Comet 103P/Hartley 2 made a close approach to the Earth in October 2010. It was the target of an extensive observing campaign including ground-and orbit-based observatories and was visited by the Deep Impact spacecraft in the framework of its mission extension EPOXI. We present observations of HCN and CH 3 OH emission lines conducted with the IRAM Plateau de Bure interferometer on 22-23, 28 October and 4, 5 November 2010 at 1.1, 1.9 and 3.4 mm wavelengths. The thermal emission from the dust coma and nucleus is detected simultaneously. Interferometric images with unprecedented spatial resolution of ∼100 to ∼500 km are obtained. A sine-wave like variation of the thermal continuum is observed in the 23 October data, that we associate with the nucleus thermal light curve. The nucleus contributes up to 30-55 % of the observed continuum emission. The dust thermal emission is used to measure the dust production rate. The inferred large dust-to-gas ratio (in the range 2-6) can be explained by the unusual activity of the comet for its size, which allows decimeter size particles and large boulders to be entrained by the gas due to the small nucleus gravity. The rotational temperature of CH 3 OH is measured with beam radii from ∼150 km to ∼1500 km. We attribute the increase from ∼35 K to ∼46 K with increasing beam size to radiative processes. The HCN production rate displays strong rotation-induced temporal variations, varying from ∼0.3×10 25 s −1 to ∼2.0×10 25 s −1 in the 4-5 November period. The HCN production curve, as well as the CO 2 and H 2 O production curves measured by EPOXI, are interpreted with a geometric model which takes into account the complex rotational state of 103P/Hartley 2 and its shape. The HCN and H 2 O production curves are in phase, showing that these molecules have common sources. The ∼ 1.7 h delay, in average, of the HCN and H 2 O production curves with respect to the CO 2 production curve suggests that HCN and H 2 O are mainly produced by subliming icy grains. The scale length of production of HCN is determined to be on the order of 500-1000 km, implying a mean velocity of 100-200 m s −1 for the icy grains producing HCN. From the time evolution of the insolation of the nucleus, we show that the CO 2 production is modulated by the insolation of the small lobe of the nucleus. The three-cycle pattern of the production curves reported earlier is best explained by an overactivity of the small lobe in the longitude range 0-180• . The good correlation between the insolation of the small lobe and CO 2 production is consistent with CO 2 being produced from small depths below the surface. The time evolution of the velocity offset of the HCN lines, as well as the displacement of the HCN photocenter in the interferometric maps, are overall consistent with this interpretation. Other localized sources of gas on the nucleus surface are also suggested.