Quantcast

2002 M.D./Ph.D. student abstracts

COMPARATIVE ANALYSIS OF MICROARRAY TRANSCRIPT DATA OF IL-6 STIMULATED LNCaP AND CWR22RV1 PROSTATE CANCER CELL LINES
 
PSTP Student: Yu-Ming Chang
Collaborator: Dr. Hsing-Jien Kung
 
  Prostate cancer is the most prevalent cancer and the second most frequent cause of cancer related death among American men. Androgens are known to drive prostate cancer growth and thus lead to patient morbidity and mortality. Although anti-androgen therapy results in cancer cell apoptosis during its growth both in vivo and in vitro, the cancer eventually develops androgen independence and grows uninhibited despite therapy. There is some evidence that the acquisition of the cancer's androgen independence may be related to its ability to develop neuroendocrine characteristics.
   
  Interleukin-6 (IL-6) is a known acute phase response factor under normal physiological conditions known to produce effects such as driving the proliferation and differentiation of B cells, proliferation of CD4 T cells, production of acute-phase proteins, mobilization of neutrophils, and mobilization of protein and energy leading to increased body temperature. In prostate cancer, however, increased serum levels of IL-6 has been correlated with poor prognosis and increased patient morbidity and mortality. In addition, the binding of IL-6 to its receptor has been shown to activate ErbB2, a growth factor receptor implicated in the neoplastic transformation of prostate cancer.
   
  We are thus interested in the effect IL-6 has on prostate cancer in addition to ErbB2 activation. In this experiment, we stimulate two prostate cancer cell lines: LNCaP and CWR22RV1 with IL-6 and examine their gene transcripts over four time points via microarray analysis. We will be focusing on neuroendocrine specific, androgen related, cell cycle regulatory, as well as other oncogenic transcripts.
   
   

 
MICROVASCULAR COMPLICATIONS IN TYPE 2 DIABETES MELLITUS (T2DM)
 
PSTP Student: Patricia L. Duong
Mentor: Dr. Anthony Cheung
 
 
  Computer-assisted intravital microscopy (CAIM) was used to noninvasively study real-time microvascular complications in the conjunctival microcirculation in T2DM patients and nondiabetic controls. Fifteen recognizable microvascular abnormalities existed in T2DM patients and not in the controls, though not all the abnormalities were found in each patient. A Severity Index (SI; the sum total of microvascular abnormalities found in each patient) was computed to give a quantitative score for correlation with medical history (T2DM SI=6.9±2.1; control SI=0.61±0.66; P<0.001). T2DM SI correlated significantly with disease severity and HbA1c level, but not with disease duration from time of diagnosis. Six patients diagnosed with T2DM for =4 years had SI (8, 8, 9, 9, 9 and 10) significantly (P<0.001) higher than patients diagnosed for 6 and 10 years (SI=5 and 7). In addition, four recently diagnosed patients (=2 yrs) had SI of 4, 6, 8 and 9. These results lend support to the hypothesis that diabetic complications may occur in the prediabetic period before onset of clinically detectable hyperglycemia. CAIM can be used as a noninvasive tool to identify pathogenesis before clinical confirmation of T2DM.
   
   
   
   
   

 
  COMPACT COMPTON SCATTERING X-RAY SOURCE FOR CANCER DIAGNOSTICS AND TREATMENT
 
PSTP Student: William Frederick
Mentor: Dr. John Heritage
 
  The Compton X-ray Source (CXS) is a bright, monochromatic, tunable x-ray source that will soon be used to conduct proof-of-principle experiments in cancer detection and treatment. Collision of a short (fs) pulsed laser beam and a beam of relativistic, bunched electrons generate X-rays in a collimated beam from microscopic focal spot (< 30 micron) in an energy range tunable through 15 keV -120 keV. Source monochromaticity offers advantages in both diagnosis and therapy. When used with heavy-metal contrast agents, the CXS will allow location and identification of potentially cancerous lesions, with selection of energies to minimize dose and image acquisition time. The same contrast agents will be used in K-edge therapy to emit Auger electrons to enhance local delivery of high radiation dose. Experiments are underway using existing synchrotron sources to verify dose enhancement models, and CXS dose enhancement experiments are scheduled to begin shortly.
   
   
   
   


  THE ROLE OF ATP IN THE EXERCISE PRESSOR REFLEX
 
PSTP Student: Ramy Hanna
Mentor: Dr. Marc Kaufman
 
  Recently adenosine-3',5'-triophosphate (ATP) has been shown to stimulate visceral afferent C-fibers transducing the sensation of pain. Somatic C-fibers, along with thinly-myelinated A-delta fibers, comprise the afferent limb of the muscle reflex arc that causes cardiovascular and respiratory adaptations to exercise. We have previously shown that injection of alpha, beta methylene ATP into the arterial supply of the triceps surae muscle of decerebrate cats elicits a pressor response that is attenuated or abolished either by denervation of the muscle or by previous injection with the P2X receptor antagonist PPADS (10 mg/kg ia). Furthermore, we showed that injection of 2-chloroadenosine (25 ug/kg ia) into the same arterial supply does not evoke a pressor response. In the current study, we tested the hypothesis that the cardiovascular and ventilatory responses to static muscle contraction would be attenuated by injection of PPADS but not by injection of CGS 15943, a potent antagonist at adenosine (P1) receptors. The portion of the tibial nerve innervating the triceps surae muscles of the hindlimb was electrically to elicit a static contraction. Blood pressure (BP), heart rate (HR) and ventilation were monitored. Either PPADS (10 mg/kg) or CGS 15943 (20 ug/kg) was injected and trapped into the arterial supply of the muscles. The contraction was repeated and the same end-points monitored. The pressor response to contraction prior to PPADS injection averaged 33.3 ± 3.8 mmHg while that after PPADS injection averaged 9.8 ± 4.7 mmHg. The response was significantly attenuated by the P2X antagonist (n= 8, P<0.001). The pressor response prior to CGS 15943 injection averaged 39.8 ± 4.9 mmHg while that after CGS 15943 averaged 37.9 ±8.9 mmHg. The response was not significantly attenuated by the P1 antagonist (n=9, P>0.05). These results implicate ATP as a metabolic activator of the exercise pressor reflex while suggesting that adenosine is not.

 
 

SODIUM/HYDROGEN EXCHANGE INHIBITION AND CELL DEATH IN MALIGNANT GLIOMAS
 
PSTP Student: Manu Hegde
Mentor: Dr. Fred Gorin
Collaborators: Cala P, Floyd C
 
  We previously demonstrated that U118 and U87 human malignant glioma cells maintain an alkalotic intracellular pH (approximately 7.5) compared with primary rat astrocytes (7.0). Given the metabolic demands and hypoxic conditions for gliomas, this finding was surprising. Altered glioma pH regulation appears to result from increased tonic activity of the type 1 sodium hydrogen exchanger (NHE1), which overcompensates for increased intracellular H+ production. We have treated glioma cells in vitro with pharmacologic NHE1 inhibitors, and observed several effects. First, NHE1 inhibitors (NHE1i) produce widespread cell death in glioma cultures, and this death is largely caspase-independent. In addition, NHE1i may have an antiproliferative effect. We are currently expanding our analyses of these findings. Second, pHi change alone may act as an initiating event in this cell death, but cannot explain the scope of glioma demise we have observed. Glioma intracellular pH drops rapidly following NHE1i treatment, but precedes cell death by several hours. Finally, we hypothesize that tonic NHE1 activity may produce a secondary rise in intracellular calcium. Pilot data indicates this is the case; baseline intracellular calcium appears elevated in glioma cells. We have therefore proposed a model which links intracellular calcium, NHE1 inhibition, and cell death, and are designing experiments to test this hypothesis.
   
   
   

 


OVEREXPRESSION OF hXBP-1 IN THYMIC B CELLS OF AUTOIMMUNE MICE
 
PSTP Student: Tom Hsu
Mentor: Dr. Eric Gershwin
 
  New Zealand Black mice, as well as various other models of murine lupus all share common characteristics in their pathogenesis. Namely, the more prominent ones include premature degeneration of the thymus, production of anti-nuclear antibodies, and glomerular damage. In looking for a molecular basis for disease development, we performed differential display on NZB versus BALB/c thymuses at 5 weeks of age, looking for differences in mRNA expression levels. Any discrepancies were resolved and analyzed through in situ hybridization. A clone was found to be consistently differentially expressed between the two strains of mice, and upon sequencing, revealed itself to be the murine homologue of the human X box binding protein (hXBP-1). XBP has been primarily implicated in the regulation of MHC-class II expression, but also putatively plays a role in various other organ systems, particular with liver development. Analysis of other strains of mice showed a corresponding increase of XBP in lupus-prone mice, including the NZBxNZW F1 hybrid, and the MRL-lpr/lpr mice. Interestingly, there was an absence of mRNA expression of XBP in the NZB/Bln-Igh6null mice, deficient in B cells. Protein expression studies have shown a corresponding increase in XBP levels in the autoimmune mice as well. Further characterization of the function of XBP still remains to be accomplished.
   
   
   
   
   

 


BIOMECHANICAL ANALYSIS OF THREE SURGICAL APPROACHES FOR LUMBAR CORPECTOMY USING ALLOGRAFT AND SYNTHETIC DEVICES
 
PSTP Student: Philbert Huang
Mentor: Dr. Nesrin Sarigul-Klijn
Collaborators: Hazelwood S, and Gupta M
 
  Fractures, tumors, bacterial infections and other diseases affecting vertebral bodies can result in either spinal instability and/or neural impingement. For many of these conditions it is mandatory to reconstruct the destabilized spine after anterior decompression of the vertebral body, or vertebral corpectomy. Anterior bone grafting is often performed. However, bone grafting alone does not provide spinal stability immediately after the surgery. Therefore, either anterior or posterior spinal instrumentation, combined with anterior bone graft, has been advocated for the restoration of vertebral column stability. To further improve patient care, it is critical to further evaluate a variety of instruments by biomechanical testing. Since it is important to perform evaluations in the context of the natural healing abilities of soft tissue and bone we are developing an in vivo sheep model for analyzing interactions between fusions and anterior spinal instrumentation. Lumbar spinal instability will be created in the animal model by a corpectomy procedure at a given level. Next, the following current vertebral body replacement methods will be implemented in the animal model: 1) femoral ring allograft; 2) Carbon Fiber Stackable Cage System; and 3) Titanium Mesh Cages.
   
  These strut implants will be used with the Kaneda device for anterior fixation. Following an appropriate postoperative period, the animals will be sacrificed and their lumbar spines harvested. This will allow for biomechanical, radiological, and histological assessment. All Kaneda instrumentation will then be removed. Testing only the vertebral bodies and the replacement for the corpectomized unit will serve as a true test of the fusion success of each graft/device. Biomechanical testing parameters will include subjecting the segment to torsion, axial compression, flexion/extension, and lateral bending. Load versus deformation data will allow for the determination of torsional, axial, and flexural stiffnesses. Another consideration will be the use of strain gauges attached to the surfaces of the vertebral bodies to provide direct measurement of axial and flexural strains. Stiffness and/or strain data can then be compared across the fused vertebral constructs to determine if one is superior to the others. These studies should provide valuable insight for the design and development of the next generation of implants for spinal stabilization.
   

 


DIFFERENTIAL EFFECTS OF MAP KINASE SIGNALING ON COREPRESSORS
 
PSTP Student: Brian Jonas
Mentor: Dr. Martin Privalsky
 
  The nuclear receptors are hormone-controlled transcription factors that regulate the expression of a complex array of target genes. In most cases, in the absence of hormone, these receptors interact with proteins called corepressors to mediate gene repression, whereas in the presence of hormone, they interact with coactivators to mediate gene activation. SMRT and N-CoR are paralogs of one another, and both appear to play critical roles in the control of normal gene expression; aberrant regulation by these corepressors has been implicated in a number of human endocrine and neoplastic diseases. Although N-CoR and SMRT share many common properties, the conservation of these corepressors as two distinct paralogs in all vertebrates studied suggest that they must also possess functional differences. We are interested in identifying and understanding these differences. Using mammalian two-hybrid analysis, we have shown that cotransfection of the MAP kinase kinase kinase, MEKK1, stabilizes the interaction between N-CoR and thyroid hormone receptors (T3Rs) whereas it strongly inhibits the interaction between SMRT and T3R. Western immunoblot analysis rules out the possibility that MEKK1 is functioning by causing different changes in the protein levels or stability of SMRT or N-CoR. We have also shown that a constitutively active EGF-receptor, which operates above MEKK1 in a growth-factor signaling cascade, inhibits the interaction of both SMRT and N-CoR with T3R. These results imply that additional downstream effectors of EGF receptor signaling exert modulatory actions on SMRT and N-CoR, and operate outside of the MAP kinase pathway previously identified in our experiments. We are currently exploring the mechanism of the MEKK1 and EGF-receptor signaling on N-CoR as compared to SMRT using co-immunoprecipitation, and we will also create GFP-corepressor fusions in order to compare changes in the cellular localization of N-CoR and SMRT in response to MAP kinase and EGF-receptor signaling.
   

 


CHARACTERIZATION OF CALCIUM RELEASE-ACTIVATED APOPTOSIS OF LNCaP PROSTATE CANCER CELLS
 
PSTP Student: Ingrid Wertz
Mentor: Dr. Vishva Dixit
 
  Apoptosis inhibition rather than enhanced cellular proliferation occurs in
prostate cancer (CaP), the most commonly diagnosed malignancy in American men. It is therefore important to characterize residual apoptotic pathways in CaP
cells. When intracellular Ca2+ stores are released and plasma membrane
"store-operated" Ca2+ entry channels open, cytosolic Ca2+ increases which is
thought to induce apoptosis. However, cells incapable of releasing Ca2+ stores
are resistant to apoptotic stimuli indicating that Ca2+ store release is also
important. We therefore investigated whether release of intracellular Ca2+
stores is sufficient to induce apoptosis of the CaP cell line LNCaP. We
developed a method to release stored Ca2+ without elevating cytosolic Ca2+; this stimulus induced LNCaP cell apoptosis. We compared the apoptotic pathways activated by intracellular Ca2+ store release to the dual insults of store
release and cytosolic Ca2+ elevation. Earlier processing of caspases-3 and -7
occurred when intracellular store release was the sole Ca2+ perturbation.
Apoptosis was attenuated in both conditions in stable-transfected cells
expressing antiapoptotic proteins BclxL and catalytically inactive caspase-9, and
in both scenarios inactive caspase-9 complexed with caspase-7. Thus,
intracellular Ca2+ store release initiates an apoptotic pathway similar to that
elicited by the dual stimuli of cytosolic [Ca2+] elevation and intracellular
store release.
   

 


SLOWED DEACTIVATION OF THE LIGHT RESPONSE IN RODS LACKING THE PROTEIN Gb5-L
 
PSTP Student: Claudia Krispel
Mentor: Dr. Marie Burns
 
  Retinal rod photoreceptors convert light into electrical responses via a G protein cascade. The timely recovery of the light response is important for temporal resolution of rod vision, and requires prompt deactivation of the G protein cascade. In particular, the active G protein which is formed during the light response must be deactivated by GTP hydrolysis. In rods, the rate of GTP hydrolysis is speeded by a protein called RGS9, which exists in a complex with another G protein b subunit called Gb5-L. The function of Gb5-L is unknown. In this study we have investigated the function of Gb5-L by recording light responses of rods from mice that do not express Gb5-L. We have found that the activation of the light response in knockout rods was indistinguishable from that of wild-type mouse rods. However, the responses of Gb5 -/- rods were very slow to recover. These results show that Gb5-L is required for the proper function of rods, and suggest that it is specifically important for deactivation of the cascade.
   
 


CIRCULATING C-REACTIVE PROTEIN (CRP), INTERLEUKIN-6 (IL-6) AND SOLUBLE TUMOR NECROSIS FACTOR RECEPTOR TYPE-II (sTNF-RII) DECREASE WITH WEIGHT LOSS IN MODERATELY OVERWEIGHT WOMEN
 
PSTP Student: Edward A. Medina
Mentor: Dr. Kent L. Erickson
Collaborators: Horn WF, and Keim NL
 
  Circulating levels of CRP and the proinflammatory cytokines TNF-a and IL-6 are elevated in overweight and obese-associated metabolic syndrome X; they are hypothesized to contribute to insulin resistance. It is unclear whether these inflammatory factors are elevated as a result of increased adipose tissue mass, which has recently been demonstrated to produce TNF-a, sTNF-RI and RII, and IL-6. The few studies that examined the effects of weight loss on these inflammatory factors yielded conflicting results, but were limited to free-living obese subjects. To examine the role of adipose tissue on plasma CRP, TNF-a, sTNF-RI and RII, and IL-6, we assessed the effects of energy-deficit-induced weight loss on these inflammatory parameters and related changes in them to changes in metabolic parameters, in moderately overweight women in a highly controlled setting. Twelve healthy women (average BMI 28.2 ± 1.3) completed a metabolic ward study consisting of a 3 wk stabilization period followed by 12 wk of moderate energy restricton and daily exercise. Body weight and composition; fasting CRP, IL-6, TNF-a, sTNF-RI, sTNF-RII, insulin and glucose concentrations, and oral glucose challenge tests were determined near the beginning and end of the intervention period. Weight, BMI, fat free mass and fat mass decreased significantly over the 8 wk intervention period in the energy-deficit group (P = 0.001). Plasma concentrations of CRP (P = 0.028), IL-6 (P = 0.039), sTNF-RII (P = 0.001), and the ratio glucose AUC/insulin AUC (P = 0.016) also decreased significantly. The percentage change in CRP was correlated with the percentage change in insulin (r = -.641; P = 0.025), and the percentage change in sTNF-RII was correlated with the percentage change in fat free mass (r = .671; P = 0.017), fat mass (r = -.572; P = 0.052), insulin (r = .588; P = 0.044) and glucose AUC (r = -.622; P = 0.016). Eight free-living control subjects with comparable body compositions demonstrated a small but significant increase in weight, BMI and fat mass over an 8 w period (P = 0.05); there was also a small but significant increase in sTNF-RI (P = 0.030) . Although plasma CRP and IL-6 decreased with weight loss, these changes were unrelated to decreases in fat mass. This suggests that other factors associated with energy-deficit or weight loss, beside the decrease in adipose tissue mass, are mediating changes in plasma CRP, IL-6 and sTNF-RII. TNF-a activity may have also increased with weight loss due to the decrease in sTNF-RII in the absence of changes in TNF-a and sTNF-R1, which could explain the inverse relationship between the percentage change in sTNF-RII and the percentage change in glucose AUC. Studies that explore the unexpected coupling between fat free mass and sTNF-RII may reveal a role for TNF-a in glucose homeostasis.
   
 
ACUTE INFLAMMATORY RESPONSE TO LIPOPOLYSACCHARIDE INDUCES TRANSIENT HYPERHOMOCYSTEINEMIA IN MICE
 
PSTP Student: Maria Medina
Mentor: Dr. Ralph Green
Collaborators: Reynolds RM, Erickson K, Miller JW
 
  Elevated plasma homocysteine (hyperhomocysteinemia, HHCY) is now recognized as an independent risk factor for cardiovascular, cerebrovascular, and peripheral vascular disease. The causal link between homocysteine and these disorders, however, has yet to be established definitively. HHCY, in fact, may be a consequence of chronic inflammation associated with vascular disease. We therefore investigated the hypothesis that HHCY can result from inflammation in an animal model. A systemic inflammatory response was induced in C57BL mice by intraperitoneal injection of lipopolysaccharide (LPS) dissolved in saline. Control mice received saline alone. LPS induced a significant increase in serum TNF-alpha levels within 1h which returned to baseline by 4h, indicative of an acute inflammatory response. Baseline serum homocysteine was 8.7±1.8 µmol/L. Within 2h of the LPS injection, serum homocysteine was significantly elevated over baseline and remained elevated for at least 5h. Peak homocysteine concentration at 3h was 11.6±2.0 µmol/L (p=0.007 compared with baseline). No change from baseline in serum homocysteine was observed for the saline controls. These results indicate that HHCY occurs in response to acute systemic inflammation. The biochemical or physiological mechanisms by which this occurs remain to be determined, as well as whether HHCY is sustained during chronic inflammation.
   
 
FACE EXPRESSION PROCESSING TESTED WITH EXPRESSION-MORPHED IMAGES
 
PSTP Student: Noah Merrin
Mentor: Dr. David Amaral
 
  We intend to study a group of patients who have undergone selective bilateral amygdalectomy. One of our primary hypotheses is that these patients will be impaired at recognizing the intensity of facial expressions due to the putative involvement of the amygdala in social cognition and face expression processing. This presentation outlines a computer-based experiment intended to quantify expression processing in control subjects, and normalize a large set of digitally manipulated images of facial expression, so that these stimuli can be used to test impairment in this and other patient populations. Pilot data demonstrating psychometric threshold curves of normal subjects generated with this experiment will be presented. The procedures used to generate the stimuli are also described.
   
 


MARKERS DISTINGUISHING ANCESTRY DEMONSTRATE STRONG EXTENDED LINKAGE DISEQUILIBRIUM IN AFRICAN AMERICANS
 
PSTP Student: Heather Schramm
Mentor: Dr. Michael Seldin
Collaborators: Chima BS, Operario DJ, Weber JL, Criswell LA, Cooper RS
 
  Conflicting reports regarding the extent of linkage disequilibirum (LD) in the African American (AA) population have created uncertainty about the usefulness of admixture mapping in this population. By examining regions of human chromosomes 5 and 22, we demonstrate that significant and impressive LD is observable for extensive distances in AA for diallelic markers with large allele frequency differences (measured by the f value) between European Americans and Africans. In contrast, only rare, marginally significant LD is present between unlinked markers in this population. Significant LD extends up to 12 cM (e.g. MID1191 at 126.2 cM and MID719 at 138.4 cM on Chr 22, p=10-4, D'=0.54). In representative founder populations, strong LD was only observed for markers separated by less than 10kb and suggestive LD was only observed at greater distances in a single comparison in the African population (MID1102 at 135.1 and MID768 at 135.5, p= 0.003, D'=0.26). For unlinked markers, only one of the 42 pair-wise analyses in African Americans was associated with a p value < 0.01 (MID619 and MID883, p=0.003, D'= 0.2). Examination of the strength of LD demonstrated that the ability to detect LD for extended chromosomal segments decays as a function of both the f values and distance between markers. Unlinked marker comparisons are clearly distinguishable from linked marker comparisons by LD strength. It is likely that the conflicting results of previous reports are due to differences in the f values of markers chosen for analysis. These results indicate that obtainable markers with large f values can detect extended regions of LD and should therefore be useful in admixture mapping in the AA population.
   
 
MARKERS THAT DISCRIMINATE BETWEEN EUROPEAN AND AFRICAN ANCESTRY SHOW LIMITED AFRICAN DIVERSITY
 
PSTP Student: Heather Schramm
Mentor: Dr. Michael Seldin
Collaborators: Kittles RA, Operario DJ, Weber JL, Criswell LA, Cooper RS
 
  Markers informative for ancestry are necessary for admixture mapping and improving case-control association analyses. For studies in African Americans (AA) this will require markers that can be used to determine if a given genomic region is of European or African ancestry. We show that despite studies indicating high intra-African sequence diversity, markers with large inter-ethnic differences have only small variations in allele distribution among divergent African populations. 32 unlinked diallelic insertion/deletion markers with large frequency differences between one African population and European Americans (EA) were individually genotyped in >150 DNA samples each from Africans in Nigeria and Zimbabwe, EAs, and AAs. Allele frequencies in Nigerians were nearly identical to those in Zimbabweans and dramatically different from those in EAs. Perhaps more surprising, there was a strong correlation between allele distributions in the above African groups and a pooled DNA sample from 21 Pygmies. The variation for the intra-African comparisons, Nigerian/Zimbabwe (mean Wahlund variance f=0.005) and Nigerian/Pygmy (f=0.02), was very small compared to inter-ethnic African/EA comparisons: Nigerian/EA (f=0.38) and Zimbabwe/EA (f=0.37). We also examined individual admixture using the Structure program. Africans were clearly separated from EAs without prior knowledge of population affiliation, and no distinction was evident between Nigerians and Zimbabweans. AA showed considerable variation in the proportions contributed from the two parental populations. These studies emphasize the importance of controlling for individual admixture ratios in admixture mapping studies, and provide evidence that the African contribution to AA subjects can be accurately analyzed with a set of obtainable markers that distinguish between European and African ancestry.
   

 

HUMAN FETAL PANCREAS-DERIVED NESTIN POSITIVE STEM CELLS FROM LONG TERM CULTURES CORRECT HYPERGLYCEMIA WHEN IMPLANTED IN DIABETIC MICE
 
PSTP Student: Frederick Wu
Mentor: Dr. Jerry S Powell
 
  Diabetes is a disease characterized by the destruction of insulin producing beta islet cells resulting in hyperglycemia and associated morbidity. With improvements in islet transplantation therapy for diabetes, there is growing interest in increasing the availability of transplantable tissue. Nestin positive cells (NPC) derived from pancreas and embryonic stem (ES) cells are able to proliferate and maintain their ability to differentiate towards an endocrine cell lineage, hinting at their potential as islet stem cells. However, differentiated NPCs have not yet shown the ability to correct hyperglycemia in animal models. Taking a cue from development, we used human fetal pancreas (HFP) as an enriched source of precursor cells from which a population of stem cells might be established in vitro. Culture preparations expressed varied percentages of nestin positive cells, with one culture preparation comprised solely of nestin expressing cells. These cells were cultured for over eighteen months, during which time, the effects of extracellular matrix proteins on cell differentiation were observed. Growth of NPC on collagens I and IV, fibronectin, laminin, and poly-L-lysine did not change the typical monolayer forming morphology of the cells. However, growth of NPC on poly-D-lysine (PDL) or Matrigel coated culture plates induced formation of nestin positive cell aggregates (NPA) that resembled islets. Other derivative cultures of non-nestin positive fetal pancreatic cells did not form aggregates with PDL or Matrigel. NPA's stained positively for insulin using both anti-insulin antibody immunohistochemistry and dithisone staining. Differential display for tyrosine kinase genes of cells from monolayer culture versus NPA demonstrated selectively increased 5-fold expression of PDGFR b and ErbB2 mRNA in the aggregated cells. The in vivo differentiation potential of NPA's was studied by implanting NPA's into streptozotocin induced diabetic mice (blood glucose concentration > 300 mg/dl). Animals treated with insulin briefly to correct their hyperglycemia prior to NPA implantation (2-5 x 105 aggregates per mouse) were then able to maintain essentially normal glucose concentrations, with occasional morning glucose measurements up to 190 mg/dl in some mice, without exogenous insulin for more than 70 days, and have had essentially normal glucose tolerance tests (n = 6 mice, versus control diabetic mice n = 10, p < 0.0005). These findings show that a potential population of human islet stem cells 1) can be identified from human fetal pancreatic tissue and expanded in vitro, 2) can maintain the ability to differentiate into functional islets after more than a year in culture, and 3) can correct hyperglycemia in a mouse model of streptozotocin induced diabetes. Further characterization of human pancreas derived nestin positive stem cells could lead to insight into the differentiation pathways for normal islets, and may provide a novel source of tissue available for transplantation therapy for diabetes.