Team:Paris Bettencourt/Project/Assay

Project
- Overview
- Assay
- Microbiology
- Enzyme
- Binding
- Indigo
- Model
Practices
- About Perc
- Dry Cleaners
- Design
- Events
Safety
- Lab Safety
- Product
  Safety
Achievements
Notebook
Team

@iGEM_Paris

Goals

To construct a quick, cheap and fast way to stain and de-stain fabrics.
To quantify fabric stain levels

Results

Built a 96-well microplate compatible assay for studying fabrics.
Accurately quantified wine stains with custom image analysis software.

Methods

Fabric Selection
Laser cutting
Flat bed scanning
Computational image analysis (Eclipse-Java)

Skip to results

Overview of the project

Abstract

Standard biochemical activity assays are performed in well-mixed solutions. But enzymes are designed to work at the fabric surface, where real stains are found. We wanted to measure enzyme performance under realistic condtions. Our protocol uses laser-cut fabric circles sized to fit in standard 96-well microplates. In this way, the fabric could be stained, washed, enzyme-treated and incubated using small volume protocols compatible with common lab protocols.
To measure the stain, we used flatbed scanners coupled to custom-made image analysis software. This software is able to identify each circular well and quantify the average pixel color and intensity, which we show is directly proportional to the stain concentration. In this way, we are able to measure the capabilities of our enzymes in conditions that exactly match their real-world application.
Finding the appropriate bacterial strain/enzyme to digest anthocyanin, the compound that gives wine its colour, requires high-throughput screening of strains on fabric. Here we developed a microplate on fabric to test 96 different strains/enzymes for stain removal at once and a measurement technique to quantify the intensity of stain. The novel microplate on fabric is DIY and can be highly automatable and provides simple solution when compared to the traditional methods of using enzymes to test stain removal from fabric of different types. Our results demonstrate the simplicity of the design and its application. The model can be further used along with intensity measurement image processing tool for every fabric and any stain in general.

Motivation and Background

First challenge: miniature washing machines

Most of the stains on fabric cannot be removed by using a traditional detergent. Research has come up with chemicals, which are specific to particular stain removal. These chemicals are used in dry cleaning to remove the stains. On the other hand, Enzyme based detergents are becoming increasingly common nowadays. There is a dearth of a high throughput assay to screen different enzymes to test their potential activity on stain degradation. Our project wanted to find a synthetic biological alternative to PERC, a likely carcinogenic agent used in dry cleaning to remove wine stains, which demands a high throughput assay to test different bacterial strains/enzymes on different fabrics to find potential alternative.

A typical garment is composed of several square meters of fabric and a typical washing machine has a volme of 100 liters. While it is possible to perform controlled experiments at this scale, only a few such experiments can be run in parallel in a normal lab. We needed a millimeter-scale system that would allow us to perform hundreds of quantitative assays in parallel.

Experiments on t-shirt scale systems are also complicated by the tendency of liquids to absorb into hydrophilic fabrics like cotton. Small liquid volumes containing stains, detergents or enzymes are quickly wicked into the fabrics and spread over a large area via capillary action. This makes liquids difficult to recover and encourages cross-contamination between different regions of a fabric. Therefore, it was important that each fabric sample in our system be physically enclosed in a micro-well. This allows solutions to be added and removed to a samples under controlled conditions during the course of an experiment.

Designing a Microplate on fabric needs effective containment of fabric portions without any diffusion from a well from an other. Diffusion of contents (crosstalk) leads to discrepancy in scientific results. Capillarity phenomenon plays a predominant role in diffusion through fabric. The ideal assay should avoid the diffusion between fabric portions that are being tested. It should be simple, cheap, robust, producible at high rate and potentially automatable with high durability. The design should be applicable to test any stain on different types of fabrics. Complementing these qualities of an assay, the strain intensity should be quantified and compared to analyze the experiments effectively in a research environment. The development of such an assay facilitates research to find new enzymatic/bacterial alternatives.

Second challenge: measuring stain level

Many conventional biochemistry assays are done in well mixed liquid solutions, taking advantage of the uniform absorbance and fluorescence properties to perform quantitative measurements. This is not possible on fabric samples, which are uneven and opaque. We knew that the human eye was able to detect and quantify stains. Therefore, we turned to photographic image analysis software to quantify the color intensity of a 2-D image.

Figure 1 Our assay platform: Microplate with lasercut fabric. The fabric to be tested is precisely laser cut to fit the diameter of a standard microplate along with the glass cover. The fabric circles are removed from the glass and placed within the microplate wells. A microplate tip holder with 5 mm diameter holes is glued onto the microplate to hold the circles in place. Fabric circle separation prevents diffusion to adjacent samples. The plates at the top left and bottom right show a row of increasing sample concentration. The plate at the top right shows the tip holder underneath the glass cover. The picture at the bottom right shows a completed microplate from the bottom.

Results

First generation: Wax-based

Wax was 3D printed on cotton. There is no crosstalk between two adjacent wells. Although it proves to be efficient, the time required to make a 3D printed wax on cotton is long which makes it unreliable and unfit for high throughput production in industry or for research. But the technique of 3D printing wax on fabric can be useful to make definite patterns of wax and dye the areas without wax and then melting the wax in hot water (The traditional technique is called Batik). This can be a boost for customizable Batic 3D printing of wax on fabric.

Second generation: Microplate-based

A Do-It-Yourself and automatable Microplate on fabric is developed which can be used to test every type of stain as long as it can be scanned using a flatbed scanner on any fabric. An automated 96 well microplate for fabric was made using a fluorescent microplate, 96 well tip holder and laser-cut cotton circles.

Spread a flat, wet cotton sheet on glass substrate of 96 well plate dimension.
96 Wet cotton circles of 6.5mm are laser cut on glass substrate.
Remove the fabric. The circles of cotton stay on the glass substrate.
Place the glass substrate of cotton circles on fluorescent microplate such that all circles face the wells and align well.
Hold it tight and move the glass substrate laterally so that all the cotton circles fall into the wells.
Put the tip plate holder on the top of the microplate and adhere it to the microplate using a glue.
Cover it with a lid and the microplate is ready to use for experiments. The same can be done for other fabrics like silk, polystyrene, nylon etc.

Open source image processing plugin for quantification

An open source image processing plugin to measure the intensity of stain in 96 wells Microplate with fabric which is used to quantitatively measure the degradation of stain on fabric by a bacterial strain/enzyme. Usage:

Add the Plugin by Team Franknstain to ImageJ, enable the plugin
Load the Image in ImageJ, Image must be in TIFF format and 12x8 well format as shown in the figures
Go to Image -> Color -> Split Channels, close Green and Blue channel images.
Keep the Red image. Go to Analyze -> Set scale -> Remove the scale
Choose Point tool, click on four different center points of extreme wells in the Image followed by pressing 'T' after choosing each point.
Click on the Result table Tab.
You get a pop up table of mean value of each well with an index for example '3D' where 3 represents the row and D represents the column.

Figure 2 Measured relative pixel intensity of stains using Image Processing plugin. We developed an open source Image Processing plugin using the ImageJ software and developed in Eclipse-Java in order to quantify stain intensity by image analysis of scanned microplates. The graph shows the relative intensity of the stain based on the fabric samples shown below. Fabric circles increase in sample concentration from left to right, which is recapitulated in the graph.

Methods

Fabric selection:

Selection of fabric plays a crucial role in the experiments. Single type of fabric is used to conduct experiments for example pure cotton, pure silk etc. Chemically treated fabrics can potentially reduce the activity of enzymes or react with the enzymes or might not support the survival of bacteria affecting their growth rate on the fabric. Most of the fabrics are treated with hydrophobic materials to repel water in industries. Chemically untreated and unbleached fabrics need to be identified and used in making microplate on Fabric. Plain color fabric is used in order to quantify the intensity without any noise.

Laser-cutting fabric on a glass plate:

Once the procurement of appropriate fabric is done, it is cut into the dimensions of the standard Microplate(11.5cmX8.5cm). The fabric is placed on the glass plate and it is wet with pure water completely. The Laser-cutter is given an input using a Computer Aided Design (CAD) model which has 96 circles of diameter 6.5mm which are precisely in the positions of wells of a standard Microplate. The glass plate along with fabric is subjected to laser cutting to cut 96 circles. The cutting speed and input power for Laser-cut need to be optimized a priori for a specific fabric. The fabric is then removed from one end of the glass plate. It is observed that 96 circles of the fabric remain adhered to the glass plate where the remaining cloth comes out clean.

Transfer of fabric to a Microplate:

The cotton circles on the glass plate can be safely transferred to the Microplate in two ways

By inverting the glass plate and aligning the circles of fabric to the wells and applying uniform pressure in vertical direction and pushing the glass plate horizontally. Since fabric has appreciable thickness, the circles of fabric fall down the wells.
By drying up the fabric circles on the glass plate using a heater which should not be convective, Convective heaters force air drying which make the fabric circles move out of their positions, and inverting the microplate on the glass plate with fabric circles with proper alignment, holding them together and inverting them at once would make the circles of fabric fall down.

Trapping Fabric:

Once the fabric is placed inside the wells, a microplate tip holder which has circles of 5mm diameter is glued on the top of Microplate to entrap the fabric circles.

Measurement using imageJ Plugin:

An imageJ, open source software for image processing, plugin is developed in Eclipse-Java to quantify the stain intensity from images taken using a flatbed scanner. The bottom surface of the Microplate is scanned using a flatbed scanner before and after the introduction of bacterial strains/enzymes. TIFF format of images are highly recommended format to preserve the quality of the intensity. The Images are then fed into the imageJ plugin ‘ij.franknstain’ to quantify the difference in the intensity of strain before and after the experiment. A decrease in intensity suggests that the strain is being degraded which is a positive control. A dilution test of wine is done to quantify the intensity measurements at different stages of dilutions to assess the efficiency of our image processing tool.

The image processing plugin takes a scanned image of 96 Well Microplate of Fabric. It identifies positions of 96 wells when provided with four extreme points. It then goes to encircle the wells precisely and takes red channel into account for calculating mean value of the pixel in the well. It then gives a result table which has a specific mean pixel value assigned to each well. Mean pixel value of the well is considered to be the measure of stain intensity.

Our failed attempts

Before achieving to develop a simple and efficient assay platform we had some failed attempts:

Figure 3 Wax applied on cellulose paper

Wax on paper:

Impressions of wax on the paper where the wax penetrates vertically into the paper and creating specific boundaries was observed.
The wells are not uniform given the impression and dipping the template in wax were done manually.
The Anthocyanin did not spread when tested and it was contained in the wells.

Figure 4 Wax block with 96 wells made by using CNC

Wax impression on cotton:

96 wells were created on a block of wax we've found in the Openlab of CRI-Paris using CNC (Computer Numeric Control)
The diffusion of wax on chemically untreated cotton is very fast even though cotton is hydrophilic this time and wax is hydrophobic when made an impression

Figure 5 3D printed two component mold

Two component mold:

Wax is poured into the two-component mold so that it would create circles of wax on cotton when the cotton is placed beneath the mold.
The base of the pillars against which we press the fabric are not uniform.
We cannot get uniform circles and avoid crosstalk when we pour wax through the mold.

Figure 6 Two-layer PDMS assay platform

Using PDMS to create our own 96-well plate:

96-well plate was made where we would sandwich cotton fabric between two layers of PDMS with wells.
The punching went well but not the spacing between the wells. Manual punching resulted in non uniform spacing of wells.
Although there is a strong bond between cotton and PDMS when the cotton was introduced in between the PDMS layers where uncured PDMS was applied to one of the faces, the PDMS got diffused into the wells which hinders the use of fabric in the microplate

Figure 7 3D printed mold to pour PDMS on

3D printed mould for PDMS:

A mold with pillars to pour PDMS on it and have a layer of PDMS with wells was 3D printed. A 96 well plate with PDMS was created using the mold.
Cotton was laser cut to minimize the area available for diffusion. Multiple layers of cotton at the same time was laser cut.
The laser-cut cotton when aligned with PDMS wells was bonded to glass using plasma cleaning, The gap (thickness of the cotton) is sufficient enough for crosstalk leading to diffusion.

Figure 8 3D printed wax assay on fabric

3D printed wax on cotton:

Wax was 3D printed on cotton
There is no crosstalk
The time required to make a 3D printed wax on cotton is long which makes it unreliable.

Figure 9 Crossroads design assay platform

Crossroads design:

3D printed mold was created in such a way to easily insert the laser cut cotton directly without any misalignment and diffusion.
No misalignment was observed between the cotton and wells of PDMS.
Diffusion through the grooves can be avoided, if the connection between the wells are filled using pdms or hydrophobic material which might lead to bumps that gives improper binding.

Figure 10 PDMS microplate glued on glass

PDMS Microplate:

Diffusion of cotton cannot be avoided if it is continuous, so cotton is cut into pieces.
PDMS should then be bound to glass using plasma cleaning trapping the cotton in the wells.
We could not bind the PDMS and the glass using plasma cleaning as the surface of the PDMS is not flat (molds made out of 3D printing have non smooth surfaces) which makes the process inefficient.
PDMS was bound to glass using a glue. There was leak at the borders due to manual gluing and because of rough surface of PDMS.

Attributions

Allison Bricknell George and Mani Sai Suryateja Jammalamadaka worked on assay development and its measurement tool.
We are grateful to Edwin ‘Jake’ Wintermute for his valuable suggestions throughout the work.
We are in great debt to the Openlab of CRI and it’s members, Kevin Lhoste and Daniel L'orfèvre, Vladimir Hermand for giving us guidance and access to 3D printers, CNC and Laser cutting machines which were quite useful in reaching successful fabrication of Microplate on Fabric and for their guidance.
We thank Song Xiaohu, Evolutionary Systems Biology Group (INSERM U1001) for his continuous monitoring for the image processing tool development.
We acknowledge Dr. Ana Santos, Evolutionary systems Biology Group (INSERM U1001), for helping us while we were experimenting with our initial designs using PDMS. Omission of credits is deeply regretted.
We would like to thank Arthur Dalaise, Founder of La Biche-Renard for helping us with 3D printing issues.

References

Cotton-based Diagnostic Devices, Nature Scientific Reports, doi:10.1038/srep06976
ImageJ User Guide, IJ 1.46r, Tiago Ferreira and Wayne Rasband