*2.14. Data Analysis*

All the replicated data for the assays and comfort analyses (Figures 3–5 and Figure S5) were plotted in ORIGINPRO® 2018 using a box plot format, a graphical format that summarizes the key statistical values. The solid brown dot in the box plot was the raw data. The height of the box represents the 25th and 75th percentiles. The whispers represent the 5th and 95th percentiles. Additional values included the median (line inside of the box) and mean (white dot) presented in the box plot. We used the mean value of each data set for our analyses.

We used one-sample Student's *t*-tests to investigate the significance between two data sets in Figures 3I,J and 5B,C The mean value of the first data set was used as the theoretical expectation. The second data set was set as the true mean. Differences in mean values were found to be statistically significant when the *p* values were greater than 0.05 (\*) or 0.01 (\*\*).

**Figure 5.** Prototype garment's comfort and bite-resistance compared to commercially available similar garments. (**A**) Manikins equipped with various garments (I, Under Armour® base layer; II, NC State base layer; III, winter army combat shirt; and IV, NC State shirt), also showing the average heat-loss maps. (**B**) Garment insulation. Since mosquitoes mostly appear in warm weather, a garment with low insulation properties is preferred. The NCSU base layer and NCSU shirt provided lower levels of insulation compared with the comparative garments tested (*p* < 0.05 and *p* < 0.01, respectively). (**C**) The *D*predicted values (predicted heat loss; Table S1) for the garments tested. II the NCSU base layer showed an equivalent thermal and moisture management compared with I. IV, the NCSU shirt exhibited better thermal and moisture management compared to III (*p* < 0.01). (**D**) Walk-in-cage bioassay with 10 min standing and 10 min sitting. The container in the hands of the subject (bottom picture) housed the mosquitoes. The mosquitoes were typically released, and the test started with the person standing (note the empty container on the stool, top picture). (**E**) Walk-in-cage bioassay results for the worst-case replicate shown (\* = one mosquito bite). Bites on the shoulder were observed where the most stretching of the garment occurred and bite resistance was reduced. A specially designed double layer was used in this part of the NCSU base layer which eliminated all bites in the walk-in-cage bioassay (data not shown).

All tested materials and garments are listed in Table 1, including information on the material type, name, abbreviation, thickness, pore diameter, model prediction, and bioassay validation. Values of thicknesses and pore diameters are the mean values calculated from the multiple measurements discussed in the section "Textile structure analysis". Model prediction is the predicted bite resistance. "Safe" represents a fabric that is predicted to have 100% bite protection predicted by the bite-resistance model and "unsafe" means the fabric is predicted to allow at least 1 mosquito bite. Bioassay results are actual measurements of bite resistance. "Pass" indicates the fabric was at least 95% bite resistant by the in vitro or in vivo bioassay. "Fail" indicates a fabric provided less than 95% bite protection.

#### **3. Results and Discussions**

#### *3.1. Mosquito-Bite-Resistant Textile Model*

Figure 1A shows an adult female *Ae. aegypti* probing human skin. Figure 1B is a scanning electron microscopy picture (SEM) of a knitted textile. The yarns used to make the textile consisted of a multitude of filament fibers knitted in an intermeshed loop configuration. In a knitted fabric, the spaces between the filaments form pores (Figure 1C) and together with its thickness determine a fabric's bite resistance to mosquitoes and its comfort to people. Pore diameter and fabric thickness are critical limiting factors for mosquito proboscis penetration of the skin that also affect the thermophysiological comfort of a textile (Figure 1D). Increasing pore diameter improves fabric breathability and comfort but increases the transmission of skin odorants, increasing mosquito landings and biting. Fabrics containing small pores are less attractive to mosquitoes and more bite resistant but have reduced comfort because of reduced air flow. Increasing fabric thickness improves bite resistance but reduces comfort by increasing thermal insulation. A model to predict bite resistance was developed that informed fabric thickness and pore diameter as they related to the morphometrics of the mosquito's head, antennae, and proboscis, and the mechanism that mosquitoes use for finding and biting through a textile. The three cases considered are illustrated in Figure 1E. Figure 1F describes our overall strategy for developing bite-resistant garments: (i) developing a predictive model based on mosquito head morphometrics; (ii) model validation using mosquito *in vitro* testing of woven filter fabrics, plastic spacers, and 3D spacer fabrics for bite resistance; (iii) development of knitted fabrics for garmen<sup>t</sup> construction using the model; (iv) *in vivo* (arm in cage) mosquito testing for bite resistance of these fabrics; (v) garmen<sup>t</sup> construction; and (vi) garmen<sup>t</sup> walk-in-cage testing for bite resistance; and (vii) manikin comfort tests of the garments.

Figure S1A shows the size of the proboscis where the stylets of the proboscis interlock forming a feeding tube covered by the labium (Figure S1B). Figure S1C shows the stylets, and Figure S1D is an SEM of the mosquito's proboscis composed of the labrum, maxillae, mandibles, and hypopharynx. The mechanical process of probing skin was described previously [31,32]. The labrum's diameter was measured in our work as a key parameter for our bite-resistance model. Preventing labrum contact with the skin prevents blood feeding. Figure 2A–D provide a detailed description of Cases 1–3. In Case 1, the pore diameter of the fabric barrier is smaller than the diameter of the labrum (Figure 2B). In Case 2, the pore size of the fabric barrier is larger than the labrum diameter but smaller than the diameter of the mosquito head (Figure 2C). Thus, fabrics with the proper thickness can prevent the labrum tip from contacting skin. In Case 3, the fabric pore size is larger than the head diameter but is smaller than the size of the head plus antennae (Figure 2D). The ice-green vertical bars are the textile barrier, and the red dotted line the critical combination of pore diameter and thickness of the textile barrier.

The critical geometrical relationships of pore diameter and thickness for each case to prevent blood feeding were defined as follows:

Case 1:

$$t = \frac{\mathbf{x}}{2 \times \tan\left(\frac{\mathbf{a}}{2}\right)}, \text{ when } 0 \le \mathbf{x} < D \tag{2}$$

$$\frac{D}{2 \times \tan\left(\frac{\theta}{2}\right)} \le t \le L\_{\text{probasics}} \text{ when } \mathbf{x} = D \tag{3}$$

Case 2:

$$\text{If } t = L\_{\text{proboci}} + \frac{D\_{\text{head}}}{2} \left\{ 1 - \cos \left[ \arcsin \left( \frac{\mathbf{x}}{D\_{\text{head}}} \right) \right] \right\}, \text{ when } D < \mathbf{x} \le D\_{\text{head}} \tag{4}$$

Case 3:

$$\text{Let } t = L\_{\text{proboics}} + \frac{D\_{\text{hand}}}{2} + \tan(\beta - 90) \times (\mathbf{x} - D\_{\text{head}}), \text{ when } D\_{\text{hand}} < \mathbf{x} \le L\_{\text{antruma}} \tag{5}$$

where *t* and *x* are the thickness and pore diameter of the mechanical barrier, respectively; *<sup>L</sup>*proboscis is the maximum proboscis length; *D* is the maximum diameter of the proboscis tip; *α* is the angle of insertion of the proboscis tip; and *β* is the angle between the antenna and proboscis.

The red dotted lines in Figure 2B–D show the limit between a textile being predicted as unsafe (biting is possible) and safe (biting cannot occur) for Cases 1–3 (for critical combinations of pore sizes and thicknesses as specified by the model). For the model to be feasible, we made the following assumptions: (1) the fabric barrier and proboscis tip were not deformable; and (2) only thickness and pore diameter were considered as structural parameters for the fabric barrier. Figure 2F,G show the correlation between the bite-resistance performance predicted by the model and fabric pore size and thickness, in which the abbreviations of all dimensional values are described in Figure 2E. In Figure 2F,G, the brown dotted lines mark the dimensions of the key factors of the mosquito anatomy, including the head diameter, labrum and its tip length, and diameter and antenna angle from the head and length. The red solid lines are the critical combinations of the fabric pore diameter and thicknesses relative to the mosquito morphometrics that would produce a safe (100% bite resistance shown in green) or unsafe (pink) fabric as predicted by the model.

#### *3.2. Mosquito Morphometrics Used to Predict Safe Fabrics*

The head diameter (*D*head), antenna length (*L*antenna), proboscis length (*L*proboscis), maximum labrum diameter (*D*), labrum tip length (*L*tip) and the tip angle (α) of *Ae. aegypti* adult females are shown in Figure 3A. Each body part was measured from twenty insects. The average values were input into our model to define the fabric thickness and pore diameter and the limit between safe and not safe (Figure 3B). We focused on these limits and produced a variety of barriers of different pore sizes and thicknesses for the experiments (Figure S5A–F) to test the model using our *in vitro* bioassay (Figure S2A). In some cases, these barriers (description follows) were not practical for garmen<sup>t</sup> construction but were used because they were optimum for model validation, as explained in the Materials and Methods.

For Case 1, single-filament (woven) filter fabrics (shown in Figure 1Fii and Figure S2B) with different pore sizes and a fixed thickness (Figure S5A,B) were tested using the *in vitro* mosquito-bite-resistance bioassay (Figure 1Fii and Figure S2A). These are technically fabrics, but they are highly resistant to stretch, uncomfortable to wear, and too costly for garmen<sup>t</sup> construction. However, they were used for model validation because they were available in precise, different pore diameters and fabric thickness. Highly precision-machined, polypropylene plastic plates (Figure 1Fii and Figure S2C) were used with different pore sizes and thicknesses (Figure S5C,D) to evaluate the model for Cases 2 and 3 using the *in vitro* bioassay. Then, two knit fabrics for Case 1 and two knitted spacer fabrics (shown in Figure 1Fii) for Cases 2 and 3 each with different pore diameters and fabric thickness (Figure S5E,F) were constructed to inform further on Cases 1–3, to better approximate a practical garmen<sup>t</sup> application than filter fabrics and plastic plates.

The number of landings and percentage blood feeding for the barriers tested are shown in Figure 4 for our model validation research. Table 1 (group = materials for model validation) relates thickness and pore diameter to the model prediction and whether the barrier failed or passed in preventing mosquito blood feeding. In these experiments, a percentage of blood feeding greater than 5% (bite resistance was lower than 95%) was considered a failure for the barrier in preventing blood feeding. In Figure 3B, the left and right graphs relate the pore size and thickness for the filter fabrics and plastic plates, respectively, with the model prediction of what would be safe and unsafe. Only one (plastic plate S7, Table 1) out of the 18 barrier materials tested (filter fabrics, plastic plates and knit fabrics) failed to provide bite protection when the model informed the barrier should be safe. This failure in the model corresponds to the red dot in the green area in Figure 3B, the right graph. Those barriers (green color dots) located in the safe area exhibited bite resistance against mosquitoes of at least 95%, as the model predicted for the filter fabrics and plastic plates. The model was 100% accurate in predicting safe and unsafe for both the knit and knitted spacer fabrics (Table 1, T1–T6).

These results sugges<sup>t</sup> that the model we developed was reliable for predicting mosquitobite resistance against the lab-reared mosquito, *Ae. aegypti*, and was 100% reliable in our studies of the knits and spacer knits tested. Additional testing will be needed in the future, to determine if our model translates to other mosquito species and to mosquitoes in the field. Regarding for the economy of time and resources, we argue concentrating on one species was a reasonable approach for our studies and proof of concept.

#### *3.3. Finite Element Analysis*

In our validation studies, a barrier was considered safe when bite resistance was 95% or higher. When pore sizes and thickness approached the limit between safe and unsafe (Figure 3B left graph for filter fabrics and right graph for plastic plates), some blood feeding occurred at a low percentage, 5% or less (Figure 4A–D). This was also the case for the knits tested (Figure 4E,F). There are two possible reasons. First, the labrum diameter of some mosquitoes may have been smaller than the average value (27.5 μm) used in the model, allowing some mosquitoes to penetrate the barriers. Second, the barrier may have deformed under the pressure of the proboscis and enlarged the pores causing failures. In the latter case, this would not be an issue with the plastic plates but could be a factor for the textiles tested.

To investigate the interaction between proboscis and textile structure, the elastic modulus and geometry of the labrum were measured to establish a finite element labrum model. Figure 3C shows the anatomy of the proboscis tip. Figure 3D is the nanoindentation curve for the labrum, which was used to obtain the elastic modulus for the property parameters needed for the model. The woven (filter) fabric used in our validation studies (Section 3.2), W1 to W4 (Table 1), were modeled to better understand how the labrum might deform textiles in general. Figure 3E illustrates the four patterns. Figure 3F shows one example of the penetration model for the labrum on the W2 woven fabric, and Figure 3G shows the time course of penetration. For W1, the labrum interaction with the textile is less since the labrum can easily go through the fabric. However, W3 and W4 in Figure 3E are more dense structures with the pore size below that of the labrum diameter, not allowing free labrum penetration through the pore. Therefore, W2 with a pore diameter of 18 um was selected to show fabric deformation subjected to labrum penetration. It was found in our research that the labrum can move the filament yarn and push through the W2 filter fabric over time (Figure 3G) for a blood meal. This is the reason that W2 located near the boundary line failed in resisting some mosquito bites. Figure 3L shows the change curves for the pore diameters of each woven structure. After labrum penetration, W1 and W2 were enlarged more than the labrum diameter and therefore would fail in preventing blood feeding because the structures were deformed. Although pores on W3 and W4 demonstrated deformation, the pore diameter was still below the labrum diameter, which enabled the structure to prevent blood feeding. In summary, in addition to the importance of pore size and thickness, the finite element analysis informs that micromechanical deformation of the fabric in response to the pressure exerted by the proboscis pushing-through the fabric can affect blood-feeding success. Yarn chemistry and methods of weaving and knitting will

impact deformation and, therefore, bite resistance. It would also be expected that variation in labrum diameter in the mosquito population will have an impact.

#### *3.4. Development of Fabrics for Garment Construction*

Once the model was validated for Cases 1–3, textiles were developed for the construction of a garmen<sup>t</sup> for final proof of concept that non-insecticide clothing could be bite resistant to mosquitoes and also comfortable. For these studies, bite resistance was measured with arm-in-cage bioassays (Figure S3A) with a textile considered safe if the bite resistance was 95% or higher. For Case 1, the knitted fabrics were H and B (Table 1) and shown in Figure S3B,C, respectively, and in Figure 1Fiii. For Case 2, the knitted spacer fabric was S (Table 1) and shown in Figure S3D, front and back, and Figure 1Fiii. Thickness and pore diameters are shown in Figure S5G,H, respectively, and the model prediction and bioassay results are in Table 1. The model was correct in all cases (see group = fabrics used in garments) in successfully predicting bite resistance. Accordingly, these textiles were used for garmen<sup>t</sup> construction.

#### *3.5. Bite Resistance of an Insecticide-Treated versus Non-Insecticidal Textile*

Permethrin-treated textiles are a widely used technology to prevent mosquitoes from biting people. Permethrin exhibits mosquito contact toxicity but also spatial repellency. Figure 3I shows the number of landings on fabric P (a permethrin-treated commercial fabric; detail on pore size and thickness in Table 1), which was lower (*p* < 0.01) than that for fabric H, the non-insecticidal superfine knit. Fabric P demonstrated spatial repellency presumably because of permethrin in the cloth whereas fabric H did not. Fabric H had a higher number of landings because mosquitoes were not repelled and landed on the fabric repeatedly in attempts to find a suitable location to penetrate the fabric. High landings without bites indicated the fabric structure has breathability but with pores sufficiently small for high bite resistance. Figure 3J shows that the percentage of blood-fed mosquitoes in the arm-in-cage studies for fabric P was three times higher than fabric H (*p* < 0.05). Although fewer mosquitoes landed on fabric P, a larger percentage of the mosquitoes that landed were able to penetrate the fabric and obtain a blood meal. In contrast, fabric H with smaller pore diameters and no insecticides resisted mosquito bites at a higher level.

These studies demonstrated that high bite resistance across a textile can be achieved that far succeed one commercial permethrin-treated fabric under high biting pressures in an arm-in-cage bioassay. Higher landings with no spatial repellency on the insecticidefree cloth would be expected to reduce biting on uncovered skin, especially when the proportion of uncovered to covered skin is small; in this case, the mosquitoes are probing the cloth and not being pushed to unprotected skin. However, more detailed studies are needed to address how an insecticide-treated textile versus a non-insecticide-treated textile, such as fabric H, would protect uncovered areas of the body.

#### *3.6. Comfort and Bioassay Evaluation of Prototype Garments*

The final step in demonstrating the proof of concept that insecticide free textiles can be used to protect humans from mosquito blood feeding and at the same time be comfortable, was to construct garments with the knits that our model predicted would be safe (fabrics H, B, and S, Table 1). These fabrics were used to construct a protective undergarment (a base layer garment; NCSU base layer, Table 1, and shown in Figure S6A and Figure 1Fv) and shirt (NCSU combat shirt, Table 1 and shown in Figure S6B and Figure 1Fv). These garments were tested in walk-in-cage bioassays to evaluate the mosquito-bite resistance where the threshold for success was no bites. A sweating manikin test was conducted to create whole-body heat loss maps for fabrics in different body zones to understand the heat and moisture resistance properties of our mosquito-bite resistant garments compared to commercially available garments.

Garments were tested for heat loss using a sweating manikin (Figure 5A). The garments included I, an Under Armour base layer; II, the NC State base layer developed using our model; III, a US army-issued combat shirt (provided by the US DOD); and IV, the NC.

State-developed, next-generation combat shirt, using our model. The same style of garments had similar heat-loss maps (Figure 5A), which indicated equivalent levels of thermal management. In the maps for garments III and IV, the blue color of IV is darker than III due to an innovative design that incorporated a 3D spacer fabric (Figures S3D and S6B and Figure 1Fiii) predicted to be bite resistant by our model but with open pores into the chest and arms area for heat managemen<sup>t</sup> (Figure 5A).

The insulation values for both of our developed garments (Figure 5B) were smaller than their counterparts of the same style. This finding indicated that the NC State base layer and the NC State combat shirt had favorable thermal exchange as well as minimal heat accumulation, making the garments more comfortable to wear in warm weather. The Predicted Heat Loss Potential ( *Q*predicted, W/m2) is a projection of the total amount of heat that could be transferred from the manikin to the ambient environment for a given condition, which was calculated using thermal and evaporative resistance values (see details in Table S1). In this case, the *Q*predicted of garments II (NC State base layer) and IV (NC State combat shirt) exhibited higher values than their counterparts (Figure 5C), which indicated they possessed superior comfort performance in both thermal and moisture management.

The NC State base layer and the NC State combat shirt were tested in walk in cage bioassays under heavy mosquito biting pressure with the human subject standing and sitting for 10 min in each posture (Figure 5D,E). The NC State combat shirt provided 100% protection against mosquito bites. However, the human subject wearing the NC State base layer received bites on the back and shoulders and the level of overall average protection was 96.5% (7 bites per 200 mosquitoes). When the base layer is used as an undergarment under a uniform, protection would be 100% (data not shown). This result on biting in the test reported was attributed to deformation of the knitted fabric on the shoulders where the fabric stretched, increasing the pore diameter of the fabric. We measured the fabric length during standing and sitting. Fabric H was estimated to have a 9.47% increase in stretch from the standing to the sitting postures. We conducted a virtual tensile experiment using an FEA model to investigate the change in the pore diameter of fabric H (see details in Figure S7). The tensile behavior of the fabric showed a directionality of stretch in which the wale direction exhibited a smaller deformation compared with the course direction, as shown in Figure S7C. The pore diameter also exhibited directional deformations in the course and wale directions, as shown in Figure S7D,E. In order to improve the bite resistance, a double layer of fabric H was stacked on the shoulder area (yoke), which partially covered the back of the human subject. The stacked orientation for both layers were perpendicularly aligned with each other, which reduced the fabric deformation during sitting and movement; this treatment also misaligned the pores of both fabrics. This improved the garment's bite resistance and provided 100% bite protection in walk-in-cage bioassays. Our two final garments listed in Table 1 were 100% bite proof in walk-in-cage tests. Notably, when the base layer was used as an undergarment under a uniform, protection was 100% (data not shown). In summary, preventing human–vector contact is an effective way to protect people from mosquito bites as well as to eliminate the threat of mosquito-borne diseases. We developed a mathematical model to predict the bite resistance of non-insecticidal textile barriers. Our model was verified through in vitro bioassays, using woven fabrics, plastic spacer plates, and knitted and knitted spacer fabrics, which showed that the model could accurately predict the bite resistance of mechanical barriers. The model was then used to develop comfortable and wearable textiles for garments. When compared with a permethrin-treated fabric, our fabrics development for garments had a higher bite resistance with a predicted higher level of protection for exposed skin; however, the latter needs further study. Then, the prototype garments were constructed with these textiles. These garments exhibited superior comfort performance compared to similar commercial garments and 100% mosquito-bite resistance. Use of our

model in the future will facilitate development of other, highly effective and comfortable bite-resistant fabrics solely based on textile structure without the need for an insecticidal treatment to prevent mosquito biting, and thus can be used to produce mosquito-bite-proof clothing for everyday use.
