INDUCTION OF DNA STRAND BREAKS AND CROSS-LINKS BY THE NEW ANTICANCER AGENT 3,6-DIAZIRIDINYL-2,5-BIS(CARBOETHOXYAMINO) -1,4-BENZOQUINONE

The DNA breakage effect of the anticancer agent 3,6-diaziridinyl-2,5-bis(carboethoxyamino)-1,4-benzoquinone (AZQ, NSC-182986) on bacteriophage PM2 DNA was investigated using agarose gel electrophoresis. AZQ caused both single-stranded and double-stranded breaks after reduction with NaBH(,4), but it was not active in the native state. At 120 (mu)M, it degraded 50% of the closed circular form I DNA into 40% form II DNA (single-stranded break) and 10% form III DNA (double-stranded break). It produced a dose-response breakage between 1 (mu)M and 320 (mu)M. The DNA breakage exhibited a marked pH dependency. At 320 (mu)M, AZQ degraded 80% and 60% of form I DNA at pH 4 and 10 respectively, but none between pH 6 to 8. The DNA breakage at physiologic pH was greatly enhanced when 10 (mu)M cupric sulfate was included in the incubation mixture. The DNA strand scission was inhibited by catalase, glutathione, KI, histidine, Tiron, and DABCO. These results suggest that the DNA breakage may be caused by active oxygen metabolites including hydroxyl free radical. The bifunctional cross-linking activity of reduced AZQ on isolated calf thymus DNA was investigated by ethidium fluorescence assay. The cross-linking activity exhibited a similar pH dependency; highest in acidic and alkaline pH, inactive under neutral conditions. Using the alkaline elution method, we found that AZQ induced DNA single-stranded breaks in Chinese hamster ovary cells treated with 50 (mu)M of AZQ for 2 hr. The single-stranded break frequencies in rad equivalents were 17 with 50 (mu)M and 140 with 100 (mu)M of AZQ. In comparison, DNA cross-links appeared in cells treated with only 1 to 25 (mu)M of AZQ for 2 hr. The cross-linking frequencies in rad equivalents were 39 and 90 for 1 and 5 (mu)M of AZQ, respectively. Both DNA-DNA and DNa-protein cross-links were induced by AZQ in CHO cells as revealed by the proteinas K digestion assay. DNA cross-links increased within the first 4 hr of incubation in drug-free medium and slightly decreased by 12 hr, and most of the cross-links disappeared after cells were allowed to recovered for 24 hr.^ By electrochemical analysis, we found that AZQ was more readily reduced at acidic pH. However, incubation of AZQ with NaBH(,4) at pH 7.8 or 10, but not at 4, produced superoxide anion. The opening of the aziridinyl rings of AZQ at pH 4 was faster in the presence of NaBH(,4) than in its absence; no ring-opening was detected at pH 7.8 regardless of the inclusion of NaBH(,4). . . . (Author's abstract exceeds stipulated maximum length. Discontinued here with permission of author.) UMI ^

Bayesian doubly optimal group sequential designs for randomized clinical trials

When conducting a randomized comparative clinical trial, ethical, scientific or economic considerations often motivate the use of interim decision rules after successive groups of patients have been treated. These decisions may pertain to the comparative efficacy or safety of the treatments under study, cost considerations, the desire to accelerate the drug evaluation process, or the likelihood of therapeutic benefit for future patients. At the time of each interim decision, an important question is whether patient enrollment should continue or be terminated; either due to a high probability that one treatment is superior to the other, or a low probability that the experimental treatment will ultimately prove to be superior. The use of frequentist group sequential decision rules has become routine in the conduct of phase III clinical trials. In this dissertation, we will present a new Bayesian decision-theoretic approach to the problem of designing a randomized group sequential clinical trial, focusing on two-arm trials with time-to-failure outcomes. Forward simulation is used to obtain optimal decision boundaries for each of a set of possible models. At each interim analysis, we use Bayesian model selection to adaptively choose the model having the largest posterior probability of being correct, and we then make the interim decision based on the boundaries that are optimal under the chosen model. We provide a simulation study to compare this method, which we call Bayesian Doubly Optimal Group Sequential (BDOGS), to corresponding frequentist designs using either O'Brien-Fleming (OF) or Pocock boundaries, as obtained from EaSt 2000. Our simulation results show that, over a wide variety of different cases, BDOGS either performs at least as well as both OF and Pocock, or on average provides a much smaller trial. ^

The molecular mechanisms of sensing nucleoside analogue-induced DNA damage

Nucleoside analogues are antimetabolites effective in the treatment of a wide variety of solid tumors and hematological malignancies. Upon being metabolized to their active triphosphate form, these agents are incorporated into DNA during replication or excision repair synthesis. Because DNA polymerases have a greatly decreased affinity for primers terminated by most nucleoside analogues, their incorporation causes stalling of replication forks. The molecular mechanisms that recognize blocked replication may contribute to drug resistance but have not yet been elucidated. Here, several molecules involved in sensing nucleoside analogue-induced stalled replication forks have been identified and examined for their contribution to drug resistance. ^ The phosphorylation of the DNA damage sensor, H2AX, was characterized in response to nucleoside analogues and found to be dependent on both time and drug concentration. This response was most evident in the S-phase fraction and was associated with an inhibition of DNA synthesis, S-phase accumulation, and activation of the S-phase checkpoint pathway (Chk1-Cdc25A-Cdk2). Exposure of the Chk1 inhibitor, 7-hydroxystaurosporine (UCN-01), to cultures previously treated with nucleoside analogues caused increased apoptosis, clonogenic death, and a further log-order increase in H2AX phosphorylation, suggesting enhanced DNA damage. Ataxia-telangiectasia mutated (ATM) has been identified as a key DNA damage signaling kinase for initiating cell cycle arrest, DNA repair, and apoptosis while the Mre11-Rad50-Nbs1 (MRN) complex is known for its functions in double-strand break repair. Activated ATM and the MRN complex formed distinct nuclear foci that colocalized with phosphorylated H2AX after inhibition of DNA synthesis by the nucleoside analogues, gemcitabine, ara-C, and troxacitabine. Since double-strand breaks were undetectable, this response was likely due to stalling of replication forks. A similar DNA damage response was observed in human lymphocytes after exposure to ionizing radiation and in acute myelogenous leukemia blasts during therapy with the ara-C prodrug, CP-4055. Deficiencies in ATM, Mre11, and Rad50 led to a two- to five-fold increase in gemcitabine sensitivity, suggesting that these molecules contribute to drug resistance. Based on these results, a model is proposed for the sensing of nucleoside analogue-induced stalled replication forks that includes H2AX, ATM, and the Mre11-Rad50-Nbs1 complex. ^

The cellular and molecular responses to a novel nucleotide analogue, GS-9219

GS-9219 is a cell-permeable double-prodrug of the acyclic nucleotide analogue 9-(2-phosphonylmethoxyethyl)guanine (PMEG). The conversion of GS-9219 to its active metabolite, PMEG diphosphate (PMEGpp), involves several intracellular enzymatic reactions which reduces the concentration of nephrotoxic PMEG in plasma. PMEGpp competes with the natural substrate, dGTP, for incorporation by DNA polymerases. The lack of a 3'-hydroxyl moiety makes PMEGpp a de facto DNA chain-terminator. The incorporation of PMEGpp into DNA during DNA replication causes DNA chain-termination and stalled replication forks. Thus, the primary mechanism of action of GS-9219 in replicating cells is via DNA synthesis inhibition. GS-9219 has substantial antiproliferative activity against activated lymphocytes and tumor cell lines of hematological malignancies. Tumor cell proliferation was significantly reduced as measured by PET/CT scans in dogs with advanced-stage, spontaneously occurring non-Hodgkin's lymphoma (NHL).^ The hypothesis of this dissertation is that the incorporation of PMEGpp into DNA during repair re-synthesis would result in the inhibition of DNA repair and accumulation of DNA damage in chronic lymphocytic leukemia (CLL) cells and activate signaling pathways to cell death.^ To test this hypothesis, CLL cells were treated with DNA-damaging agents to stimulate nucleotide excision repair (NER) pathways, enabling the incorporation of PMEGpp into DNA. When NER was activated by UV, PMEGpp was incorporated into DNA in CLL cells. Following PMEGpp incorporation, DNA repair was inhibited and led to the accumulation of DNA strand breaks. The combination of GS-9219 and DNA-damaging agents resulted in more cell death than the sum of the single agents alone. The presence of DNA strand breaks activated the phosphatidylinositol 3-kinase-like protein kinase (PIKK) family members ataxia-telangiectasia mutated (ATM) and DNA-dependent protein kinase (DNA-PK). The activated ATM initiated signaling to the downstream target, p53, which was subsequently phosphorylated and accumulated to exert its apoptotic functions. P53-targeted pro-apoptotic genes, Puma and Bax, were upregulated and activated when DNA repair was inhibited, likely contributing to cell death. ^

A global perspective on intimate partner violence during pregnancy: An analysis of twelve developing countries

Violence against women has been recognized as a significant worldwide human rights issue and public health problem. Women of reproductive age may be particularly at risk, and pregnancy may trigger or escalate violence. Using data available from Demographic and Health Surveys on 271,103 women of reproductive age (15-49) from Bolivia, Cameroon, Colombia, Dominican Republic, Egypt, Haiti, India, Kenya, Nicaragua, Peru, South Africa, and Zambia, this study examined the nature of domestic violence during pregnancy in developing countries, including prevalence, demographic and risk factors, maternal and child health outcomes, perpetrators of violence, help-seeking behavior, and social support. In the majority of countries analyzed, violence during pregnancy consistently occurred at approximately one-third the rate at which domestic violence occurred overall. Younger women and women with more children were particularly at risk. Abuse during pregnancy was significantly associated with history of a terminated pregnancy and under-5 child mortality in most countries, and with neonatal and post-neonatal mortality in most Latin American countries. Women who were abused during pregnancy were most often abused by their current or former husband or boyfriend and most never attempted to seek help. In most countries that examined social support, women abused during pregnancy had significantly less contact with family and friends. Implications for practice and research are discussed. ^

THE POTENTIAL EFFECTS OF A COMMUNITY DEVELOPMENT PROGRAM ON DEMOGRAPHIC VARIABLES IN HIMALAYAN INDIAN VILLAGES

A community development program operating in the mountains of North India was studied to assess its potential effects on mortality, fertility and migration patterns in the community which it served. The development program operated in Jaunpur Block, Tehri-Garhwal District, Uttar Pradesh State. Two comparable villages in the district were studied. The development program had been working in one for two years, and the other was completely untouched by the program.^ Since not enough time had elapsed since the beginning of the development program's work for any effects on demographic patterns to be visable in Jaunpur Block, this study looked to attitudes of village residents as indicators of future demographic trends. Existing demographic patterns and their interrelationship with socio-religious customs were examined in the test village. A questionnaire was then administered to ascertain attitudinal differences between the residents of the test village and the control village.^ The primary work of the community development program was to train women as village health workers. The results of the attitudinal comparison of the residents of the two villages showed a marked difference in attitudes relating to the position of women in society. The data showed a higher esteem for women in the test village than in the control village, and it is argued that this difference may be attributable to the work of the development program.^ Predicting future demographic trends in Jaunpur Block on the basis of the observed difference in villagers' attitudes toward the status of women is speculatory. Jaunpur Block appears to be in the demographic stage of pre-transition, maintaining relatively high rates of both mortality and fertility. Based on demographic transition theory the next significant change in demographic patterns in Jaunpur is predicted to be a decline in mortality, and an increase in the status of women is unrelated to this prediction.^ The community development program which was studied terminated unexpectedly during the time of this study. A case study of the program's final months is presented, and speculation on the future course of demographic trends in Jaunpur Block is related to the possible alternatives for future development in the area. ^

Structure and biosynthesis of a pyruvate -dependent enzyme---phosphatidylserine decarboxylase from {\it Escherichia coli\/}

Phosphatidylserine decarboxylase of E. coli, a cytoplasmic membrane protein, catalyzes the formation of phosphatidylethanolamine, the principal phospholipid of the organism. The activity of the enzyme is dependent on a covalently bound pyruvate (Satre and Kennedy (1978) J. Biol. Chem. 253, 479-483). This study shows that the enzyme consists of two nonidentical subunits, $\alpha$ (Mr = 7,332) and $\beta$ (Mr = 28,579), with the pyruvate prosthetic group in amide linkage to the amino-terminus of the $\alpha$ subunit. Partial protein sequence and DNA sequence analysis reveal that the two subunits are derived from a proenzyme ($\pi$ subunit, Mr = 35,893) through a post-translational event. During the conversion of the proenzyme to the $\alpha$ and $\beta$ subunits, the peptide bond between Gly253-Ser254 is cleaved, and Ser254 is converted to the pyruvate prosthetic group at the amino-terminus of the $\alpha$ subunit (Li and Dowhan (1988) J. Biol. Chem. 263, 11516-11522).^ The proenzyme cannot be detected in cells carrying either single or multiple copies of the gene (psd), but can be observed in a T7 RNA polymerase/promoter and transcription-translation system. The cleavage of the wild-type proenzyme occurs rapidly with a half-time on the order of 2 min. Changing of the Ser254 to cysteine (S254C) or threonine (S254T) slows the cleavage rate dramatically and results in mutants with a half-time for processing of around 2-4 h. Change of the Ser254 to alanine (S254A) blocks the cleavage of the proenzyme. The reduced processing rate with the mutations of the proenzyme is consistent with less of the functional enzyme being made. Mutants S254C and S254T produce $\sim$15% and $\sim$1%, respectively, of the activity of the wild-type allele, but can still complement a temperature-sensitive mutant of the psd locus. Neither detectable activity nor complementation is observed by mutant S254A. These results are consistent with the hydroxyl-group of the Ser254 playing a critical role in the cleavage of the peptide bond Gly253-Ser254 of the pro-phosphatidylserine decarboxylase, and support the mechanism proposed by Snell and co-workers (Recsei and Snell (1984) Annu. Rev. Biochem. 53, 357-387) for the formation of the prosthetic group of pyruvate-dependent decarboxylases. ^

Molecular and evolutionary genetics of the X -linked visual pigment genes in humans and New World monkeys

Normal humans have one red and at least one green visual pigment genes. These genes are tightly linked as tandem repeats on the X chromosome and each of them has six exons. There is only one X-linked visual pigment gene in New World monkeys (NWMs) but the locus has three polymorphic alleles encoding red, yellow and green visual pigments, respectively. The spectral properties of the squirrel monkey and the marmoset (both NWMs) have been studied and partial sequences of the three alleles are available. To study the evolutionary history of these X-linked opsin genes in humans and NWMs, coding and intron sequences of the three squirrel monkey alleles and the three marmoset alleles were amplified by PCR followed by subcloning and sequencing. Introns 2 and 4 of the human red and green pigment genes were also sequenced. The results obtained are as follows: (1) The sequences of introns 2 and 4 of the human red and green opsin genes are significantly more similar between the two genes than are coding sequences, contrary to the usual situation where coding regions are better conserved in evolution than are introns. The high similarities in the two introns are probably due to recent gene conversion events during evolution of the human lineage. (2) Phylogenetic analysis of both intron and exon sequences indicates that the phylogenetic tree of the available primate opsin genes is the same as the species tree. The two human genes were derived from a gene duplication event after the divergence of the human and NWM lineages. The three alleles in each of the two NWM species diverged after the split of the two NWMs but have persisted in the population for at least 5 million years. (3) Allelic gene conversion might have occurred between the three squirrel monkey alleles. (4) A model of additive effect of hydroxyl-bearing amino acids on spectral tuning is proposed by treating some unknown variables as groups. Under the assumption that some residues have no effect, it is found that at least five amino acid residues, at positions 178 (3 nm), 180 (5 nm), 230 ($-$4 nm), 277 (9 nm) and 285 (13 nm), have linear spectral tuning effects. (5) Adaptive evolution of the opsin genes to different spectral peaks was observed at four residues that are important for spectral tuning. ^

Mid-Wisconsin to Holocene permafrost and landscape dynamics based on a drained lake basin core from the northern Seward Peninsula, Northwest Alaska

Permafrost-related processes drive regional landscape dynamics in the Arctic terrestrial system. A better understanding of past periods indicative of permafrost degradation and aggradation is important for predicting the future response of Arctic landscapes to climate change. Here, we used a multi-proxy approach to analyze a ~4 m long sediment core from a drained thermokarst lake basin on the northern Seward Peninsula in western Arctic Alaska (USA). Sedimentological, biogeochemistical, geochronological, micropaleontological (ostracoda, testate amoeba) and tephra analyses were used to determine the long-term environmental Early-Wisconsin to Holocene history preserved in our core for Central Beringia. Yedoma accumulation dominated throughout the Early to Late-Wisconsin but was interrupted by wetland formation from 44.5 to 41.5 ka BP. The latter was terminated by deposition of 1 m of volcanic tephra, most likely originating from the South Killeak Maar eruption at about 42 ka BP. Yedoma deposition continued until 22.5 ka BP and was followed by a depositional hiatus in the sediment core between 22.5 and 0.23 ka BP. We interpret this hiatus as due to intense thermokarst activity in the areas surrounding the site, which served as a sediment source during the Late-Wisconsin to Holocene climate transition. The lake forming the modern basin on the upland initiated around 0.23 ka BP, which drained catastrophically in spring 2005. The present study emphasizes that Arctic lake systems and periglacial landscapes are highly dynamic and permafrost formation as well as degradation in Central Beringia was controlled by regional to global climate patterns and as well as by local disturbances.

Preliminary results and documentation of sediment core CRP-1 from the Ross Sea off Cape Roberts, Antarctica

The first hole of the Cape Roberts Project, CRP-1, was drilled in October, 1997, to a depth of 148 metres below the sea floor (mbsf) before being terminated unexpectedly the loss of fast sea-ice seaward of the rig following a severe storm. The site lies in 150 m of water at 77.008°S and 163.755°E, 16 km off Cape Roberts. This part of the report outlines the geologic setting, a gently tilted sequence near the margin of the Victoria Land Basin, and describes the history of the growth of sea ice, which provided the drilling platform, as well as the history of the drilling itself. Core recovery was around 77% in soft and brittle strata to 100 m and 98% below that. The sequence was found to comprise a Quaternary glacigenic interval down to 43.55 mbsf and below this an early Miocene interval that was also glacigenic. Core properties that were studied include fracture patterns, porosity, sonic velocity and magnetic susceptibility. Velocity in particular was useful in relating the cored sequence to the regional seismic stratigraphy. A preliminary assessment suggests that the bottom of the hole is 15 m short of the boundary between seismic sequences V3 and V4. Analytical facilities new to the Antarctic and used for processing samples for the project are described here and include a bench top palynological processing system and a palaeomagnetic laboratory. The core management and sampling system, which recorded over 2000 samples, is also outlined.

Age determination of sediment cores from the western Svalbard margin

The deglaciation of the continental shelf to the west of Spitsbergen and the main fjord, Isfjorden, is discussed based on sub-bottom seismic records and sediment cores. The sea floor on the shelf to the west of Isfjorden is underlain by less than 2 m of glaciomarine sediments over a firm diamicton interpreted as till. In central Isfjorden up to 10 m of deglaciation sediments were recorded, whereas in cores from the innermost tributary, Billefjorden, less than a meter of ice proximal sediments was recognized between the till and the 'normal' Holocene marine sediments. We conclude that the Barents Sea Ice Sheet terminated along the shelf break during the Late Weichselian glacial maximum. Radiocarbon dates from the glaciomarine sediments above the till indicate a stepwise deglaciation. Apparently the ice front retreated from the outermost shelf around 14.8 ka. A dramatic increase in the flux of line-grained glaciomarine sediments around 13 ka is assumed to reflect increased melting and/or current activity due to a climatic warming. This second stage of deglaciation was interrupted by a glacial readvance culminating on the mid-shelf area shortly after 12.4 ka. The glacial readvance, which is correlated with a simultaneous readvance of the Fennoscandian ice sheet along the western coast of Norway, is attributed to the so-called 'Older Dryas' cooling event in the North Atlantic region. Following this glacial readvance the outer part of Isfjorden became rapidly deglaciated around 12.3 ka. During the Younger Dryas the inner fjord branches were occupied by large outlet glaciers and possibly the ice front terminated far out in the main fjord. The remnants of the Barents Sea Ice Sheet melted quickly away as a response to the Holocene warming around 10 ka.

Planktonic foraminifera in the Maastrichtian of Hor-Hahar

The lengthy warm, stable climate of the Cretaceous terminated in the Campanian with a cooling trend, interrupted in the early and latest Maastrichtian by two events of global warming, at ~70-68 Ma and at 65.78-65.57 Ma. These climatic oscillations had a profound effect on pelagic ecosystems, especially on planktic foraminiferal populations. Here we compare biotic responses in the tropical-subtropical (Tethyan) open ocean and mesotrophic (Zin Valley, Israel) and oligotrophic (Tunisia) slopes, which correlate directly with global warming and cooling. The two warming events coincide with blooms of Guembelitria, an extreme opportunist genus best known as the main survivor of the Cretaceous-Paleogene (K-Pg) catastrophe. In the Maastrichtian, Guembelitria bloomed in the uppermost surface water above shelf and slope environments but failed to reach the open ocean as it did at K-Pg. The coldest interval of the late Maastrichtian (~68-65.78 Ma) is marked by an acme of the otherwise rare species Gansserina gansseri, a deep-dwelling keeled globotruncanid. The G. gansseri acme event can be traced from the deep ocean even onto the Tethyan slope, marking copious production and circulation of cold intermediate water. This acme is abruptly terminated by extinction of the species, a dramatic reversal attributed to a short-term global warming episode. This extinction corresponds precisely with the second bloom of Guembelitria that began ~300 kyr prior to the K-Pg event. The antithetical relationship between blooming of Guembelitria and the G. gansseri acme reflects planktic foraminiferal sensitivity to warm-cool-warm-cool climatic oscillations marking the end of the Cretaceous.

Pollen records from two outcrop profiles in the Kakitu Mountain of Tibet

There is a great similarity between pollen types which occur in the early Holocene NE Tibetan pollen spectra and those which are commonly considered to be typical for the Würm Late Glacial period in Central Europe and for the Würm Pleniglacial period in Southern Europe. Evidently, this similarity is due to a remarkable general conformity of plant taxa growing in cold-arid regions of the northern hemisphere. The improvement of the climate and the retreat of the glaciers that commenced at the end of the Würm period had already terminated definitely before 9500 BP. In addition, the climatic situation as well as the vegetation belts must have remained rather constant during the following 3000 yr, i.e. through most parts of the climatic optimum of the Holocene.

Immobilisation data of juvenile and adult male and female Weddell seals at Drescher Inlet from expeditions DRE1990 and DRE1992

Fourten Weddell seals (Leptonychotes weddellii) and two crabeater seals (Lobodon carcinophaga) were immobilised at Drescher Inlet (Riiser Larsen Ice Shelf), eastern Weddell Sea coast, between January and February 1990 using a combination of ketamine, xylazine, and diazepam. Eleven Weddell seals were drugged once, and two and one were drugged two and three times each, coming to a total of 18 immobilisation procedures. Another 16 seals were immobilised between January and February 1992. Ten seals were drugged once, and three and two were drugged two and three times each, coming to a total of 25 immobilisation procedures. Narcoses were terminated with yohimbine. Data as given by doi:10.1594/PANGAEA.438920 were selected for publication. Data sets doi:10.1594/PANGAEA.438921 and doi:10.1594/PANGAEA.438926 followed the same methods and dose regimes.

Mineral composition and major element oxides of gabbros from ODP Hole 179-1105A, southwest Indian Ridge

Hole 1105A penetrated 158 m of gabbros at a site offset 1.3 km east-northeast from Hole 735B on the Atlantis Bank near the Atlantis II Fracture Zone. A total of 118 m of dominantly medium- to coarse-grained intercalated Fe-Ti oxide gabbro and olivine gabbro was recovered from Hole 1105A that shows many petrographic features similar to those recovered from the upper part of Hole 735B. The main rock types are distinguished based on the constituent cumulus phases, with the most primitive gabbros consisting of olivine, plagioclase, and clinopyroxene. The inferred crystallization order is subsequently Fe-Ti oxides (ilmenite and titanomagnetite), followed by orthopyroxene, then apatite, and finally biotite. Orthopyroxene appears to replace olivine in a narrow middle interval. The magmatic evolution is likewise reflected in the mineral compositions. Plagioclase varies from An66 to An28. Olivine varies from Fo78 to Fo35. The gap in olivine crystallization occurs between Fo46 and Fo40 and coincides approximately with the appearance of orthopyroxene (~En50). The clinopyroxenes show large compositional variation in Mg/(Mg + Fe total) from 0.84 to 0.51. The nonquadrilateral cations of clinopyroxene similarly show large variations with Ti increasing for the olivine gabbros and decreasing for the Fe-Ti oxide gabbros with the decrease in Mg/(Mg + Fe total). The apatites are mainly flourapatites. The compositional variation in the gabbros is interpreted as a comagmatic suite resulting from fractional crystallization. Pyroxene geothermometry suggests equilibration temperatures from 1100°C and below. The coexisting Fe-Ti oxide minerals indicate subsolidus equilibration temperatures from 900°C for olivine gabbros to 700°C for the most evolved apatite-bearing gabbros. The cryptic variation in the olivine gabbros defines two or three lenses, 40 to 60 m thick, each characterized by a distinct convex zoning with a lower segment indicating upward reverse fractionation, a central maximum, and an upper segment showing normal fractionation. The Fe-Ti oxide gabbros show cryptic variations independent of the host olivine gabbros and reveal a systematic upward normal fractionation trend transgressing host olivine gabbro boundaries. Forward fractional crystallization modeling, using a likely parental magma composition from the Atlantis II Fracture Zone (MgO = 7.2 wt%; Mg/[Mg + Fe2+] = 0.62), closely matches the compositions of coexisting olivine, plagioclase, and clinopyroxene. This modeling suggests cosaturation of olivine, plagioclase, and clinopyroxene from 1155°C and the addition of Fe-Ti oxides from 1100°C. The liquid line of descent initially shows increasing FeO with moderately increasing SiO2. After saturation of Fe-Ti oxides, the liquid strongly decreases in FeO and TiO2 and increases in SiO2, reaching dacitic compositions at ~10% liquid remaining. The calculations indicate that formation of olivine gabbros can be accounted for by <65% fractionation and that only the residual 35% liquid was saturated in Fe-Ti oxides. The modeling of the solid fractionation products shows that both the olivine gabbro and the Fe-Ti oxide gabbros contain very small amounts of trapped liquid (<5%). The implications are that the gabbros represent crystal mush that originated in a recharging and tapping subaxial chamber. Compaction and upward melt migration in the crystal mush appear to have been terminated with relatively large amounts of interstitial liquid remaining in the upper parts of the cumulate mush. This termination may have been caused by tectonic disturbances, uplift, and associated withdrawal of magma into the subaxial dike and sill system. Prolonged compaction and cooling of the trapped melt in the mush formed small differentiated bodies and lenses by pressure release migration and crystallization along syntectonic channels. This resulted in differentiation products along lateral and vertical channelways in the host gabbro that vary from olivine gabbro, to Fe-Ti oxide gabbro, gabbronorite, and apatite gabbros and show large compositional variations independent of the host olivine gabbros.