In the first part of the thesis, changes of the Atlantic meridional overturning circulation (AMOC) in the mid-Holocene compared to the pre-industrial state are explored in different coupled climate models. Using time-slice integrations by a newly developed global finite-element model ECHAM6-FESOM with unstructured mesh and high resolution, our simulations show an enhanced mid-Holocene AMOC, accompanied by an increase in the ocean salinity over regions of deep water formation. We identify two different processes affecting the AMOC: 1) a more positive phase of North Atlantic Oscillation (NAO) increases water density over the Labrador Sea through anomalous net evaporation and surface heat loss; 2) a decreased import of sea ice from the Arctic causes a freshwater reduction in the northern North Atlantic Ocean. Using the coupled model ECHAM6-MPIOM in T63GR15 and T31GR30 grids, we find that the simulated AMOC is strongly affected by the model resolution. In detail, stronger-than-present mid-Holocene AMOC is revealed by simulations with the T63GR15 grid, which resembles the result of ECHAM6-FESOM, while a decline of the mid-Holocene AMOC is simulated by the low resolution model with the T31GR30 grid. Such discrepancy can be attributed to different changes in Labrador Sea density which is mainly affected by 1) NAO-induced net precipitation, 2) freshwater transport from the Arctic Ocean, and 3) the strength of AMOC itself. Finally, we analyzed available coupled climate models showing a diversity of responses of AMOC to mid-Holocene forcings, most of which reveal positive AMOC changes related to northern high latitudes salinification. Sensitivity of the simulated climate to the early-Holocene (9k) insolation, greenhouse gases (GHGs) and topography is examined in the second part of the thesis, by performing timeslice experiments under pre-industrial and 9k regimes using ECHAM6-FESOM. Under the early-Holocene orbit and GHGs, the ECHAM6-FESOM simulation shows a warming in boreal summer and a cooling in boreal winter from over mid and high latitudes compared to pre-industrial, with amplification over the continents; as well as a reduction of sea ice in the Arctic and Southern Oceans. A reduced sea ice transport through the Fram Strait leads to a stronger-than-present Atlantic Meridional Overturning Circulation (AMOC) in the early-Holocene. Including the early-Holocene topography and continental ice sheet over North America leads to an additional regional cooling year-round. The resulted enhanced sea ice thermodynamic production over Baffin Bay and North Atlantic subpolar gyre is the cause for a more saline surface over the region of deep water formation. There are big discrepancies in the oceanic responses to different locations of freshwater discharge. Laurentide Ice Sheet (LIS) coastal melting only leads to a freshening over the Gulf Stream and Canary Current, with no meltwater advection to the deep water formation sites, therefore not affecting the strength of the thermohaline circulation. In contrast, adding freshwater into the Labrador Sea produces a significant decrease in ocean salinity over the North Atlantic region from sea surface to 200 m depth, contributing to a decline of the Atlantic meridional overturning circulation (AMOC). All early-Holocene experiments reveal a change of the westerlies over the North Atlantic section, accompanied by a more positive North Atlantic Oscillation (NAO) phase, which is led by the corresponding divergence anomalies of the Eliassen-Palm (E-P) flux. The enhanced westerly wind at 50 $^{\circ}$N provides a barrier which prevents the Arctic cold air from invading into the lower latitudes. This circulation change in the atmosphere leads to less frequent episodes of blocking patterns which further results in decreased cold surges over most parts of the Northern Hemisphere continents, in particular the Europe and Asia in the early-Holcene compared to pre-industrial. Finally, the reduced cold air outbreak events, together with a relatively dry atmospheric condition, are the causes for a reduced snowfall over Europe and Asia. Another aim of the thesis (the third part) is to examine to what degree the area-thickness distribution of new ice formed in open water affects the ice and ocean properties. Two sensitivity experiments are performed which modify the horizontal-to-vertical aspect ratio of open-water ice growth. The resulting changes in the Arctic sea-ice concentration strongly affect the surface albedo, the ocean heat release to the atmosphere, and the sea-ice production. Furthermore, our simulations show a positive feedback mechanism among the Arctic sea ice, the Atlantic Meridional Overturning Circulation (AMOC), and the surface air temperature in the Arctic, as the Fram Strait sea ice import influences the freshwater budget in the North Atlantic Ocean. Anomalies in sea-ice transport lead to changes in sea surface properties of the North Atlantic and the strength of AMOC. For the Southern Ocean, the most pronounced change is a warming along the Antarctic Circumpolar Current (ACC), especially for the Pacific sector. Another insight of this study lies on the improvement of FESOM-ECHAM6. FESOM-ECHAM6 is a newly developed global climate model with an unstructured mesh and multi-resolution. It has relatively high ability for producing many aspects of the observed climate, but also has shortcomings when simulating the subpolar sea-ice boundary in the Northern Hemisphere and the strength of AMOC. We find that such disadvantages can be improved by tuning the process of open-water ice growth, which strongly influences the sea ice concentration in the marginal ice zone and affects the North Atlantic circulation through regulating the import of Arctic sea ice volume. In reality, the distribution of new ice on open water relies on many uncertain parameters, for example, surface albedo, wind speed and ocean currents. Knowledge of the detailed processes is currently too crude for those processes to be implemented realistically into models. Our sensitivity experiments indicate a pronounced uncertainty related to open-water sea ice growth which could significantly affect the climate system sensitivity.