Maven BiotechnologiesMiniaturized, Multiplex, Point-of-Care Solutions
July 2018OMNIBUS
ANA
Maven Confidential
Maven Assay CompilationModel Systems for Development and Diagnostics
July, 2018
Contact: William Rassman, M.D, Managing Partner, Maven Biotechnologies, LLC11620 Wilshire Blvd #280, Los Angeles, CA 90025 (310) 505-5393, http://mavenbiotech.com/
Table of Contents• LFIRE Detection Principle
• Maven Analytical Units
• Model Systems Tested
• Kinetics Model Systems
• Ag-Ab, Drug-Target, Drug Detection, Enzymatic Amplification
• Analytical Sensitivity vs Clinical
• Clinical Analyte Ranges
• Competitive Platforms’ Analytical Sensitivity and Multiplexing,
• Diagnostics Proof of Concepts• Autoimmune, Infectious Disease, Biomarker
� Flu testing directly from drop of blood
� Note: The table of contents is mostly self-explanatory, but this should prepare you for the flow of what follows. We’ll start with how it works at the physics level. The notes are in a small font at the bottom of each slide
Detection Principle: “LFIRE"Label Free Internal Reflection Ellipsometry
Real-Time Polarimetric Imaging
IrpI
IrsI
IrsI
Irp
Iδ
or
φCritical Angle
EvanescentField (~150 nm)
. ...
Output:An Image of Change
Input: Light
Maven Confidential
Event: Any binding at or near the surface.
Note: LFIRE uses a beam of polarized light, which enters the underside of a piece of glass at a particular “critical” angle. The light beam reflects from the glass-water interface, producing an electromagnetic wave that responds to refractive index in the water. The reflected beam’s polarization changes if the refractive index changes at the interface, in real-time. Thus anything that contacts the interface can be detected, from a small molecule, to a protein, and even whole cells. The hardware is simply a inexpensive LED, some lenses, some polarizers, and a CCD camera. No lasers, scanners, monochromators, photon multiplying tubes, etc. The surface is just silicon dioxide, aka glass, not precious metals or expensively nanofabricated structures.
Measurements Resolved in X,Y, Z, and Time
Protein Arrays Bacterial Motility(among spots on a protein array)
Wound Healing(pharma cell culture assay)
Maven Confidential
Imaging: 5mm FOV @1.5x zoom, ~7 μm resolutionZ Sensitivity: Sub-Picometer, >3000 thickness units/nanometer
“RU” or pg/mm2 sensitivityRange of thickness +/- sub-monolayer to micronsng/ml Label-Free kinetic immunoassay, pg/ml μELISA
Speed: 35ms to 2 seconds depending on sensitivity needs
As mentioned before, we use a camera to capture the reflected beam in real-time, making this a measurement that has spatial (xy), thickness (z), and time components, truly 4-D! The three videos here demonstrate the breadth of applications, from proteins binding other proteins to detect human disease, to microbes moving on a surface, to live tissue culture healing a wound.The performance metrics are listed; no technology combines the resolution, sensitivity, and speed like LFIRE; yet LFIRE is simple and robust by comparison.
Maven Analytical Units� The primary units of ellipsometry are phase shift, or Δ, expressed as degrees or radians, and amplitude, or Ψ.
Maven mostly uses images of in-situ phase shift, +/- 90 degrees.
� The camera used for measurement has a 12-bit sensor, and we integrate (average) to obtain an effective 16-bit dynamic range, and we handle it in raw unsigned 32-bit.
� We take several images per measurement, at different polarizations, and combine them with a image-based signal processing algorithm that produces proportional units between -1 and 1. These units are variously referred to as “y” or “w” internally at Maven. REUs (used in 2007 and 2008) are Relative Ellipsometry Units, protein equivalent nanometers.
� We scale this range of +/- 1 over -32000 to +32000 pixel units, in a 16-bit .tiff image, so that increases and decreases of thickness may be observed without clipping.
� We have AFM-calibrated thickness standards that produce 2400 pixels units of change per nanometer of a material that has a refractive index matching that of solid protein.
� The time domain standard deviation of noise is ~3 pixels, thus we can see a picometer of change over an averaged area, a pg/mm2, less than a monolayer of small molecules.
� A typical layer of antibodies is ~4 nanometers thick, or about 10,000 pixel units, thus >3.5 logs of detection is possible in immunoassays, not considering multi-layer assays and amplification with “optical mass” such as precipitates, large molecules, beads
� Note: This is a very dense slide… because it steps through the derived units and conventions we use in LFIRE, from the physics to the CCD to final data storage in .tiff image files. It further explains how we have calibrated our measurements to actual physical objects, and how they compare to some common biological molecules.
Model Systems TestedImmunoassays (Our initial clinical target market)� Various antigen-antibody and kinetic analyses� HCG (Human Chorionic Gonadotropin) � Gentamicin� Species-specific and isotype-specific industrial secondary antibodies � Phosphorylation antibodies to detect pSerine, pTyrosine, pThreonine� Protein A, G, and Chimeric A/G binding to immunoglobulinsAutoimmune Antigens� Myelin Basic Protein – peptide epitope 85-99 for Murine Autoimmune Encephalitis / MS� Proteolipid Protein – peptide epitope 139-151� JO-1 antigen for Polymyositis� Histone H1 for Lupus� dsDNA antigen for Lupus� Smith antigen for Lupus� RNP antigen for Mixed Connective Tissue Disease / Lupus� SCL-70 antigen for Progressive Systemic Sclerosis / Lupus� SSA (Ro) antigen for Sjogren's Syndrome / Lupus� SSB (La) antigen for Sjogren's Syndrome / Lupus� Rheumatoid Infectious Disease Antigens� Influenza – Flu A H1 and H3, Flu B Brisbane, Phuket recombinant & 2016 FluVirin (analytical study in
progress); New Caledonia 20/99, Wyoming 03/2003, Jiangsu 10/2003 as 2005 FluVirin.� Epstein Barr – Early Antigen, Nuclear Antigen� ToRCH – Toxoplasma, Rubella, CMV, HSV1, HSV2
Note: For more than a decade, we have been applying LFIRE to observe phenomena from pure chemical and physical, to very complex living systems. This slide and the following simply list all of these projects.
Model Systems TestedVarious� DNA hybridization of oligonucleotides, 23 base pairs � Transcription factor (p53) binding to double-stranded DNA � Anti-p53 antibody� Blood typing, binding of red blood cells to a microarray by surface antigens� Polyanion/polycation layer-by-layer (LBL) deposition – PSS-PDADMAC� Lectins (ConA, WGA, SNA) binding to a microarray of 128 different carbohydrates � Streptavidin binding to Biotin and vice-versa� Hydrogel-based humidity sensing� Physical thickness standard, calibrated� Drug (vancomycin) binding to target peptide (Acetyl-Lys-D-Ala-D-Ala)� Blocking material studies� Enzymatic Amplification of Signal � Nanoparticle Ampification of Signal� Lectin Arrays for Glycoprotein Capture, Glycoprotein Arrays for Characterization with Lectins� Click Chemistry Capture from lysate� Phosphatase Activity AssayCell-Based � Extracellular Matrix Screening – Rat Tail Collagen, Fibronectin etc.� GPCR Agonist Screening – CHO-M1 and carbamoylcholine� Induction of apoptosis with Fas ligand; inhibition of Fas with Vegf� Ion channel activation with bithionol� Bacterial & Protozoan Motility
Early Kinetic, Sensitivity Determinations Ab-Ag
Printed Rabbit IgG
Goat Anti-Rabbit IgGPolyclonal
Time Constant = 1/KaC
C = concentration
Ka = adsorption constant
Kinetics yield analyteconcentration
Antigen-Antibody Interaction (IgG) Is Resolved In Real-Time, and Quantitative
Note: The earliest ideas for applications centered on the dynamics of antibody binding, which has many uses in immunology, such as detecting active infections, autoimmune diseases, and monitoring and developing vaccines. • Goat antibodies that specifically bind rabbit antibodies can be obtained from many suppliers. • We printed rabbit IgG on a surface, injected the goat-anti-rabbit at 7 different concentrations, and monitored the binding real-time. • Knowing the concentration, the height of the curve at a given time can be used to derive an adsorption constant, or vice-versa.
Early Kinetic, Sensitivity Determinations Ab-Ag
Printed Rabbit IgG
Goat Anti-Rabbit IgGPolyclonal
Time Constant = 1/KaC
C = concentration
Ka = adsorption constant
Kinetics yield analyteconcentration
Antigen-Antibody Interaction (IgG) Is Resolved In Real-Time, and Quantitative
Note: Then if we plot the different concentrations’ time constants we can gain certainty that we have decent quantitative behavior, and fitting the slope we get a Adsorption Constant describing how quickly things bind. • The midpoint of the spot height vs Concentration curve is an approximation of an Affinity Constant that is variously referred to in molarity or
inverse molar terms,, depending on the context.• To report the complete affinity constants, the dissociation, or separation of the binding molecules has to be measured to get an “offrate” –
which is more varied than often assumed.
Early Kinetic, Sensitivity Determinations Ab-Ag
Printed:
• Rabbit IgG, • 2005 FluZone (recombinant)• 2005 Flushield (inactivated)• Human Serum Albumin
(background)
• Spike permits quantitation
The unusual kinetic curve shape is due to crude fluidics and resulting poor mixing dynamics.
Antigen-Antibody Interaction (IgG) Is Resolved In Real-Time, and Quantitative
0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
Sy
50454035302520151050Time (minutes)
Injection of 5% human serumSpiked with 125 nM goat anti-rabbit IgG
Flu Shield Flu Zone rabbit IgG Background
Note: Before moving ahead with such experiments, we jumped ahead with the human health relevant tests, trying to satisfy our curiosity.
We printed some flu vaccines and rabbit IgG, and injected 5% donor serum, spiked with 125nM of the Goat-Ant-Rabbit IgG.
Early Kinetic, Sensitivity Determinations Ab-Ag
Assuming the donor’s immunoglobulins have affinity equivalent to polyclonal goat IgG, nanomolar, the calculated Flu-specific iconcentration is 240nM.
The Detection limit for specific IgG is approximately 12nM at 1:20 dilution.
Antigen-Antibody Interaction (IgG) Is Resolved In Real-Time, and Quantitative
Note: And repeating the experiment at various concentrations, we again get some affinity numbers and a sense of just how sensitive the system may be.
Judging it to be just adequate for most diagnostics, we successfully undertook an improvement program increasing sensitivity almost 50-fold, bringing the detection limit to the picometer thickness scale.
Early Kinetic, Sensitivity Determinations Drug-Target
• Vancomycin kills Gram-positive bacteriaby binding to cell wall precursors
terminating in Lys-D-Ala-D-Ala
• Prevents peptidoglycan polymerization
• Drug of choice for treating methicillin-resistant Staphylococcus aureus (MRSA)
• Kinetics relevant to drug efficacy
• Acetyl-Lys-D-Ala-D-Ala: 330.4 Da
• Vancomycin: 1449.2 Da
Chloroeremomycin binding to acetyl-Lys-D-Ala-D-Ala; A. Try, et. al., J. Chem. Soc., Perkin Trans. 1, pg. 2911-2917, 1997
Note: With sensitivity improved, we confidently looked at more challenging systems, where the molecules were smaller than antibodies, and the affinity numbers weaker.
This system, which uses the specific binding of an antibiotic to a peptide that is found in bacterial cell walls, was deemed ideal.
Kinetic, Sensitivity Determinations Drug-Target
•1.8 nanometer thickness change confirms that the 3D hydrogel surface holds ~3 layers of peptides.
•Kinetic data provides onrate, offrate, equilibrium. Affinity constants agree with literature.
•Detection limit of 1 picometerestablished. This implies that a small molecule probe binding to an average sized protein target is detectable at low micromolar affinity.
• Note: So this is a single slide showing pretty much the whole experiment – a map of what we printed, how it looks when we image it with LFIRE, kinetic traces of on and off binding at various concentrations, and then concentration vs signal to get the binding constants.
• Lower limit of sensitivity is proven.
Small Molecule Immunoassay GentamicinNote, this assay was tuned for linearity in the 0.1-100ng/ml range.
1REU (nm) dynamic range of response, a layer of HRP.
Note: Gentamicin is an antibiotic which has a narrow therapeutic concentration range (ineffective low, toxic high) of 5-10ug/ml, and people vary in their ability to metabolize it, so testing for how much is in a patient’s blood to guide dosing is important.
This experiment shows we’re able to make tests at least 1000-fold more sensitive than necessary for that particular application.
Small Molecule Immunoassay GentamicinEnzyme- Amplified (Micro-ELISA)
This is not optimized, simply the result at 5 minutes, after substrate introduction at 10ug/ml.
Can be tuned 3 orders of magnitude in either direction, and linearity range expanded.
Note 28 REUs (nm) dynamic range, peroxidase precipitate is quite thick.
25X signal with Enzyme Amplification
Note: And what’s more, with use of some standard chemistry, we can amplify the signal magnitude – 25 REUs rather than 1 REU.
The curve does appear to move to the right (“less sensitive”), but the robustness of the signal is more important, we’re already 3 logs more sensitive than required!
Analytical Needs
Engineers’ Sensitivity: Bounded by first principles, i.e. Noise Floor, LLOD, Dynamic Range.Enables utility, defines possibilities.
Clinicians’ Sensitivity and Specificity:Based on a complex matrix of patient-dependent factors and preceding medical decisions, which correlated to the test’s performance, i.e. does having CMV Igg mean the patient has a disease?
FDA Sensitivity and Specificity depends on the predicate device
Note: When asked by different people about sensitivity, we have to carefully qualify and explain terms.An Engineer’s definition can define possibilities in a usefully broad way; system applicability, the statistical performance metrics.The Clinician’s view is all about accurately making the call of “disease” or “not disease”; what percentage of the time does it correctly identify the diseased and healthy samples?
When going to the FDA the demand is to be about as good as a predicate device. Many diagnostics are only 85% sensitive and specific!
Clinical Needs vs Technologies
1.E-181.E-151.E-121.E-091.E-061.E-031.E+00
Classic Biochem
Maven LFIRE POC
Genalyte
Luminex
ELISA
Electrochemiluminescence (MSD)
Classical Clinical Proteins
Tissue Leakage Biomarkers (Inj, CA)
Horm., IL & Cytok (Infl, Injury, CA)
Signaling Biomarkers (RUO)
ProteinDetectionTechnologies
Clinical Ranges
g/mL of plasma
Analytical Sensitivity of current platforms adequate for clinical use.
Note: A more useful way to group the various analytes a bit, and look at this is in relation to standard and emerging testing technologies. “Classic” clinical proteins like CRP are still useful, but the methods are commodified and therefore not subject to market disruption. Electrochemiluminescence (MSD), which as yet is best for measuring controlled systems in research settings, not complex samples, is overkill for diagnostics. The middle is well-served by any of the “middle mass” technologies… so what is the differentiator?
Clinical Needs vs Technologies – Maven is Multiplexing
Sensitivity Multiplexing
1.E-181.E-121.E-061.E+00
Classic Biochem
Maven LFIRE POC
Genalyte
Luminex
ELISA
Electrochemiluminescence (MSD)
0 100 200 300 400 500
g/mL of plasma tests per sample
Note: The differentiator is multiplexing, or the ability to test for many things at once with a single sample. Maven’s LFIRE POC can test for up to 450 analytes without signal cross-talk. Other considerations should include scalability and cost structure, because of course to help the most, it has to be affordable.
Genalyte Comparison
See 20170628 Genalyte Comparison.docx
Sensitivity: ParSpeed: MavenApplications: ParSensor Fabrication & Cost: Maven by farSurface Chemistry: MavenContent Loading: Near Par, but edge to MavenEase of Assay Development: Par (though Genalyte is ahead)Hardware Simplicity, Cost: Maven by farCartridge Assembly: Edge to GenalyteMicrofluidics: Par, but Genalyte is aheadSafety: Maven
Note: And if you’re following technology in the space, you know that Genalyte is Maven’s closest competitor with $92 million raised to-date.
We believe that there is a fundamental limit to the cost structure of Genalyte’s system, due to the $800 wafer itself, and sensitivity to geometric variability in their silicon-chip manufacturing process.
Maven in contrast uses a commodity glass chip, $100/square meter, standard optical coatings, $0.35 led light source and a simple stencil to build our sensors.
Genalyte Comparison
From “Multiplexed Diagnostics Enabled by Silicon Photonics”, a 2014 presentation by Ryan C. Bailey, a professor at UIUC. Bailey is a current consultant, stock option holder, and SAB member of Genalyte.
Note: As you can see the structure has critical thicknesses, multiple critical widths, and positioning requirements, as well as material constraints. If a resonator’s dimensions vary by a few nanometers in any direction it won’t work correctly.
There will be variation of all these things. This results in qc cost (”binning” and probably >25% waste) in producing their sensors, and still there are expensive hardware workarounds (monochromator light source and detector) to compensate for the remaining variability.
Maven in contrast uses a commodity glass chip, and simple (flat) standard optical coatings to build our sensors, and a LED that costs less than $0.50 and lasts five instrument lifetimes.
Diagnostics: Proof-of Concept Assays
Autoimmune and Infectious Disease for POC
Note: As mentioned before, we’ve worked on almost every class of biomolecule and organism, and we believe that we can do the most good by addressing the Autoimmune and Infectious Disease markets, and Allergy, with a Point Of Care System. We have also studied live cells.
The chemistries all involve detecting antibodies in a sample, a similarity that we address with one set of hardware and a consumable system that is easily tailored to the specific test. Cell testing detects the cells affixed to the sensor surface and we have grown cultures of cell directly on our glass surfaces.
From this point on, we’ll refer to Maven’s systems for Point of Care as “Polaron”.
Diagnostics: Autoimmune And Infectious Disease Testing
YAdd patient’sserum or plasma
immobilized antigenson the chip surface
Specific antibodies from the patient binding to the antigen is “positive”
Concept: One-Step “Screening” *Note: Being a label-free detection system, Polaron can see, real-time, the binding of patient’s antibodies to an antigen.
In allergy, this is typically IgE against a substance that is not normally the source.
In infectious disease, IgM against a pathogen is present early to fight off a new infection, while later, IgG is made to remain in the blood and combat subsequent infections.
In autoimmune disease, antibodies against proteins in the patient’s own body are present, sometimes several classes which correlate with disease stage.
Diagnostics: Autoimmune And Infectious Disease Testing
Concept: “class typing” For Early and Memory Immunity
YAdd Anti-IgGand Anti-IgM
Patient antibody and antigens on the chipsurface
IgM is an early-stage immune molecule, IgGis the memory for fighting off subsequent infections
**Y Y Y
IgE and IgA assays can be donefor allergy.
Note: Since the different classes of antibodies relate to the nature and stage of disease, in Polaron assays we use a “secondary antibody”, a non-human antibody against human antibodies, to confirm what kind of patient antibody is binding to the antigen.
Polaron can do this serially, showing proportions of different classes of antibodies from the same sample. As you’ll see later, this allows the combination of screening and class typing into a single test, effectively providing a 3-for-1 result in under an office visit’s time.
All while multiplexing, so a whole menu of tests is done at once.
Diagnostics: Autoimmune Antibodies
Collaboration with INOVA / Werfen, the #1 Worldwide OEM
Two Stages: 1) Pilot Screen with Smith Ag Only, 6 Negative, 5 Positive, incl. 2
incidental RNP+2) Semi-Blind 30 Samples, ENA5 x Screen, IgG &
IgM
24-sample silicone microwell format, 45 ulcapacity, instead of flow cells, glass substrate.
Note: Upon showing our early data with Rabbit IgG – Goat Anti-Rabbit IgG and Flu Antigen – Human Serum to a few diagnostic OEMs, all were interested, and we selected INOVA (now part of Werfen Group) as they are the leading US OEM in autoimmunity.
They proposed a two-stage process to see how well we could transfer one assay and discriminate between a pair of similar antigens, then to do a whole ENA panel.
The ENA panel is for differentiating Sjogren’s, Lupus, MCTD, Scleroderma, Polymyositis.
Diagnostics: Autoimmune Antibody
1) Pilot Screen Smith and Ribonuclear Protein (RNP) Ags Only, 13 Samples (6 Negative, 5 Sm Positive + 2 RNP+)
INOVA supplied: • 100ug each Sm and RNP Antigens• Samples• ELISA results
Note: INOVA gave us materials and samples to minimize the variables, to make sure we’re just comparing technologies.
INOVA Pilot #1
Sm Ag Only, 13 samples, IgG ELISA vs LFIRE Serum Screen
All 5 positive and 8 negatives agreed, ~proportional, semi-quantitative6.5
6.0
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Abso
rban
ce (O
D)
110
100
90
80
70
60
50
40
30
20
10
0
Phase Change (m
Rad)
N1
N3
SM1
SM2
SM3
SM4N4
N5
SM5
ELISA Signal LFIRE Signal
N6
RN
P1
RN
P2N2
SM2D
Low
C
Low ControlELISA
LFIRE ControlSM2 - 15 fold Dilution
(Amplified)
(Label-free)
Note: The summary looks like so.
Looking semi-quantitatively we clearly were able to discriminate between negative, Sm+, and RNP+ samples.
INOVA Pilot #1 - Printing Optimization
Printing Buffers Affect Sensitivity and Specificity
Smith Ag RNP Ag
Pri
nt B
uffe
rs a
-f
Sm: d>f>bRNP: d+
Sm: d>e>b=f>cRNP: d++>c>a
Array Before IncubationZ-axis is nanometers
At all stages, before or during the assay, quality of the printing is readily observed, so defects or absence is readily accounted-for and relative response can be factored in.
Sample Sm1: Sm+ Sample Sm3: Sm++
Note: Looking a bit deeper into the process, we printed each antigen six different ways, knowing that to match the ELISA sensitivity and specificity we have to get the protein immobilized with the antigenic site (the “epitope”) correctly oriented and exposed (folding and denaturing). Some print poorly, like C for example Some buffers print efficiently, but the antibody doesn’t bind at all, or only spuriously. See Row A. Some don’t bind every Positive Sample. See E. Some bind extraordinarily well, but can’t discriminate between Sm+ and RNP+ samples! D is very “hot”, but not reliable. Some spots bind well and discriminate. See B and F. The value of seeing print the quality with Polaron is obvious!.
INOVA Pilot #1
Sm Sample SM3 Quantitative Dilution and Kinetic Modeling
Time Constant = 1/KaC
C = concentration
Ka = adsorption constant
Kinetics yield analyteconcentration for puresamples
Kinetic curves are qualitatively distinct; first-order binding is typical for simple samples, the double exponential fit suggests several interactions.
Quantitative Behavior with Dilution
Note: With the printing strategy established and sense that the assay was working, we explored the real-time quantitative behavior of the system a bit.
Looking at this, it is reasonable to consider using the real-time binding curve as a substitute for a titered result in making disease severity judgments, positive or negative calls, or to simply make calls earlier.
INOVA Pilot #2
Scale Up to ENA5 IgG Plus Screen and IgM
INOVA Supplied:• 16 De-Identified Clinical Surplus Serum Samples• 100ug Sm, RNP, SSA, SSB, SCL-70 Antigens• Goat Anti-Human IgG and Goat Anti-Human IgM• ENA-5 IgG ELISA Kits to run in parallel
• After we shared our LFIRE-ELISA Comparison Data, they supplied some limited BioPlex (Illumina) data.
Printed ENA 5-Plex + Controls ArrayENA-5 Panel is used to help diagnose autoimmune syndromes,including Lupus, Scleroderma, and Sjogren’s Syndrome
IgM indicates a recent disease onset, IgG is historical or flare-up
Note: We used the data to get INOVA to supply us with the materials for the next stage, the ENA-5 full panel.
Their customers typically use IgG results in the approved manner to aid in diagnosis; many are using IgM to detect earlier disease.
INOVA Pilot #2
ELISA vs LFIRE Process Comparison
Maven LFIRE ENA5-Plex Array
Buffer in WellsAdd Sample 15 Min (Screen Data!)WashAdd Anti-IgG 10 Min (Isotype Data)Add Anti-IgM “Add Anti-IgA “
4.5ul Serum, 6 Steps, ~45 minutes15+ Assays: 5 Ags x 3X Iso + Screen
Scalable to 200 Ags
INOVA ENA-5 IgG ELISA
Buffer in WellsAdd Sample 1hr (no data!)WashAdd Anti-IgG 1hr (no data)WashAdd Streptavidin-HRPWashAdd SubstrateAdd Stop, Read (data!)
4ul Serum, 9 Steps, ~3 Hours5 Assays: 5 Antigens x 1 Isotype
Linear Single-Plex
Note: If we just compare the assay processes side by side, clearly Maven is simpler, faster, and more economical
INOVA Pilot #2
Printed Array Layout
Array Before Incubation, Z-axis is nanometers
SCL-70 is printed at low density, <1nM thick; Sm is 4.5nM; all others 2.75nm
IgM
IgG
IgA
RNP
SCL-70
Sm
SSBSSA
Printed ENA 5-Plex + Controls Array
Note: This is what the final array looked like. We printed human IgG, IgM, and IgA as controls for the second antibody class confirmation step, and each of the spots was printed in quintuplicate for statistical purposes.
The summary results for each antigen follow.
Images - Screen & Class Typing IgG & IgM & IgA
Serum 8171 Raw Screen RNP +++++
GAHIgA Class + 0%GAHIgG Class Confirms +++
GAHIgM Class Confirms ++
INOVA Pilot #2 ENA-5
Note: So in looking at the images from one sample you can get a sense of how easy it can be, from a developer’s perspective, to interpret results. Images like these are shown real-time on the prototype instruments. Some prototypes also do the image analysis, signal processing, and tabular data recording in real-time. It is very exciting to watch – the first couple times -- or when a lot is riding on a result!
Control IgAntigen Spots
IgG proportionately stronger but IgM quite significant.
Ig Class Typing - Screen at 1:40 & IgG & IgM & IgAResults Indicate Recent Disease Onset!
INOVA Pilot #2 ENA-5
Note: Looking at this one sample further – specifically that RNP result – we can use the serum screen and serial class typing to say something about this patient’s disease state. The Serum Screen is very strong, over 2000 units, as if a packed monolayer of globulins binds specifically to the RNP antigen. Given the brief incubation, assuming the affinity of immunoglobulins is somewhere in the high micromolar to low nanomolar, we can estimate that there are single digit micrograms per milliliter of RNP specific immunoglobulins in the serum! The significant IgM result suggests relatively recent onset. There is at least twice as much IgG, which conforms with this idea of staged development of antibody classes.
Sm IgG
SNR ~15/nM
INOVA Pilot #2 ENA-5
Cutofffor positive>
Note: And moving to the comparison data for all samples… In using our linear signal to noise units rather than a fully normalized unit system like ELISA, we’re afforded a look at the actual concentration and/or affinity of the specific diagnostic antibodies in the sample. In making our calls, we used the lowest “positive” sample’s Polaron signal-to-noise-ratio as the cutoff. We found that no Smith samples in this set approached the titer or affinity of the earlier sera. As mentioned, we have some luminex results for each of the samples, so some method-to-method-to-method comparison is possible.
Patient Serum ID Numbers
RNP IgG
SNR ~675/nM
INOVA Pilot #2 ENA-5
Cutofffor
positive>
Patient Serum ID Numbers
Note: RNP had a very wide range of signal responses, literally ten times more dynamic range than the Smith samples.
SCL-70 IgG
SNR ~375/nM
INOVA Pilot #2 ENA-5
Cutofffor positive>
Patient Serum ID NumbersNote: Surprisingly, as we had been advised that it was the most difficult assay, SCL-70 yielded quite clear calls. There was no sample provided with near-equivocal ELISA values.
SSA IgG
SNR ~10/nM
INOVA Pilot #2 ENA-5
Cutofffor
positive>
Patient Serum ID Numbers
Note: SSA yielded the single discordant point. Sample 8009 should have been just above the positive value, and our LFIRE result was not in the equivocal range. So we have a single false negative.
SSB IgG
SNR ~675/nM
INOVA Pilot #2 ENA-5
Cutofffor
positive>
Patient Serum ID Numbers
Note: SSB also had a surprising dynamic range.
Comparing Commercial ENA Profile 5 ELISAs and LFIRETM
LFIRETM
+ -+ 26 1- 0 123
ELISA
Concordance > 99%
Analytical Sensitivity 96%
Analytical Specificity 100%
INOVA Pilot #2 ENA-5 Clinical Utility Summary
Economics:ELISA: ~$375 in kits and 3 hours, no screen, 5 tests, 9+ stepsLFIRE: ~$ 45 materials and 45 minutes, 15+ tests, 5 steps
Expandable to a >80 test full panel for pennies/test
Note: And a quick summary shows excellent concordance, as well as outstanding economy of materials, labor, and time.
Economics – More Data, Less Time, Less MoneyLFIRE ENA5 Profile ELISAs
Sample Consumption 4.5ul 20ul (4ul x 5 kits)
Primary Incubation 15 min 1hr x 5 kits
Aspirate & Wash Once 3 Times x 5 Kits
Screen; Kinetic Titer Included Not possible; “x” wells
Ag Consum’n 3ng ~20ng x 5 kits
Anti-IgG, IgM Consum’n 45ng 20ng “
Streptavidin-HRP None 20ng “
Substrate None 100ng “
Stop None 200ul “
Hands-On Time <15 mins >60 mins
Time to All Results <1hr 3-12 hrs
Kit Price for 15 Assays <$100 (r&d, <10 samples) >$3000 (msrp, 40 samples)
Price per data point <$1 >$7.50
INOVA Pilot #2 ENA-5 Summary
Note: Digging deeper into how we can qualify and quantify these savings, the chart speaks for itself. We are simply doing so much more, for so much less!
Diagnostics: Model System for Infectious Disease
ToRCH (Toxoplasma, Rubella, Cytomegalovirus, Herpes) is a standard prenatal and newborn immunology panel.
Note: We have looked at broader infectious disease panels, and feel that ToRCH is a good model system.
Diagnostics: Infectious Disease
Antigens purchased from Meridian and Fapon.
Anonymous Clinical Surplus Samples from SLR Research, San Diego.
Limited clinical disease information, no reporting for multiple positivity
Reference values from various platforms depending on source:• Diasorin Eti-Max 3000• DPC Immulite• Bio-Rad Bioplex• Roche E170
Note: Unlike the ENA5 tests, which were supported by a commercial partner, we went at this independently.
We requested single disease state samples, and a lab result for each, not realizing that this, being the US, ToRCH is not always applied as a full panel, so the “negatives” are not reliably assured.
Diagnostics: Infectious Disease
ToRCH Array Layout
BSA HSA HIgG
HIgM T. gondii A T.gondii B
Rubella A Rubella B CMV A
CMV B HSV1 A HSV2 A
HSV2 B Fetuin RIgG
GAHIgG GAHIgM GAM-HRP
PBS-T BSA
QC Image, Pre-AssayNote: The array was printed as above.
Note that in addition to the controls for the class typing, the next-to-last row contains anti-human antibodies, to serve as a meter for the sample’s IgG and IgM content.
Diagnostics: Infectious DiseaseToRCH Sample Kinetic Screening Results at 1:50 Dilution -- small signal, big info!
Signal magnitudes andcurves suggest CMV Abhas high affinity and lowconcentration, while ToxoAb has lower affinity andhigher concentration.
CMV (Purple) +Saturated at 15min
Toxo (Red) +Continuing to bindat 45 minutes
Abundance and clonalities (isotype, affinity) for the different Ags!
deg
rees
0 10 20 30 40minutes
Note: A quick look at some real-time data from one sample which we found interesting. This is a small but clearly significant result from the Toxoand CMV antigens, but look at the shapes of the two curves. Clearly different! CMV antibody has a high on-rate, but there isn’t so much of it, while Toxo may have a high affinity but low concentration, or vice versa.
Diagnostics: Infectious DiseaseToRCH Sample Kinetic Class Typing Example
IgG and IgM proportionsdetect current or pastInfection.
Signals enhanced> five-fold
Every spot in the array(below) produces a kinetictrace.
anti-IgG
anti-IgM
diff
ere
ntia
l elli
pso
metr
yd
egre
e u
nits
Note: More contextual real-time observations from a multiply positive serum sample. When class-typing the response we see there are some antigens that are only IgG (like the yellow line at top), some that are IgG and IgM (the brown line at the middle). There are no IgM-only responses suggesting that there are no “very early and totally naïve” responses here.
Diagnostics: Infectious Disease
ToRCH IgG Result Images
All Samples
Reference Image Result Image IgG,HSV1+ Sample
Note: A sample of an IgG result, particularly strong CMV, and all results at the right.
Diagnostics: Infectious Disease
ToRCH IgM Result Images
All Samples
Reference Image Result Image IgM,HSV1+ Sample
Note: And the same sample’s IgM result.
Diagnostics: Infectious DiseaseToRCH IgG and IgM HSV1+ Compared
IgM Result
IgG Result
BSA HSA HIgG
HIgM T. gondii A T.gondii B
Rubella A Rubella B CMV A
CMV B HSV1 A HSV2 A
HSV2 B Fetuin RIgG
GAHIgG GAHIgM GAM-HRP
PBS-T BSA
Microarray with Spot IDs
Note: And again comparing IgG and IgM results. Interestingly we are seeing that our Anti-IgG may not be as specific as thought. Some reactivity with the IgM control spots and the IgM capture sandwich spots is seen.
Diagnostics: Infectious Disease
ToRCH Summary
Analytical Sensitivity 91%, Analytical Specificity 87%*
*Significant number of “false positives”
1. We used 1:50 dilutions (1ul / sample) to see the label-free binding curve at significant levels, while the reference systems use 1:100.
1. None of the “disease state” samples were tested by the supplier for the other antigens, effectively masking any multiple positives.
2. The default printing conditions were used, which may not be diagnostically most comparable without optimization.
Note: Looking at the overall results, we can say it was fair for a first stab proof of concept at >85% sensitivity and specificity. There is of course much to be done to make this a great ToRCH assay system, but clearly Polaron’s label-free imaging and real-time features make some of this far less obscure than they would be otherwise.
Diagnostics: Biomarkers, Interleukin 6 and IFN-gamma
IL-6 is clinically significant in Inflammation and Cardiovascular disease, but reference ranges are not well-characterized.
IFN-gamma is anti-viral, anti-parasitic, and anti-proliferative, and similarly, it’s utility and reference ranges are yet to be established.
Typical Testing Ranges:• Routine 1-500 nanograms/mL• High Sensitivity 0.1-100 picograms/mL
Note: Looking at the overall results, we can say it was fair for a first stab proof of concept at >85% sensitivity and specificity. There is of course much to be done to make this a great ToRCH assay system, but clearly Polaron’s label-free imaging and real-time features make some of this far less obscure than they would be otherwise.
Diagnostics: Biomarkers, Interleukins and Cytokines
Maven approached this as a “sandwich assay micro-ELISA” in which the microarray provides multiplex capabilities, while two antibodies and enzyme amplification increase sensitivity and dynamic range.
1) Anti-IL6 CaptureAntibody
3) Biotinylated Anti-IL6Detection Antibody
2) IL-6 Cytokine
4) Streptavidin-HRP
5) Enzymatic Substrate
silicon dioxide (glass) substrateNote: Recognizing that the size and concentration of the analytes and the affinity of the antibodies meant that a label-free assay would take several hours of incubation to be reliable, we decided a “sandwich assay micro-ELISA” would be necessary. Having enzyme-amplified a competitive gentamicin assay in the past, we were confident the approach would work.
Diagnostics: Biomarkers: Interleukin 6
Workflow matches traditional ELISA for simple adoption, ease of use.
VS
Print Slides
Blocking
Analyte
Detection Antibody
Streptavidin-HRP
DAB Substrate
Maven LFIRE Standard ELISA
Time Resolved Analysis
Coat Plates
Blocking
Analyte
Detection Antibody
Streptavidin-HRP
TMB Substrate
End Point Analysis
Note: Comparing the workflow, it looks just like an ELISA, except that we’re able to run several to hundreds of biomarker assays from a single sample in a single well or a single flow cell.
Diagnostics: Biomarkers, Interleukins and Cytokines
Spotting Scheme
� 750 um dot pitch (typical is 250 um)
� 10 drops / spot (~3.5 nL)
� ~ 300 um spot diameter (typical is 125um)
� 6 x 9 spots / well (typical is 225 spots)
� 20+ analytes per well (duplicate spots)
� Results confirm spot size & density can be adjusted to typical 1000/cm2 or 200+ tests per sample.
� Despite oversize spots, uses 1/80th the antibody of ELISA, 1/40th of Bio-Plexbeads.
Note: For the array features we “went big” at twice our typical spot diameter because we were transitioning to a new spotting instrument and had not received the small spot print head yet.
Diagnostics: Biomarkers, Interleukins and Cytokines
Time-Resolved Enzymatic Readout t0, t2.5, t5, t10 minutes
Typical Ag-Ab is 4000 pixel units, precipitate amplifies 5-100 fold.
High Dose4 ng/mL hIL-6
Medium Dose400 pg/mL hIL-6
t0 t2.5
t5 t10
t0 t2.5
t5 t10
t5.0 t5.0t10 t10
Note: And the first pass often includes some printing fine-tuning. We print antibodies at a presumptive saturating concentration, then bring it down in 20% increments. And what do we see? At high analyte concentrations, spots of antibodies at any concentration saturate quite quickly; at lower concentrations, response is moderated. The key is to choose conditions that maximize sensitivity and dynamic range within the time-to-result required.
Diagnostics: Biomarkers, Interleukin 6
Time-Resolved Dose-Dependent Response
Note: Mostly self-explanatory, just some real-time data from a dose-response series.
Diagnostics: Biomarkers, Interleukin 6
Well-to-Well n=4 and Dose-Dependent Variability is Acceptable
CV ≅ 0.5%
CV ≅ 3%
CV ≅ 17%
CV ≅ 9%
Note: Mostly self-explanatory, just some real-time data from a dose-response series. CV at higher concentrations very good
Diagnostics: Biomarkers, Interleukin 6
Dose-Response Snapshot, 2 Replicate Wells
Linear Response 5pg/mL to >1ng/mL
Note: And the dose-response is overall exactly on target for interleukin 6.
Diagnostics: Biomarkers IFN-ϒ
Dose-Response Snapshot
Linear Range 20pg/mL to >1ng/mL
10
100
1000
10000
100000
1 10 100 1000 10000
Sig
nal (
TIFF
uni
ts)
ifn gamma pg/ml
Signal vs [IFN-gamma]
Note: And not to neglect IFN-gamma, it also looks pretty good.
Diagnostics: Biomarkers, Interleukin 6 and IFN-ϒ
Assay Development Status / Conclusions
� Rapid development, less than two FTE weeks to transfer 2 high-sensitivity ELISAs to Maven Polaron Micro-ELISA.
� Detection range from 5 pg/mL - 5 ng/mL IL-6, 15 pg/ml – 1ng IFN- demonstrated on the first pass, one order of magnitude improvement in LLOD and range is anticipated.
� Acceptable reproducibility within each well and between wells.
� Familiar user workflow compatible with manual use, whole plate HT automation, or single samples for random access or point-of-care.
� Multiplexing capable (cytokine panels within each well) up to 200, notwithstanding antibody specificity limitations.
� Tremendous savings per assay, plus multiplex sample conservation.
Note: Did I mention that this was all done with a single assay developer on it? This is how quickly assays can be proved viable.
Diagnostics: Summary
Capable, Sensitive, Multiplexed
Interleukins to Immunodiagnostics
Note: And we wrap up the main presentation by emphasizing what should be obvious. Maven’s Polaron technology is perfect for point-of-care, from small biomarkers, to the largest macromolecules in fully real-time multiplexed format with over 200 assays/cassette.
Recent Work:
Development Assay Platform
Engineering The Instrument System
Note: Recent work involves commercializing the Point of Care platform. We’re developing a series of test assays that can be done anywhere, with any blood sample, to serve as an evaluation tool during development. And then there is the engineering of the hardware and consumable subsystems. There is extensive requirements development, project planning, vendor and material sourcing underway with good documentation plus budgets for the prototype development in hand.
� Flu Chip - Content
Spot # name n Description, stock concentration Printing Concentration Source1a,b Cy3/Cy5 Fiducial 2 Cy3 and Cy5 AMI AMI
2 Blocker 3 Protein Blocker (BSA-based or FCS) 1% from stock or as-used by AMI Candor (Maven) or AMI3 Buffer 3 Just the buffer, without proteins. AMI AMI4 HIgG 7 Positive Control for 2ndary, Amplification, 11.8mg/ml 10ug/ml, 90ug/ml BSA Jackson5 HIgM 3 Negative Control, 4.7mg/ml 10ug/ml, 90ug/ml BSA Jackson6 Naïve Goat IgG 3 Negative Control, 11.2mg/ml 100ug/ml Jackson7 Naïve Rabbit IgG 3 Negative Control, may + w/Rh+, 11.4mg/ml 100ug/ml Jackson8 “AP” 3 BSA + enzyme Positive Control for Amplif., 15 mg/ml in Tris 100ug/ml Jackson (ordered)9 GAHIgG FAb 3 Positive Control for Human, 2ndary, Amplif. 1.7-1.8mg/ml 100ug/ml Jackson
10 Flu1 Ag 3 A/California/7/2009 H1N1 0.639mg/ml 100ug/ml Protein Sci11 Flu2 Ag 3 A/Hong Kong/4801/2014 H3N2 0.624mg/ml 100ug/ml Protein Sci12 Flu4 Ag 3 B/Brisbane/60/2008 0.681mg/ml 100ug/ml Protein Sci13 Flu5 Ag 3 B/Phuket/3073/2013 Yamagata 1.2mg/ml 100ug/ml Protein Sci14 Flu6 Inact 3 B/Hong Kong/5/72 1.75mg/ml conc. supernatant 100ug/ml bio-rad-Ab 15 Flu7 Inact 3 A/New Caled./20/99 H1N1 1.60mg/ml conc. allant. 100ug/ml bio-rad-Ab 16 Jab 3 Fluvirin concentrated to > 1mg/ml TBD 100ug/ml Gift
Infectious Disease Model System
Note: As a training tool for our staff, and validation platform for new consumables, instruments, and processes, we are using an array we print with various controls, and a series of Flu antigens.
� Flu Chip – QC – Label-Free View
Infectious Disease Model System
IgG low control
Patient SerumTotal IgGMeter
HARA
T0 before sample T5 “difference” after sample
Note: The array is made to fit within a 3.5mm footprint and is laid out in a way that is geometrically distinct. The qc image of the array is on the left, an example of response to serum on the right. The sample’s serum metering spots are quite active while it appears there is some human anti-rabbit IgG antibody present.