Recombinant protein production in microorganisms is of great interest for the production of biopharmaceuticals, therapeutics and industrial enzymes. However, recombinant protein production has always shown a harmful effect on the microorganism cell physiology when excessively produced. Cell resources (i.e. metabolites, energy, molecular machinery, cytosolic space, etc.) are used to produce the host's proteins and the overproduced gratuitous protein. As a result, this unnatural extra load typically leads to slower growth and lower protein yields, a phenomenon known as ʻburdenʼ. This burden comes from the fact that the recombinant protein has no benefit for the microorganism, and that it only uses cell resources at the expense of the production of the endogenous essential proteins. In my PhD project, the issues were (1) to decipher the consequences of gratuitous protein overproduction on the cell physiology, (2) to identify the limiting type of resources, and (3) to overcome this limitation to improve protein production. To address the first issue (1), we analyzed growth rates, production of several proteins of interest, and genome-wide proteomes of Bacillus subtilis strains overproducing various levels of reporter proteins. The reporter proteins were chosen so that they were easily quantifiable by fluorescence and β-galactosidase activity assays (i.e. GFP, mKate2, LacZ, etc.). To obtain the various levels of expression, we built synthetic sequences made of the assembly of various constitutive and inducible promoters and translation initiation regions (TIR, RBS). Hence, we showed that higher was the amount (and size) of the protein produced, lower were the rates of growth and higher were the cell sizes. For instance, the growth rate decreased down by over 20% when GFP was overproduced above 5% of the total soluble protein amount according to both biochemical and fluorescence assays. To further identify the limiting type of resources (2), we performed a relative protein quantification on the strains overproducing GFP at different levels. Hence, we showed that some non-essential proteins were less abundant in the strains overproducing GFP. We next targeted the reporter proteins for degradation using a synthetic tool previously engineered in B. subtilis, so that amino acids can be recycled back to the pool of cell resources. Degrading the reporter gratuitous protein should also relieve the constraint on the cytosolic density by liberating intracellular space. With a degradation of 50-60% of GFP and mKate2, we observed a 50% restoration of the growth rate. This result together with the proteome analysis suggested that the amount of amino acids (and consequently their utilization in protein synthesis) was the main limiting type of resources. To overcome this limitation and improve protein production (3), we aimed at exploring a synthetic, amino acid recycling system based on the above mentioned degradation system. We decided to improve the targeted degradation system by overproducing the E. coli and B. subtilis ClpXP proteases together with an E. coli adaptor protein SspB. This tool may allow to target proteins for degradation in order to save resources and improve the production of a protein of interest. We showed that the overproduction of either ClpXP or SspB/ ClpXP were sufficient to allow a complete degradation of the proteins produced low and intermediate levels, and up to 50% of degradation of the proteins highly produced. As ClpXP is a protease involved in stress responses, we aimed to know whether the overproduction of ClpXP may have negative consequences on the cell physiology. We therefore performed relative protein quantification on a strain overproducing ClpXP. The results showed that ClpXP overproduction causes a global reorganization on the proteome without affecting the growth rate of the cell.