Spatial induction of genes encoding secreted proteins in micro-colonies of Aspergillus niger

Aspergillus niger is used by the industry to produce enzymes and metabolites such as citric acid. In liquid cultures, it can grow as a dispersed mycelium or as micro-colonies with a width in the micrometer to millimeter range. Here, it was assessed whether expression of genes encoding secreted enzymes depends on mycelium morphology. To this end, expression of the reporter gene gfp from the promoters of the glucoamylase gene glaA, the feruloyl esterase gene faeA and the α-glucuronidase gene aguA was causally related to micro-colony size within a liquid shaken culture. Data could be fitted by hyperbolic functions, implying that the genes encoding these secreted proteins are expressed in a shell at the periphery of the micro-colony. The presence of such a shell was confirmed by confocal microscopy. Modelling predicted that the width of these zones was 13 to 156 µm depending on growth medium and micro-colony diameter. Together, data indicate that the highest productive micro-colonies are those colonies that have a radius ≤ the width of the peripheral expression zone.

micro-colonies with diameters < 2.5 th , > 2.5 th < 25 th , > 25 th < 75 th , > 75 th < 97.5 th and > 97.5 th percentile (Fig. 2). Transfer from CMX to MMX resulted in an increased proportion of micro-colonies within the smallest 2.5% of diameters as well as the largest 25% of diameters. At the same time, the frequency of micro-colonies of median sized micro-colonies was reduced. This shift was exacerbated when CMX grown micro-colonies were transferred to MMM, while transfer of micro-colonies from CMX to CMX only led to an increase in proportion of the micro-colonies within the largest 25% diameters and a decrease in proportion of micro-colonies within the median and smaller 25% of diameters. These findings were confirmed by multinomial logistic regression (data not shown). Together, data show that transfer to fresh medium promotes radial growth of N402 micro-colonies and, depending on the medium, can also result in partial micro-colony fragmentation.
Like wild-type N402, biomass of strains expressing gfp from the glaA, faeA or aguA promoter increased upon transfer of 16 h old mycelium of non-inducing CM to inducing MM or CM. Biomass of the glaA::GFP, aguA::GFP and faeA::GFP strains had increased in MM from 0.71 to 0.78 g, 0.73 to 0.83 g and 0.73 to 0.93 g during a 4 h period, respectively. When transferred to inducing CM, biomass of the glaA::GFP, aguA::GFP and faeA::GFP strains had increased to 0.85, 1.47 and 1.49 g, respectively. Also in these strains, fragmentation occurred when mycelium was transferred to fresh medium (Fig. 3). In fact, mean micro-colony diameter of glaA::GFP (1361 µm), aguA::GFP (1142 µm) and faeA::GFP (1230 µm) was reduced by 140, 145 and 250 µm, respectively, after a 4 h transfer to inducing MM. This reduction was 180, 130 and 79 µm for glaA::GFP, aguA::GFP and faeA::GFP, respectively, after transfer to CM. Transfer of reporter strains to inducing MM or CM coincided with a decrease in the proportion of micro-colonies within the largest 25% of diameters and an increase in proportion of micro-colonies within the smallest 25% of diameters (Fig. 4). Proportions of median sized micro-colonies also decreased for the aguA::GFP and faeA::GFP strains. In contrast, the glaA::GFP strain showed a marked increase in the proportion of median sized micro-colonies. Increase and decrease of proportions of size categories were of lesser magnitude after transfer of fluorescent strains to inducing CM. Diameters of small, median and large micro-colonies of the faeA::GFP and aguA::GFP strains had reduced after transfer to inducing MM. For glaA::GFP, diameter of median and large microcolonies had also reduced after transfer to inducing MM. For the faeA::GFP and aguA::GFP strains transferred to inducing CM the smallest 2.5% of micro-colonies had increased in diameter, while small and median micro-colonies had reduced diameter. Large micro-colonies of aguA::GFP were also reduced in diameter, while large micro-colonies of faeA::GFP showed a slight increase in diameter. Micro-colony diameter of the glaA::GFP strain was reduced for median and large sized colonies but had increased for small micro-colonies. These findings were confirmed by multinomial logistic regression (data not shown). Together, biomass of the reporter strains increased after transfer to inducing medium but size of micro-colonies was reduced by micro-colony fragmentation.
Relation between micro-colony size and expression of genes encoding secreted proteins. The relation between micro-colony diameter and GFP fluorescence intensity was studied in the reporter strains glaA::GFP, aguA::GFP and faeA::GFP. As expected, non-induced strains or the wild-type strain N402 showed low or no fluorescence (Supplemental Fig. 1E). Total fluorescence intensity of reporter strains transferred to inducing medium increased quadratically with increasing diameter (Supplemental Fig. 2). Consequently, fluorescence intensity per micro-colony volume (FV −1 ) per diameter was best fitted by a hyperbolic function ( Fig. 5A-E, Supplemental Fig. 3A-E). In this function, FV −1 is maximal at the vertical asymptote x b of the hyperbolic function, while with increasing diameter x, FV −1 decreases. Maximal FV −1 of glaA::GFP, aguA::GFP and faeA::GFP micro-colonies that had been induced in MM was found at a diameter of 288, 211 and 26 µm, respectively (Table 1). When strains were induced in CM maximal FV −1 was found at a diameter of 169 and 216 µm for aguA::GFP and faeA::GFP, respectively. Conservative selection parameters derived from micro-colony profiles could lead to micro-colonies being omitted from analysis unintentionally. Therefore, the diameter where FV −1 is maximal could be smaller in reality. This goes particularly for the CMM induced glaA::GFP strain as a is equal to 0, making b an arbitrary number intersecting γ with a minimum concurrent with the smallest recorded diameter. A hyperbolic function also has a horizontal asymptote (γ) that approaches a steady state. For glaA::GFP, aguA::GFP and faeA::GFP induced in MM this steady state was reached at 1556, 1381 and 954 µm, respectively. The horizontal asymptote was reached at a diameter of 1254 and 1532 when aguA::GFP and faeA::GFP were induced in CM (Table 1). Slopes did not converge to a steady state in the case of CMM-induced glaA::GFP.
Increase in fluorescence intensity in the region between x b and x γ cannot be explained solely by increase in micro-colony volume when parameter α in Eq. 2 is shown not to be zero. This suggests that fluorescence is not evenly distributed over micro-colony volume but occurs in a concentric peripheral shell (Supplemental Fig. 4). In the case of MM, change in width was < 10%. Modelling inferred that in the case of MM expression of faeA occurs in a relatively small concentric shell with a constant radius of 13 µm. In contrast, glaA is expressed in a concentric outer shell that increases in width from 144 to 157 µm as micro-colonies increase in diameter from 288 to 1556 µm. On the other hand, aguA was predicted to be expressed in a concentric outer shell with a width decreasing from 105 µm to 92 µm in the 210 to 1381 µm size range of micro-colonies. When reporter strains were induced in CM, glaA was predicted to be expressed in a concentric outer shell that increases with a square root relationship. On the other hand, aguA was predicted to be expressed in a concentric outer shell that has a width decreasing from 85 µm to 37 µm in the 169 to 1245 µm size range of micro-colonies. Moreover, faeA was predicted to be expressed in a concentric outer shell with a width decreasing from 85 µm to 31 µm in www.nature.com/scientificreports www.nature.com/scientificreports/ the 230 to 1532 µm size range of micro-colonies. Together, upon growth of the micro-colonies, the volume of the expression zone increases. At the same time its relative radius and volume decreases compared to the total diameter and volume of the pellet.

Discussion
Morphology of the mycelium in liquid cultures such as bioreactors depends on various factors including inoculum size, surface composition of the inoculum, medium composition, and mixing conditions 1,7,11 . In addition, presence of molecules such as chelating agents or anionic polymers may affect morphology. Here, it was shown for the first time that transfer of mycelium to fresh medium, in particular when changing the carbon source, also affects the morphology of the mycelium by inducing its fragmentation. Future studies will assess the underlying mechanisms and whether the observed fragmentation can be used to optimize productivity in liquid cultures. So far, it had not been established how secretion of proteins relates to the morphology of the mycelium. Driouch et al. 12 showed that 400-µm-wide micro-colonies express the glucoamylase gene glaA throughout the   www.nature.com/scientificreports www.nature.com/scientificreports/ mycelium while expression in mm-sized microcolonies is only observed at the outer periphery. However, this study made use of the presence or absence of titanate micro-particles to control morphology. Therefore, changes in spatial gene expression may have been the result of the addition of the micro-particles. Here, we did not compare two cultures with different composition but rather made use of the heterogeneity of micro-colony morphology within a liquid shaken culture. Expression of glaA, faeA and aguA per micro-colony volume generally decreased with increasing micro-colony size in a hyperbolic way, eventually approaching a constant expression per micro-colony volume. The hyperbolic relationship between colony diameter and fluorescence per volume is explained by expression taking place in a concentric shell at the periphery of the micro-colony. Expression in such a concentric shell was confirmed by fluorescence microscopy. When compared to confocal microscopy data models were reasonably accurate in predicting the fluorescent radius of glaA and aguA after transfer to inducing minimal medium; predicting the fluorescent radius of glaA::GFP within its 95% confidence interval and underpredicting the fluorescent radius for aguA::GFP by 16%. Yet, the expression zone of faeA deviated by 72%. For reporter strains transferred to inducing complete medium the fluorescent radius of the glaA::GFP strain could not be predicted because there were only 3% entirely fluorescent colonies in the culture. The fluorescent radius of aguA::GFP and faeA::GFP was under-predicted by 64% and 74%, respectively. Underprediction of the expression zone of faeA may be caused by the relatively low expression of this gene, while underprediction of expression zones may also be caused by selection of micro-colonies in the confocal analysis that had not yet reached their steady state (i.e. they were too small). Particle analysis is therefore the preferred tool to assess expression zones because it takes into account all micro-colonies sizes in a culture instead of analysing single micro-colonies.
GFP expression was also found at the outer part of macro-colonies 13 . In contrast, the spatial expression in micro-colonies resulting from the aguA and faeA promoters was not in accordance with those in macro-colonies 13 . These genes were found to be expressed throughout the macro-colony and the colony centre, respectively. The different spatial expression patterns in micro-and macro-colonies may be explained by inducer penetration. Micro-colonies in liquid shaken cultures are 3D structures. Penetration of the inducer into the centre of the colony may be difficult. In contrast, macro-colonies grown on solid medium are near 2D and therefore the inducer in the underlying medium has access to all zones of the colony. However, this does not explain why the radius of the fluorescent concentric shell of the faeA::GFP strain is relatively small when compared to the aguA::GFP strain despite the fact that both genes are induced by xylose. Possibly, these genes respond to different concentrations of the inducer.
The fact that expression of the genes encoding secreted proteins occurs at the periphery of the micro-colony in a shell with a relatively constant width means that the volume of this shell decreases relatively to the total volume of the micro-colony when the micro-colony becomes larger. Thus, cultures with uniform small micro-colonies (diameters at x b ) would be more productive than cultures with uniform large micro-colonies (diameters at) assuming similar biomass is formed per volume culture medium. Glucoamylase, α-glucuronidase and feruloyl esterase production would then be 54, 56 and 91% less efficient at steady state diameters compared to micro-colonies with diameters at x b (i.e. diameter where expression per volume unit is maximal) if grown in minimal medium. In inducing complete medium, α-glucuronidase and feruloyl esterase production would be 61 and 65% less efficient at steady state diameters compared to micro-colonies with diameters at x b .  Table 1. Values of relevant parameters from models describing the relation between micro-colony size and expression of gfp from the glaA, aguA and faeA promoters. Numbers between brackets indicate the 95% confidence interval. x b is micro-colony diameter where fluorescence intensity per volume (FV −1 ) is maximal, x γ represents the diameter where FV −1 remains constant when diameter increases, r γ is the predicted radius of the fluorescent band at this steady state and r γ /0.5x γ represents the percentage of the radius of the micro-colony that shows fluorescence.

Methods
Strains and culture conditions. Aspergillus niger strains N402, UU-A005.4, AR9#2 and AV11#3 were used in this study ( Table 3). The former strain was used as a control, while the latter three strains (called faeA::GFP, glaA::GFP and aguA::GFP in this chapter) express gfp from the faeA, glaA and aguA promoter, respectively. Strains were grown at 30 °C on minimal medium (MM 14 ,) with 25 mM xylose and 1.5% agarose (MMXA). MMXA cultures were grown for three days, after which conidia were harvested using Saline-Tween (0.8% NaCl and 0.005% Tween-80). 250 ml liquid cultures were inoculated with 1.25*10 9  . This formula was inferred from the TOF of beads with diameters of 42 µm, 250 µm and 500 µm. Micro-colony volume (V) was calculated from the diameter assuming spherical morphology. To analyze the relation between micro-colony morphology and gene expression, particles with a TOF <165 and >13005 arbitrary units (AU) were removed from the dataset because they are outside the object size range of the FOCA (i.e. <30 µm and >1500 µm). In addition, particles were removed with  www.nature.com/scientificreports www.nature.com/scientificreports/ fluorescence peaks <80 AU. Peaks >80 AU but with a fluorescence width <2000 AU were also removed. Finally, particles were removed that had an integrated density <25 AU for the green, red and yellow channels. The integrated green value (in AU) from the Biosorter was taken as a measure for fluorescence intensity (F). Micro-colonies of the non-fluorescent wild-type strain N402 were selected similarly but in this case particles were removed that had fluorescence peaks <100 AU. Additionally, integrated density of fluorescent signals was no longer a selection variable.
Differences in size distributions were quantified using two-sample Kolmogorov-Smirnov tests. Particles were distributed into five size categories based on diameter. Micro-colonies with diameters < 2.5 th , > 2.5 th < 25 th , > 25 th < 75 th , > 75 th < 97.5 th and > 97.5 th percentile were designated smallest, small, median, large and largest, respectively. Differences in ratio between categories were analyzed using chi-square tests. Change of diameter after transfer to different media was described by resampling with 1000-fold replacement. For each resample the mode was recorded as the central tendency. The median of the modes and the 2.5 th and 97.5 th percentile provided an approximation of the mode and its confidence intervals. For biomass quantification, micro-colonies were not fixed, filtered using coffee filters, transferred to 50 ml falcon tubes and freeze-dried. Differences in dry weight were analyzed using Kruskal-Wallis H test.
Modelling GFP fluorescence relative to micro-colony diameter. Median fluorescence intensity data of micro-colonies that had been transferred to inducing medium were modelled with the R package quantreg 15 using non-linear quantile regression. Data pertaining to the fluorescence intensity per unit of volume (mm 3 ) were best described by the non-linear hyperbolic functions Eqs. 1 and 2.
In these equations, F denotes fluorescence intensity and V micro-colony volume (mm 3 ). In Eq. 1, x denotes micro-colony diameter (µm), a represents the maximal fluorescence intensity per volume, c denotes the arbitrary intercept of this function and b represents the location of the vertical asymptote (i.e. the diameter x b of the micro-colony where fluorescence intensity per volume is maximal). In Eq. 2, α denotes the maximal fluorescence intensity per volume, while γ represents a steady state between increase in fluorescence intensity and increase in volume. The x coordinate x γ of the intersect of the lower 95% confidence limit of Eq. 1 and γ gives the diameter of micro-colonies that have reached this steady state.
Parameters x b and x γ were transformed to their corresponding surface areas, S b and S γ (mm 2 ) and used as input in EQ3 to determine if fluorescence intensity increases proportionally to micro-colony size. Using S b and S γ as input for S, EQ3 gives the minimal  ( ) 1 should be input in Eq. 6 for fluorescent shells that decrease or increase in width with surface area, respectively. The relative fluorescent radius of a micro-colony is then given by Eq. 7.

= .
+ . I x z x 0 5 0 5 (7) x V s b Finally, the percentage of fluorescent micro-colony volume at the steady state is given by Eq. 8, in which V 0 is the total volume of the micro-colony and V 1 is the non-fluorescent volume. When micro-colonies were transferred to CM, equations described above remained valid for the aguA::GFP and faeA::GFP strains. However, the glaA::GFP strain was now best described by a horizontal line at γ, with b as an arbitrary intersect of this line. confocal laser scanning microscopy. Fluorescence of GFP was localized in micro-colonies using a DMI 6000 CS AFC confocal microscope (Leica, Mannheim, Germany). Micro-colonies were fixed and washed (see above), transferred into a glass bottom dish (Cellview ™ , Greiner Bio-One, Frickenhausen, Germany, PS, 35/10 MM) and embedded in 1% low melting point agarose at 45 °C. Micro-colonies were imaged at 20x magnification (HC PL FLUOTAR L 20 × /0,40 DRY). GFP was excited by white light laser at 472 nm using 50% laser intensity (0.1 kW/cm 2 ) and a pixel dwell time of 72 ns. Fluorescent light emission was detected with hybrid detectors in the range of 490-525 nm. Pinhole size was 1 Airy unit. In all cases, z-stacks were made of imaged micro-colonies using 100 slices with a slice thickness of 5.4−10 µm. Image analysis was performed with Fiji 16 .

Data availability
Data are available for readers upon request.