putamen, cerebral cortex, thalamus, and deep cerebellar nuclei, and at low levels in the cerebellar cortex. Activation of both hmGluR5a and hmGluR5b transiently expressed in Xenopus oocytes and HEK293 cells was coupled to inositol phosphate (InsP) formation and elevation of the intracellular free calcium ([Ca2+]i). The agonist rank order of potency for activating recombinant hmGluR5a receptors in either system was quisqualate > L-glutamate > 1S,3R-ACPD. Both the quisqualate stimulated InsP and [Ca2+]i were inhibited by (+)-MCPG. Recombinant human mGluR5a was also stably expressed in mouse fibroblast Ltk- cells, in which the efficacy and potency of quisqualate were unchanged for more than 30 cell passages.

Calcium oscillations in human T and natural killer cells depend upon membrane potential and calcium influx.

With the use of single-cell digital imaging of the fluorescent Ca2+ indicator dye fura-2 we investigated Ca2+ signaling in human T lymphocytes and NK cells during activation by a variety of stimuli. A low percentage of resting T cells or T cell blasts displayed oscillations in cytosolic Ca2+ when stimulated with the mitogenic lectin PHA or by the addition of OKT3 mAb followed by a secondary cross-linking antibody. Lymphokine-activated T killer cells were more responsive than resting cells. A comparison of PHA, cross-linked anti-CD3, and a heteroconjugate mAb showed that at least 20% of the cells from these T cell preparations oscillated. Addition of PHA or cross-linked anti-CD16 caused NK cells to oscillate. In contrast, thapsigargin, a microsomal ATPase blocker, resulted in a relatively uniform, slowly rising and sustained Ca2+ response in all cell types studied. The maintenance of both thapsigargin- and receptor-induced responses required Ca2+ influx driven by a negative membrane potential. Because Ca2+ oscillations occurred in response to stimuli which mimic the normal activation of lymphocytes, and inasmuch as the percentage of oscillating cells increases with state of activation, these oscillations may play an important role in mitogenic activation.

A functional role for GTP-binding proteins in synaptic vesicle cycling.

The squid giant synapse was used to test the hypothesis that guanosine-5'-triphosphate (GTP)-binding proteins regulate the local distribution of synaptic vesicles within nerve terminals. Presynaptic injection of the nonhydrolyzable GTP analog GTP gamma S irreversibly inhibited neurotransmitter release without changing either the size of the calcium signals produced by presynaptic action potentials or the number of synaptic vesicles docked at presynaptic active zones. Neurotransmitter release was also inhibited by injection of the nonhydrolyzable guanosine diphosphate (GDP) analog GDP beta S but not by injection of AIF4-. These results suggest that a small molecular weight GTP-binding protein directs the docking of synaptic vesicles that occurs before calcium-dependent neurotransmitter release. Depletion of undocked synaptic vesicles by GTP gamma S indicates that additional GTP-binding proteins function in the terminal at other steps responsible for synaptic vesicle replenishment.

Locomotor activity of the terrestrial slug, Limax maximus: response to progressive dehydration.

The circadian locomotor rhythm of the terrestrial slug, Limax maximus, was measured with activity wheels during exposure to both humid and drying conditions. Slugs kept in wet wheels (100% RH) remained fully hydrated while those in dry wheels (less than 30% RH) experienced progressive dehydration. Transfer of slugs from a wet wheel to a dry wheel resulted in an increase in the intensity and duration of their patterned locomotor activity that persisted for 3 days. Once the slugs were returned to wet wheels, their locomotor activity returned to normal.

Paracellular water uptake and molecular sieving by the foot epithelium of terrestrial slugs.

A paracellular pathway in the foot epithelium of Lehmannia valentiana can be opened by dehydrating the slug. Movement of water from a wet pad through the opened pathway into the haemolymph of this terrestrial slug is rapid. The sieving properties of this paracellular pathway have been determined using the reference isotope 3HOH and various 14C-labelled solutes. Paracellular uptake of 14C-insulin (Fig. 1) and 3HOH (Fig. 2) is initial rate for at least 3 min. If the wet pad contains 1,000 cpm of 14C per ml of 3HOH, slugs absorb only about 400 cpm of 14C with each ml of 3HOH absorbed representing a sieving ratio of 0.4 for insulin. The sieving ratio of 14C-inulin does not change when the concentration is increased from 0.1 to 2.5 mmol/l. Moreover, the sieving ratio of 14C-inulin was not affected significantly by the nature of the labelling, i.e., 14C-carboxyl vs 14C-methoxy. Sieving ratios for 14C-mannitol (182 Da), 14C-polyethylene glycol (4,000 Da), and 14C-inulin (5,250 Da) were 0.92, 0.63, and 0.39, respectively (Table 1), indicating that sieving is dependent on molecular size. 14C-Dextran (70,000 Da) and blue dextran (200,000 Da) were excluded from the paracellular pathway (Fig. 4). The effective pore size of the paracellular pathway was estimated using the relationships between sieving ratio and molecular weight of 3HOH and the various solutes that can pass through the pathway. The extrapolated pore size is equivalent to that of a sieve having a molecular weight cutoff of about 10,000 Da (Fig 3).