*3.4. Directionality*

Figure 3 shows the 2D anisotropy for an exemplary network, while Figure 4 compares the 2D anisotropy of a collagen and a collagen–elastin network. The sections of all samples show two preferred directionalities, one around 65◦ and another around −80◦. The directions seem to be inversely populated by collagen and elastin–collagen. Regarding the origin of this preferred orientation, one explanation might be the gelation condition. In fact, the gels were gelled within an incubator placed on a microscope. The incoming air and humidity induced a mild current which might have oriented the fibers accordingly. This is also visible in the Videos S1–S4 where the material is drifting until the polymerization starts and the flow is restricted. Although this drift was unavoidable when using this experimental approach, this technique was used to specifically prepare oriented gels [32]. While the addition of elastin did not change the network directionality, the standard deviation of the angle distribution might have been slightly increased (Figure 4). This direct comparison between the respective standard deviations across the angle distributions of all 10 position reveals a minor significant difference as the Mann–Whitney U test was *p* = 0.08, which was larger than the generally accepted threshold of 0.05. Under the assumption of this threshold,

both distributions would not originate from one set of data, i.e., the addition of elastin would lead to a flattened angle distribution. It can be concluded that elastin's presence interferes with the formation of larger, similarly oriented domains. Mostaço-Guidolin et al. have found the interesting observation that a similar orientation of collagen and elastin fibers in the arterial wall of rabbits was greatest when they were middle-aged and lowest when they were young or old [33]. In the context of our experiments, this might imply that an elastin addition creates less mature networks as it seems to slightly interfere with a proper alignment of collagen fibers. The above-described plate reader experiments support this claim, as they showed an elastin concentration-dependent increase in the polymerization rate. A faster rate means less time for the monomers to perfectly align, such that stacking irregularities can occur. This was already discussed earlier, where a faster rate was suspected to contribute also to the cluster formation. Nonetheless, our analysis of the 2D anisotropy in the collagen and elastin–collagen networks could not conclusively portray a difference in the 2D anisotropy, as the *p*-value was just slightly larger than 0.05, which supports our claim of a lateral elastin–collagen alignment. In fact, a predominantly perpendicular alignment of the collagen and elastin fibers should significantly increase the standard deviation of the angle distribution. Additionally, the above-mentioned bubbles were removed prior to the 2D anisotropy analysis so that the observed potential increase in the standard deviation might as well have originated from this preprocess, implying that there was no real difference at all between the collagen and elastin in terms of the 2D anisotropy. The above arguments are in line with the narrative that if 2D anisotropy as a network metric does not significantly change upon an elastin addition, then the structure cannot be changed, i.e., elastin is incorporated mostly homogeneously into collagen.

**Figure 3.** (**a**) Exemplary skeletonized image of a 2 mg/mL collagen network. (**b**): Angle distribution of (**a**). The degree values are given in the mathematical sense, i.e., 0◦ is pointing to the right. The fit is a Gaussian function, provided by ImageJ.
