In this study, we developed a high-throughput growth measurement system, Colony-live, in order to strike a better balance between throughput and measurement quality. The Colony-live system is designed to provide three growth characteristic values based on growth kinetics (Figure 1), information which is absent in the conventional method. Our system increases the ability to identify unique phenotypes among the wide range of phenotypes in genome-wide studies. Such detailed examination of phenotype alterations allows a deeper understanding of the effects of gene knockout on a cellular system.
To measure growth kinetics accurately and precisely, we focused on reducing the neighbor effect, which can cause reduced measurement quality. We took a new approach to this issue: we referred to the colony growth model as a basis for reducing the neighbor effect. According to the model  and experimental observations, cells in peripheral regions of a colony replicate faster than cells in the central region. Since the peripheral region becomes closer to neighboring colonies as the colony size increases, the neighbor effect will be more marked in this region as the colony size increases. To minimize the neighbor effect, we derived quantification values of colony mass of the center region (mass*), thereby excluding growth of the peripheral region of large colonies. As expected, the neighbor effect decreased when we took mass* as a quantification value, and the measurement accuracy and precision also successfully improved (Figure 2).
The utility of the Colony-live system was tested by genome-wide screening experiments with the Keio collection, in which we found two practical advantages of the Colony-live system over the conventional method. First, owing to the improvement of measurement accuracy and precision, the Colony-live system detected mild growth defect phenotypes, which were undetectable by the conventional method (Figure 3). Under the conditions examined in this study, we found that growth measurement at 20 h incubation time for an average diameter of 17 pixels produced the most accurate growth measurements for Colony-live. For growth under other conditions or other cells, both the incubation time and average diameter for evaluation would need to be determined empirically. Second, owing to the measurement of growth kinetics, the Colony-live system classified two distinct growth alterations, the long LTG group and the low SPG group (Figure 4). Importantly, the functions of knockout genes in each group differed (Figure 5).
The long lag time group (LTG) was linked to two aspects of protein synthesis, the sulfur relay pathway for tRNA modification  and ribosome maturation (Figure 5). Why did these mutants display a prolonged lag time? tRNA modification is thought to be functionally important for translational efficiency and fidelity , e.g., ribosome maturation. These mutants may prolong the turnaround time of new protein synthesis. The biosynthesis of a large number of new proteins is required during early growth because the global expression profile changes dramatically during the transition from stationary phase to exponential growth phase . Accordingly, the time required for early growth, lag time, was prolonged in these mutants. Since we found other gene classes in the long LTG group, which have functions unrelated to protein synthesis, it is likely that there are multiple mechanisms that can prolong lag time.
The less efficient group (SPG) was linked to aerobic respiration (Figure 5). Interestingly, we found that mutations that disrupt the cysteine biosynthetic pathway also results in a low SPG (see Additional file 1: Figure S5). Our interpretation is that an insufficient supply of cysteine can decrease the activity of the respiration pathway. Physiological connection between cysteine and respiration is that cysteine has defense activity against oxidative stress , which is mainly generated by aerobic respiration . In addition, cysteine is required to form Fe-S clusters; essential cofactors of NADH dehydrogenase I, a central enzyme in the respiratory chain. Indeed, ongoing genetic interaction studies within our laboratory using double knockout mutants of E. coli, strongly support a functional connection between the cysteine biosynthetic pathway and respiration (unpublished data).
The current Colony-live system has several issues that we need to take into consideration. The first is that morphological abnormalities of the colonies are overlooked. Colony morphology is sometimes changed by SKO mutation, as exemplified by the dnaK mutant (see Additional file 1: Figure S2) whose growth defect was missed by the Colony-live system (Figure 3A). Although Colony-live system detected severe and mild growth defects successfully (Figure 3), development of an analysis method to quantify the morphological abnormality of a colony should further enhance the measurement quality. The second issue is the growth effect of light exposure by periodic scanning. All colonies were exposed to strong light for about a second during every scan. Since E. coli cells have the potential to sense visible light [37, 38] or heat generated, the strong light of the scanner may affect the growth of E. coli. We confirmed that the light does not affect growth of individual SKO mutants during scanning (see Additional file 1: Figure S6). However, light and scanner heat have the potential to affect colony growth under specific conditions of interest and should be considered. The optimization of parameters, such as incubation time, diameter of the center region, etc., may be required for growth measurements under other conditions or for other species.