Geckos

The study was conducted at Midreshet Ben-Gurion in the northern Negev desert, Israel (30°51′8.27″N 34°47′0.24″E) from summer 2003 until autumn 2004. The study site was a complex of guest rooms surrounded by a two-meter high wall over an area 13x150m. A dense population of the Israeli fan-toed gecko Ptyodactylus guttatus inhabited the premises [30 –32 (link)]. The Israeli fan-toed gecko is a medium-sized, insectivorous, rupicolous, scansorial lizard [6 , 45 ] Zlotkin et al. 2003 [50 (link)], Sion et al. 2020 [33 (link)] in the family Phyllodactylidae [7 (link)] common in mesic and arid parts of the Middle East (Israel, Egypt, Saudi Arabia, Oman, Palestine, Jordan, and Syria). It often inhabits cliffs or masonry walls where it can easily be observed from a distance ([48 , 47 ], 2016 [6 , 15 (link), 30 –32 (link), 46 ]).
Fifty-five geckos were hand captured, measured (morphometrics) and scanned (DXA) and released at the site of capture. Of these, 30 gecko’s scan data were included in the comparison, since their body mass was above the lowest possible accurate reading with minimal body mass (> 4.8 g) as indicated in the results (Table 1). The snout vent length (SVL) of these 30 geckos was 60.6–91.7 mm and their body mass 4.99–22.5 g. We used these 30 geckos to compare the real wet mass (as measured by a scale) and the wet mass measure by DXA (see below). From each captured gecko, we recorded the mass using Ohaus digital scale to 0.1 g precision, snout-vent length (SVL), using digital calipers, and the width at the base of the tail. Six additional individuals were captured and euthanized for the calibration necessary for this study (two males, three females and one too small to be sexed without a probe) The smallest gecko (55.6 mm) with body mass 4.2 g was excluded to improve accuracy from 55 to 8.5% error. The snout vent length (SVL) of these geckos was 61.5–91.7 mm and their body mass 4.8–11.5 g. We killed only six geckos in order to minimize destructive sampling as much as possible.
Table 1

Sex, snout-vent length (SVL), live wet body mass, DXA mass reading and fat mass chemical extraction of six Israeli fan-toed geckos Ptyodactylus guttatus used to validate the application of dual-energy X-ray absorptiometry (DXA) to non-invasively calculate body fat indices in small lizards

Sex	SVL (mm)	Live Wet Mass (g)	DXA Mass Reading (g)	Fat Extraction (g)	DXA Fat Reading (g)
Male	61.51	4.8	5.7	0.26	1.5
Male	87.45	11	12	0.37	1.65
Female	80.74	10.9	11.9	1.60	2.4
Female	65.97	11.5	13	2.09	2.8
Female	67.14	6.5	7.6	0.93	2
Unknowna	55.64	4.2	5.2	0.43	0.95

^a Indicates the lizard that was removed from the validation experiment (see text for details)

GECKO uses the flux balance analysis (FBA) approach (Orth et al, 2010). In FBA, the stoichiometry of the cell's metabolism is represented in a stoichiometric matrix. The columns of this matrix indicate the stoichiometry of reactions, and the rows indicate mass balances for each metabolite. By assuming pseudo‐steady‐state conditions (i.e., no accumulation), imposing constraints on fluxes and assuming a cellular objective function, an optimal solution for all metabolic fluxes can be found. In this study, we further constrained the solution subspace by limiting fluxes with enzyme levels. For any enzyme E_i that catalyzes a reaction R_j, it holds true that: $v_{\mathrm{j}}\le k_{\mathrm{cat}}^{\mathrm{ij}}·[E_{\mathrm{i}}]$ where v_j is the metabolic flux of R_j (mmol gDWh⁻¹), [E_i] is the intracellular concentration of E_i, and kcatij is the turnover number of E_i catalyzing R_j. With GECKO, we impose this constraint on each reaction in the model (Fig 1). The procedure also accounts for different relationships between enzymes and reactions, as described in the following:

If the reaction is reversible, two reactions are defined, one in the forward direction and one in the backward direction, both utilizing the same enzyme but possibly with specific k_cat values, depending on the enzyme's substrate affinity.

In the case of a reaction having isozymes, one reaction for each enzyme is specified in the model, with different k_cat values based on affinity, when available. Additionally, to keep the same original upper bound in the reaction's flux, an “arm reaction” is introduced to constrain the overall flux, creating a pseudo‐metabolite that acts as an intermediate between the substrates and the products, as previously introduced (Zhang et al, 2015).

If an enzyme is promiscuous, then the same enzyme will be used by all the respective reactions, possibly with different k_cat values (because of different substrates). This implies that there will be more than one non‐zero coefficient in some rows of the lower left submatrix of the stoichiometric matrix (Fig 1B). Additionally, because only one enzyme utilization constraint is defined, the reactions will share the amount of enzyme available.

Finally, in the case of a complex, the reaction will utilize all of the subunits belonging to it. This implies that there will be more than one non‐zero coefficient in some columns of the lower left submatrix of the stoichiometric matrix (Fig 1B). Additionally, each subunit's stoichiometric coefficient in the reaction will be multiplied by the corresponding subunit's stoichiometry in the complex.

With these formalisms, complicated constraints (Adadi et al, 2012) are circumvented, and we can directly overlay proteomic data in the form of an abundance vector on top of the enzymes’ usage in the model. It should be noted that enzymes are not consumed in our approach, but rather occupied; given that we are operating under the steady‐state assumption, for a fraction of a second, there is a limited amount of enzyme occupied by its substrates to catalyze the corresponding flux. Therefore, by imposing a mass constraint on the enzyme level, our framework prevents reactions from having higher fluxes than that allowed by the enzyme concentrations.
The developed framework is explained in further detail in the Appendix, where an example for a toy model is also supplied (Appendix Fig S1). GECKO is implemented in MATLAB, with a small section implemented in Python (for querying k_cat values from BRENDA). Both GECKO and ecYeast7 are compatible with the COBRA toolbox (Schellenberger et al, 2011) or any constraint‐based approach.

Approximately, 5 × 10⁷ unsorted and 2 × 10⁷ twice-sorted PIGS-HRD1-DKO cells were extracted for genomic DNA using a Wizard Genomic DNA Purification Kit (Promega). The gRNAs were amplified from genomic DNA of unsorted and twice-sorted cells. PCR (25 cycles) was performed to amplify the gRNAs using KOD FX Neo Polymerase (TOYOBO LIFE SCIENCE), making up a total of 65 tubes for unsorted cells and 12 tubes for twice-sorted cells (oligos for amplification of gRNAs are shown in Data S1). PCR products for unsorted cells (1,050 μl) and for twice-sorted cells (250 μl) were applied to a 2% agarose gel for purification. The PCR products were concentrated, mixed (unsorted to sorted 4.2:1), and analyzed by paired-end sequencing with a NovaSeq 6000 system (Illumina). Deep sequencing raw data were processed for gRNA counting using Python scripts. The high-throughput sequencing reads were demultiplexed using the 5-bp adapter by cutadapt version v1.18 (Martin, 2011 (link)). By using MAGeCK workflow version 0.5.8, the adapters of the demultiplexed reads were trimmed to obtain 20-bp gRNA sequences, and the single-guide RNA (sgRNA) sequences were mapped to the sequences of the Human GeCKO v2 sgRNA library to determine the total number of gRNA counts. The robust rank aggregation values and P values were determined using the MAGeCK algorithm (Li et al., 2014 (link)).

Genome-wide CRISPR–Cas9 knockout (GeCKO) v2.0 library plasmids were kindly provided by Feng Zhang [14 (link)]. The GeCKOv2.0 library A consists of 65,383 single guide RNAs (sgRNAs) that target 19,050 genes and 1864 miRNAs, causing frameshift indel mutations that lead to loss-of-function alleles. MCF7 cells were transduced with lentivirus carrying the GeCKOv2.0 library at an MOI of 0.2–0.4 for 24 h to achieve 100 × coverage of each sgRNA construct and then selected with puromycin (3 μg/mL) for 7 days. Puromycin-resistant cells were expanded for another 10 days to allow gene editing. Transwell invasion assays were performed as described, and invasive and noninvasive cells were harvested. Genomic DNA was isolated from the cells using a Blood & Cell Culture Midi kit (Qiagen). The sequences targeted by sgRNAs were amplified by the two-step PCR method, and the primer sequences for lentiCRISPR sgRNAs are listed in Additional file 1: Table S1 [14 (link), 15 (link)]. The PCR products were purified by agarose gel electrophoresis, quantified using a Qubit 3.0 Fluorometer (Invitrogen) and an ABI7900 real-time fluorescence quantitative PCR instrument (Applied Biosystems) and sequenced on a HiSeq 2500 instrument (Illumina) in single-end mode. The MAGeCK algorithm was used to analyze the FASTQ files and identify metastasis-related genes.

Taking inspiration from the toes of tree frogs and geckos, we designed an adhesive that comprises three key functional components: (i) a tacky surface layer that can conform to micro- to nanoscopic substrate roughness features (henceforth ‘adhesive surface’), analogously to the micropatterned surface of the adhesive pads of geckos and tree frogs; (ii) a fibre-reinforced base of the adhesive surface that is stiff in shear loading (henceforth ‘fibre-reinforcement’; we refer to the free end of the fibre-reinforcement as ‘tail’); and (iii) a soft ‘backing’ below the adhesive surface and fibre-reinforcement that enables conformation to macroscopic substrate roughness features. We refer to the combination of adhesive surface and fibre-reinforcement as ‘adhesive base layer’ (ABL), and to the entire system as ‘adhesive’.
To investigate the effect of the novel backing element on the adhesive’s friction performance on a non-flat substrate, we kept the ABL constant while varying the backing design and its mechanical properties. With the friction force $F_{\parallel }$ being the product of maximum shear stress ${\sigma }_{\parallel }$ and contact area A, and contact area depending on the adhesive’s compressibility, we expect that the adhesive’s friction scales inversely with its compressive stiffness on a non-flat substrate. Due to COVID-19 restrictions, all adhesive manufacturing and experimentation were conducted in a home-office setting using custom-made procedures and setups.