Device and method for analysing liquid samples

Abstract

The invention relates to a device (1), a method, and a kit for analysing liquid samples. The device (1) comprises a sample layer (111) having a plurality of liquid permeable test sites (112) separated by a liquid impermeable barrier region (113), and an inlet part (2) comprising a plurality of inlet channels (211), which lead to respective test sites (112) of the sample layer (111), such that a flow connection between said inlet channels (211) and said respective test sites (112) is established or can be established, wherein said inlet channels (211) comprise first openings (218) and second openings (219), wherein a second surface area defined by the positions of said second openings (219) is smaller than a first surface area defined by the positions of said first openings (218) The invention further relates to a method for functionalizing a sample layer (111).

Claims

1. A device (1) for analysing liquid samples, wherein the device (1) comprises at least one sample layer (111) comprising a plurality of liquid permeable test sites (112) separated from each other by a liquid impermeable barrier region (113), wherein said device (1) comprises an inlet part (2), wherein said inlet part (2) comprises a plurality of inlet channels (211), and wherein said inlet channels (211) lead to respective test sites (112) of said at least one sample layer (111), such that a flow connection between said inlet channels (211) and said respective test sites (112) is established or can be established, wherein said inlet channels (211) comprise first openings (218), which are positioned in a first plane (p.sub.1), wherein said first openings (218) are accessible from the outside of said inlet part (2), such that liquid samples are loadable into said inlet channels (211) by means of said first openings (218), and wherein said inlet channels (211) comprise second openings (219), which are positioned in a second plane (p.sub.2) adjacent to said test sites (112), such that liquid samples can flow from said inlet channels (211) to respective test sites (112) via said second openings (219), characterised in that a first surface area is defined by the positions of the first openings (218) in said first plane (p.sub.1), and a second surface area is defined by the positions of said second openings (219) in said second plane (p.sub.2), wherein said second surface area is smaller than said first surface area, and in that at least one of said inlet channels comprises an angled section, wherein said angled section is arranged at an angle of 5 to 89 with respect to a plane (p) defined by said at least one sample layer.

2. The device (1) according to claim 1, wherein the device (1) comprises at least a top sample layer (115) and a second sample layer (116), and wherein said top sample layer (115) and said second sample layer (116) are positioned such that the test sites (112) of said top sample layer (115) overlap with respective test sites (112) of said second sample layer (116), such that a liquid permeable sample channel (114) extending through said top sample layer (115) and said second sample layer (116) is formed by the test sites (112).

3. The device (1) according to claim 1, wherein said angled section (220) is positioned at an angle () of 5 to 50 with respect to the plane (p).

4. The device (1) according to claim 1, wherein said inlet channels (211) comprise a cylindrical reservoir section (212), a conical transition section (221) and a connecting section (213), wherein said conical transition section (221) connects said reservoir section (212) and said connecting section (213), and said connecting section (213) leads to a respective test site (112).

5. The device (1) according to claim 4, wherein said reservoir section (212) has a volume in the range of 10 l to 1000 l.

6. The device (1) according to claim 4, wherein said reservoir section (212) has a volume of 3 l to 50 l.

7. The device (1) according to one claim 4, wherein said reservoir section (212) comprises a first diameter (d.sub.1 ), and said connecting section (213) comprises a second diameter (d.sub.2), wherein the ratio between said first diameter (d.sub.1 ) and said second diameter (d.sub.2) is at least 2 to 1.

8. The device (1) according to claim 1, wherein neighbouring first openings (218) are arranged at a first centre-to-centre distance (D.sub.1) with respect to each other in the first plane (p.sub.1), and wherein neighbouring second openings (219) are arranged at a second centre-to-centre distance (D.sub.2) with respect to each other in the second plane (p.sub.2), and wherein the ratio between the minimal first centre-to-centre distance (D.sub.1) and the minimal second centre-to-centre distance (D.sub.2) is at least 3 to 2.

9. The device (1) according to claim 1, wherein said device (1) comprises a separation membrane (3), wherein the separation membrane (3) is positioned in at least one of said inlet channels (211).

10. The device (1) according to claim 1, wherein said inlet channel (211) comprises at least one air passage (5), which connects said inlet channel (211) to the exterior.

11. The device (1) according to claim 1, wherein said device (1) comprises an optical unit (6) adapted to provide excitation light to a fluorophore and/or to measure light emitted by a fluorophore.

12. A method for analysing liquid samples by means of the device (1) according to claim 1, comprising the steps of: loading a liquid sample into a respective inlet channel (211) of said inlet part (2) in a loading step, passing said liquid sample through a respective test site (112) and/or sample channel (114), which is connected to said respective inlet channel (211), in an assay step, analysing substances bound to a sample layer (111) of the device (1) in an analysis step.

13. The method according to claim 12, wherein at least one of said liquid samples is a viscous sample having a dynamic viscosity of at least 3.Math.10.sup.3 Pa.Math.s, and wherein said viscous sample is diluted by a dilution factor in a dilution step prior to the loading step.

14. The method according to claim 13, wherein said viscous sample comprises a first component and a second component, and wherein said first component is separated from said second component in a separation step after said dilution step and prior to said loading step.

15. A method for functionalising a sample layer (111), comprising the steps of: providing a sample layer (111), wherein said sample layer (111) comprises a plurality of liquid permeable test sites (112) separated by a liquid impermeable barrier region (113), providing a reagent configured to capture analytes in one or more liquid samples, which is able to bind to said test sites (112), providing an inlet part (2) comprising a plurality of inlet channels (211), wherein said inlet channels (211) comprise first openings (218), which are positioned in a first plane (p.sub.1), wherein said first openings (218) are accessible from the outside of said inlet part (2), such that said one or more liquid samples are loadable into the inlet channels (211) by means of said first openings (218), and wherein said inlet channels (211) comprise second openings (219), which are positioned in a second plane (p.sub.2), wherein a first surface area is defined by the positions of said first openings (218) in said first plane (p.sub.1), and a second surface area is defined by the positions of said second openings (219) in said second plane (p.sub.2), wherein the second surface area is smaller than the first surface area, assembling said inlet part (2) and said sample layer (111), such that said test sites (112) of said sample layer (111) are aligned with respective second openings (219), such that liquid samples can flow from said inlet channels (211) of said inlet part (2) to said respective test sites (112) via said second openings (219), loading said reagent into at least one inlet channel (211), and passing said reagent through said respective test site (112), which is in flow connection with said at least one inlet channel (211).

16. A kit for performing the steps of the method according to claim 15 comprising: a sample layer (111), wherein the sample layer (111) comprises a plurality of liquid permeable test sites (112) separated by a liquid impermeable barrier region (113), a reagent, which is able to bind to said test sites (112) and an inlet part (2) comprising a plurality of inlet channels (211), wherein said inlet channels (211) lead to respective test sites (112) of said sample layer (111), such that a flow connection between said inlet channels (211) and said respective test sites (112) is established or can be established, wherein said inlet channels (211) comprise first openings (218), which are positioned in a first plane (p.sub.1), wherein said first openings (218) are accessible from the outside of said inlet part (2), such that liquid samples are loadable into the inlet channels (211) by means of said first openings (218), and wherein said inlet channels (211) comprise second openings (219), which are positioned in a second plane (p.sub.2), such that liquid samples can flow from said inlet channels (211) to respective test sites (112) via said second openings (219), wherein a first surface area is defined by the positions of said first openings (218) in said first plane (p.sub.1), and a second surface area is defined by the positions of said second openings (219) in said second plane (p.sub.2), wherein said second surface area is smaller than said first surface area.

17. The device (1) according to claim 1, wherein said angled section (220) is positioned at an angle of 10 to 45 with respect to the plane (p).

18. The device (1) according to claim 1, wherein the angled sections are arranged at different angles.

19. The device according to claim 18, wherein the angle increases from an outer inlet channel to the center of the inlet part.

Description

SHORT DESCRIPTION OF THE FIGURES

(1) FIG. 1 shows a schematic of the FoRe microarray. A) Each layer of nitrocellulose is patterned with an array of 400 m wax channels and functionalised with a different capture antibody. The layers are aligned with help of four pins, creating an array of multiplexed affinity columns. B) The nitrocellulose stack is clamped between a solid PMMA inlet and outlet to ensure contact and prevent leaking. Samples are funnelled from a 1.3 mm inlet well through 500 m angled channels in PDMS to the wax-patterned nitrocellulose channels.

(2) FIG. 2A shows a schematic of the model sandwich assay detecting mouse IgG and a side view of one channel to illustrate how the functionalised nitrocellulose was stacked.

(3) FIGS. 2B & C shows dose response curves at different antigen amounts and sample sizes. B) Six different volumes of mouse IgG (1 to 6 l, in 1 l increments) were injected into the FoRe device. Three different concentrations (5 pM, 25 pM, and 100 pM) were selected. The data points are the average of three spots from a single experiment (six for the negative controls for 25 pM and 100 pM) and the error bars are the standard deviation between the spots. C) For the data points depicted as squares, mouse IgG was diluted from 100 pM to 17 pM and the volume injected was adjusted to keep the antigen amount constant. The data points depicted as circles represent the same volumes, but without diluting the mouse IgG (constant 100 pM concentration). The error bars are the standard deviation between three spots in a single experiment (six for the negative controls).

(4) FIG. 3 shows the effect of sample volume on the limit of detection. A) Six different volumes of mouse IgG (7 pM) were compared to samples containing 5 mg/ml of BSA (negative control). Each data point represents the average of 10 spots from two independent experiments. The cut off for the limit of detection (dashed line) was calculated from 3 the standard deviation of the negative controls, incremented by the mean. The six intensities at 7 pM were each connected to the average negative control value with a linear line. B) The limit of detection for each of the 10 replicates was calculated independently and the average and standard deviation are plotted against the injected volume.

(5) FIG. 4 shows pictures of blood samples of different volumes after centrifugation (A) and a dose response curve for detecting rabbit IgG spiked into blood (B). A) Comparing the plasma available after spinning 5 l of blood to 5 l of blood diluted with PBS. While the supernatant on the left image is easy to separate from the pellet, the supernatant on the right is barely visible. B) The dose response curve is the average of four independent experiments. In individual experiments each concentration of rabbit IgG is repeated three times. The dotted line is the cut-off for the limit of detection, calculated from 3 the standard deviation of 12 negative controls (blood only) incremented by the mean.

(6) FIG. 5 shows a dose response curve for detecting TNF- spiked into blood. The curve is the average of three independent experiments. In each experiment the concentrations are repeated three times. The dotted line is the cut-off for the limit of detection, calculated from 3 the standard deviation of 9 negative controls (blood only) incremented by the mean. The concentrations represent the amount of recombinant human TNF- spiked into blood and assume the native concentration is negligible.

(7) FIG. 6 shows an example of multiplexed detection in blood. A) Schematic showing how the layers of nitrocellulose were assembled (side view) and how the three different samples (anti-rabbit IgG Alexa Fluor 488, anti-mouse IgG Alexa Fluor 488 and both combined) were injected into the device (top view). The injection pattern was designed to form an R on the rabbit layer and an M on the mouse layer. B) The fluorescence images show the sample and analyte multiplexing.

(8) FIG. 7 shows images comparing the adsorption of anti-mouse IgG Alexa Fluor 488 using three different techniques. A) Functionalisation by shaking. Anti-mouse IgG (100 g/ml, 150 l) was passively adsorbed on the exposed nitrocellulose spots during an hour long incubation step. B) Functionalisation by centrifugation. Anti-mouse IgG (20 g/ml, 10 l) was pulled through the patterned nitrocellulose at 201g (12 min). C) Functionalisation by spotting. Anti-mouse IgG (400 g/ml) was manually spotted using low-bind pipette tips (0,2 l per spot). Please refer to the section, FoRe Microarray Device Assembly in the Examples for additional details.

(9) FIG. 8 shows an example of an inlet part. A) A side view of the stack before clamping. The different layers are intentionally separated for clarity. B) A top view of the larger inlet wells micro-machined in PMMA (1.3 mm diameter). C) One channel filled with food dye to illustrate the sample flow from the large inlet wells through the angled 500 m PDMS inlet channels to the top layer of nitrocellulose.

(10) FIG. 9 shows an example of three inlet designs. A) The inlet reservoir volume was increased by stacking additional layers of PDMS and PMMA. The tapered inlets in the top PMMA layer (700 m top to 500 m bottom) fit a pipette tip for manual injection. B) Additional layers of PDMS and PMMA can be added to increase the sample volume to 6 l C) The angled channel inlet system for up to 12 l of sample. The PMMA wells (1.3 mm in diameter) can be easily enlarged to increase the reservoir volume.

(11) FIG. 10 shows dose response curves of a sandwich assay detecting mouse IgG. Six different volumes of 1000 pM and 500 pM mouse IgG were analysed, ranging from 1 l to 6 l (in 1 l increments). This resulted in the amount of antigen analysed overlapping for the two concentrations. The curves highlight both the sensitivity of the system to antigen amount (instead of concentration) and the large dynamic range (the system has not saturated with 1 ng of the antigen). The data points are the average of three replicates, except for the negative control, which is the average of six spots injected with 6 l of 1 mg/ml BSA. The error bars are the standard deviation.

(12) FIG. 11 shows an example fluorescence image of a rabbit IgG concentration series. Concentrations from (2) 6,7 pM to (7) 7,9 fM were analysed in triplicate. The reference spot (1) is 694 pM and the negative control (blood only) is indicated by (8).

(13) FIG. 12 shows a cross-section of a device for analysing liquid samples in an embodiment with angled connecting sections of the inlet channels.

(14) FIG. 13 shows a top view of a device for analysing liquid samples.

(15) FIG. 14 shows a cross-section of a device for analysing liquid samples in an embodiment with angled, conical inlet channels.

(16) FIG. 15 shows a cross-section of a device for analysing liquid samples in an embodiment with angled inlet channels with an additional built-in hydrophobic membrane and an additional air passage.

(17) FIG. 16 shows a cross-section of a device for analysing liquid samples in an embodiment with angled inlet channels with additional air passages.

(18) FIG. 17 shows a cross-section of a part of a device for analysis of liquid samples comprising an optical unit.

(19) FIG. 18 shows a cross-section of an inlet part according to the invention (A) and a device for analysis of liquid samples (B) in a further embodiment, wherein the inlet channels comprise angled sections.

EXAMPLES

(20) Volume Dependency.

(21) A sandwich assay using different sample volumes demonstrated that the FoRe array captures all the analyte as it flows through the layers. The stack was assembled as shown in FIG. 2A; the third layer was functionalised with anti-mouse IgG and the two layers above and one layer below were blocked with BSA. Three experiments were performed, each with a different concentration of mouse IgG (i.e. 5 pM, 25 pM, or 100 pM) spiked into 1 mg/ml of BSA to represent the high abundance serum proteins. For a given experiment, each volume (1 to 6 l, in 1 l increments) was injected in triplicate and the negative control consisted of six spots exposed to 6 l of 1 mg/ml BSA (three for the 5 pM sample). These three concentrations were chosen because in a system sensitive to antigen amount the curves overlap in this volume range (i.e. 5 l of 5 pM equals 1 l of 25 pM and 4 l of 25 pM equals 1 l of 100 pM). After manually injecting the samples, the device was spun at 129g for 12 min to ensure that all the liquid passed through the nitrocellulose. The third layer was then incubated in anti-mouse IgG Alexa Flour 488 before imaging. The three overlapping curves in FIG. 2B demonstrate that the device is sensitive to the total antigen amount. While these results only illustrate that we always capture the same proportion of the analytes passing through the membranes (independent of concentration), because of the high excess of capture probes the most plausible explanation is that all of the analytes are captured in the array spot. The high binding capacity also results in a large dynamic range and we could inject 6 l of a 1000 pM solution without reaching saturation (FIG. 10).

(22) We tested the influence of dilution on the amount of captured antigen (FIG. 2C). Again we used the four-layered stack and sandwich assay presented in FIG. 2A. The 100 pM sample of mouse IgG was diluted in BSA (100 pM to 17 pM) and the injected volume was adjusted to keep the amount of mouse IgG in each sample constant (i.e. we injected 2 l of the sample diluted 2, 3 l of the sample diluted 3, etc.). The six injected volumes ranged from 1 to 6 l. In FIG. 2C we compare this result to a 100 pM sample where the concentration was kept constant but the volume increased at the same rate as for the dilution series. The dilution series plateaus at 100 pM and the constant concentration series continues to linearly increase. This result indicates that dilution does not affect the sensitivity of the FoRe array, as long as we increase the sample volume by the same factor.

(23) Improving the Sensitivity.

(24) By capturing all the analyte in a sample the FoRe array is uniquely able to tailor the sensitivity based on the sample volume. FIG. 3 shows how the limit of detection (LOD) decreases with increasing sample volume. We assembled a four layer stack, where the third slice was functionalised with anti-mouse IgG and the other three were blocked with BSA. For these experiments we used the angled inlet channels (FIG. 1B) to increase the sample volume to 10 l. Mouse IgG (7 pM) was spiked into 5 mg/ml of BSA and six volumes from 5 l to 10 l were injected into the device. The FoRe array was centrifuged at 201g for 12 min before incubating in anti-mouse IgG Alexa Flour 488. Each volume is represented by 10 spots from two independent experiments, and the average and standard deviation for the six different volumes are plotted in FIG. 3A. The cut-off for the limit of detection was calculated by taking the average signal of the 10 negative controls (channels injected with only BSA) increased by 3 its standard deviation. To determine the LOD for a given volume, we plotted the 10 normalised data points and fit each with linear line from the value at 7 pM to the negative control. We determined where each fit intersected with the limit of detection line. For each volume the average and standard deviation of the LOD concentrations are plotted in FIG. 3B. As expected, for the factor two increase in volume the sensitivity of the system also increased by a factor of 2 (1,760,33).

(25) Analysis in Complex Samples.

(26) The FoRe microarray is compatible with whole blood analysis using a simple dilution trick. Without pre-processing, viscous or complex samples rapidly clog the nitrocellulose membranes, preventing the samples from flowing through and inducing leaking between the layers. While plasma readily flows through the device (de Lange & Vrs, 2014, Anal Chem 86(9), 4209-4216), the cells in whole blood are too large to pass through the 0.45 m pores (data not shown). Plasma separation membranes (e.g. the Vivid Plasma Separation Membrane, Pall Corporation) have been successfully incorporated into 3D paper-based analytical devices for multiplexed analysis from a finger prick of whole blood (Vella et al., 2012, Anal Chem 84(6), 2883-2891). However, these membranes can only process 50 l of blood per cm.sup.2, and with the small microarray test sites this would limit our device to 100 nl sample volumes. In another approach, Ge et al. mixed whole blood with an agglutination factor and used the top layer of cellulose to filter out the large multi-cellular aggregates (Ge et al., 2012, Lab Chip 12(17), 3150-3158). This was also not possible with our micron channels as the blood cells quickly blocked the membranes during filtration and the plasma could not pass through. Pre-separating the blood cells from plasma is very challenging in the 4 l volume attained from an infant heel prick (Vella et al., 2012, Anal Chem 84(6), 2883-2891). However, as we anyway capture everything that passes through the layers we are allowed to dilute the blood with buffer, and easily separate the larger volume of diluted plasma from the blood cells (FIG. 4A). Then passing the entire supernatant through the device is equivalent to analysing the smaller volume of undiluted plasma.

(27) We demonstrated this concept with a sandwich assay detecting rabbit IgG spiked into whole blood. The FoRe array was assembled using the angled inlet channels and four layers of functionalised nitrocellulose (i.e. BSA, BSA, anti-rabbit IgG, BSA). Six concentrations of rabbit IgG ranging from 6,7 pM to 7,9 fM were spiked into blood. We then mixed 5 l of each concentration with 10 l of PBS. The samples were spun at 14 100g for 3 min to separate the blood cells. We injected 10 l of the supernatant into the device and centrifuged the samples through the nitrocellulose layers (201g for 12 min). Each concentration was analysed in triplicate for a given experiment and the dose response curve in FIG. 4B is the average of four independent experiments (see FIG. 11 for the fluorescence image). The LOD was 21 fM, calculated by taking the average signal of 12 negative controls (blood samples without spiked in rabbit IgG) increased by 3 its standard deviation. For three of the experimental repeats the functionalisation was done by passively adsorbing the capture antibodies during an hour long incubation step with gentle shaking. The functionalised slice for the fourth repeat was prepared by flowing the capture antibody through the patterned nitrocellulose (as described in the Experimental Methods section). There was no noticeable difference in the dose response curve from this experiment, indicating that flow-through functionalisation is a feasible alternative. This is advantageous both to reduce the cost of expensive reagents and when the capture antibody buffer is not compatible with wax printing techniques (e.g. contains a surfactant which compromises the hydrophobic barriers) (Deiss et al., 2014, Angewandte Chemie 53(25), 6374-6377).

(28) We demonstrated the importance of flow-through functionalisation with a sandwich assay detecting TNF-. The capture probe was provided in liquid, and the storage buffer was not compatible with passive functionalisation. When the anti-TNF- capture antibody was passively adsorbed on the surface we observed considerable leaking on the functionalised slice after running the assay. To better control the flow, 1-mm thick PDMS pieces with an array of holes matching the wax pattern were placed above and below the anti-TNF- layer. However, this only prevented leakage when we switched to the flow-through functionalisation. We functionalised the layers by spinning 1 l (200 g/ml) of anti-TNF- through one layer of nitrocellulose (129g, 3 min). The nitrocellulose was rinsed in 1 ml of arraying buffer (5 min, gentle shaking), dried first under a stream of nitrogen and then for 1 h at 37 C. The layer was blocked with BSA as described in the Experimental Methods section. The functionalised slice was placed in the second position of a four layer stack. Six concentrations of TNF- (240 pM to 7,5 pM) were spiked into blood and processed as described above for the rabbit IgG sandwich assay, using TBS instead of PBS as the dilution buffer. The device was spun at 201g for 15 min (3 min longer than usual) because of the extra PDMS layers. FIG. 5 is the dose response curve for three independent experiments detecting TNF-. The concentrations represent the amount of recombinant human TNF- added to the whole blood, and we assumed that the native concentration (pg/ml) was negligible in this range. The limit of detection was 18 pM, calculated by taking the average signal of 9 negative controls (blood samples without spiked in TNF-) increased by 3 its standard deviation. The differences in the sensitivity of the device for the different analytes (i.e. mouse IgG, rabbit IgG or TNF-) can be attributed to differences in the binding affinities, which is also illustrated by the shape (i.e. the position of the inflection point) of the dose-response curves.

(29) We used a direct labelled assay to demonstrate target multiplexing in blood. While the binding of target proteins can suffer from the presence of a label and introducing a detection antibody improves the specificity (Hartmann et al., 2009, Anal Bioanal Chem 393(5), 1407-1416), the assay is faster (one incubation step is eliminated) and less expensive (Wilson, R., 2013, Expert Rev Proteomics 10(2), 135-149). The direct-labelled assay is well-suited to our multiplexing experiment because it allows us to directly visualise the target binding and highlights the compatibility of the device with different immunoassays.

(30) The layers in the stack were functionalised with: BSA, mouse IgG, rabbit IgG, and BSA (FIG. 6A). The three sample solutions were anti-mouse IgG Alexa Fluor 488 (5 g/ml), anti-rabbit IgG Alexa Fluor 488 (5 g/ml), or a combined sample (5 g/ml of each) spiked into blood. As before, each channel analyses 5 l of blood, diluted with 10 l of PBS. The samples were spun at 14 100g for 3 minutes and 10 l of the supernatant was injected following the pattern shown in FIG. 6A. The samples containing anti-mouse IgG should bind to layer 2, the anti-rabbit IgG to layer 3 and the combined sample to both layers. The pattern from the sample injection forms an R on the rabbit layer and an M on the mouse layer. The fluorescence images in FIG. 6B clearly show that the FoRe array is capable of multiplexed analysis in blood; the mouse and rabbit samples bound specifically to the correct layers and the combined samples appeared on both layers with no obvious loss of intensity.

(31) Detailed Description of the FIGS. 12 to 18.

(32) FIG. 12 shows a cross-section of a device 1 for analysing liquid samples in an embodiment with angled, connecting sections 213 of the inlet channels 211.

(33) The device 1 comprises a top plate 214 with an array of top plate openings 217, each large enough to fit a pipette tip. A reservoir part 215 comprising an array of reservoir sections 212 is positioned directly below the top plate 214, such that each opening 217 overlaps with a respective reservoir section 212.

(34) A connecting part 216 is arranged below the reservoir part 215. The connecting part 216 comprises an array of connecting sections 213, which are arranged such that the top part of each connecting section 213 overlaps with a respective reservoir section 212 of the reservoir part 215, wherein a respective inlet channel 211 is formed from each connecting section 213 and the respective reservoir section 212. Each connecting section 213 is arranged at an angle with respect to the plane defined by the at least one sample layer 111, depicted as the width w, wherein the angle differs from 90 for some connecting sections 213. That is, the connecting part 216 comprises angled sections 220.

(35) The inlet part 2 is comprised of the top plate 214, the reservoir part 215, and the connecting part 216.

(36) The device 1 further comprises a stack of sample layers 119 comprising a top sample layer 115, a second sample layer 116, and a bottom sample layer 116a. The stack of sample layers 119 is arranged between an upper sealing part 117a, and a lower sealing part 117b, which seal the sample layers 111 against leakage. Each sample layer 111 comprises a plurality of liquid permeable test sites 112, and a liquid impermeable barrier region 113, wherein the barrier region 113 separates the test sites 112 of the respective sample layer 111 from each other. The test sites 112 of the sample layers 111 are arranged such that respective test sites 112 of neighbouring sample layers 111 overlap, thereby forming a plurality of sample channels 114 extending through the stack of sample layers 119.

(37) The upper sealing part 117a comprises a plurality of upper sealing part openings 122a, and the lower sealing part 177b comprises a plurality of lower sealing part openings 122b. Therein the upper part of each upper sealing part opening 122a overlaps with a respective connecting section 213 of the connecting part 216. The lower part of each upper sealing part opening 122a overlaps with a respective test site 112 of the top sample layer 115. The upper part of each lower sealing part opening 122b overlaps with a respective test site 112 of the bottom sample layer 116a.

(38) The device 1 further comprises a frame 120, which is positioned in parallel to the height h, and surrounds the reservoir part 215, the connecting part 216, the upper sealing part 117a, the lower sealing part 117b, and the stack of sample layers 119. The frame 120 ensures the correct alignment of the parts of the device 1.

(39) The device 1 further comprises a bottom plate 118, which is arranged in parallel to the width w and forms the lower boundary of the device 1. The bottom plate 118 comprises a plurality of outlets 123, wherein each outlet 123 overlaps with the lower part of a respective lower sealing part opening 122b.

(40) The device 1 further comprises a clamp or spring-loaded tension lock 121, which is arranged in parallel to the height h, wherein the clamp or spring-loaded tension lock 121 covers the side walls of the device 1, and part of the top and bottom boundaries of the device 1, wherein the top plate openings 217, and the outlets 123 are left open. A mechanical force is applied by means of the clamp or spring-loaded tension lock 121 on the components of the device 1 by the top plate 214 and the bottom plate 118 to ensure sealing of the device 1 to the exterior and avoid leakage of samples.

(41) The device 1 is arranged such that a flow connection between a top plate opening 217, a respective reservoir section 212, a respective connecting section 213, a respective upper sealing part opening 122a, a respective sample channel 114, comprising a plurality of test sites 112 of a plurality of sample layers 111, a respective lower sealing part opening 122b, and a respective outlet 123 can be established.

(42) FIG. 13 shows a top view of a device 1 for analysing liquid samples. The device 1 is characterised by a width w, and comprises a clamp or spring-loaded tension lock 121, and a top part 214 with a plurality of top plate openings 217. Through the top plate openings 217, the respective inlet channels 211 are visible.

(43) FIG. 14 shows a cross-section of a device 1 for analysing liquid samples in an embodiment with angled, conical inlet channels 211.

(44) The device 1 comprises a top plate 214, an upper sealing part 117a, a stack of sample layers 119, a lower sealing part 117b, a bottom plate 118, a frame 120, and a clamp or spring-loaded tension lock 121 arranged analogously to the device 1 shown in FIG. 12.

(45) A connecting part 216 is arranged between the top plate 214 and the upper sealing part 117a. The connecting part 216 comprises an array of inlet channels 211, which are arranged such that the top part of each inlet channel 211 overlaps with a respective top plate opening 217. Each inlet channel 211 is arranged at an angle with respect to the width w, wherein the angle differs from 90 for some inlet channels 211. That is, the connecting part 216 comprises angled sections 220.

(46) The inlet part 2 is comprised of the top plate 214 and the connecting part 216.

(47) Each inlet channel 211 overlaps with a respective upper sealing part opening 122a at the bottom part of the connecting part 216, which is positioned adjacent to the upper sealing part 117a.

(48) Each inlet channel 211 has a conical shape, wherein the first diameter d.sub.1 of the inlet channel 211 at the connection to the respective top plate opening 217 is larger than the second diameter d.sub.2 of the inlet channel 211 at the connection to the respective top sealing plate opening 122a.

(49) The device 1 is arranged such that a flow connection between a top plate opening 217, a respective inlet channel 211, a respective upper sealing part opening 122a, a respective sample channel 114, comprising a plurality of test sites 112 of a plurality of sample layers 111, a respective lower sealing part opening 122b, and a respective outlet 123 can be established.

(50) FIG. 15 shows a cross-section of a device 1 for analysing liquid samples in an embodiment with angled inlet channels 211 with an additional hydrophobic membrane 4 and an additional air passage 5. That is, the connecting part 216 comprises angled sections 220.

(51) The device 1 comprises a top plate 214, a reservoir part 215, a connecting part 216, an upper sealing part 117a, a stack of sample layers 119, a lower sealing part 117b, a bottom plate 118, a frame 120, and a clamp or spring-loaded tension lock 121 arranged analogously to the device 1 shown in FIG. 12.

(52) A hydrophobic membrane 4 is positioned between the connecting part 216 and the upper sealing part 117a. The hydrophobic membrane 4 comprises a plurality of holes 411, wherein each hole 411 overlaps with a respective connecting section 213 of the connecting part 216, and a respective test site 112 of the top sample layer 115.

(53) The frame 120 comprises an air passage 5 positioned adjacent to the hydrophobic membrane 4, so that air trapped at the hydrophobic membrane 4 may escape through the air passage 5.

(54) FIG. 16 shows a cross-section of a device 1 for analysing liquid samples in an embodiment with angled inlet channels 211 with additional air passages 511, 512. That is, the connecting part 216 comprises angled sections 220.

(55) The device 1 comprises the parts described for FIG. 12 in an analogous arrangement. The connecting part 216 comprises at least one first air passage 511, wherein the first air passage 511 is connected to at least one connecting section 213 of the connecting part 216 in a flow connection, such that air trapped in the connecting section 213 may escape the connecting section 213 through the first air passage 511. The frame 120 comprises at least one second air passage 512, wherein the second air passage 512 is connected to the respective first air passage 511 and to the exterior in a flow connection, such that air may escape from the first air passage 511 through the second air passage 512 to the exterior.

(56) FIG. 17 shows a cross-section of a part of a device 1 for analysis of liquid samples comprising an optical unit 6. The optical unit 6 comprises a light source 611, a first optical fibre 612, a second optical fibre 613, and a photo detector 614.

(57) The light source 611 provides light, particularly excitation light, which is able to excite a fluorophore. The light is guided through the first optical fibre 612 onto the test site 112 of the sample layer 111, particularly such that fluorophores positioned at the test sites 112 are excited. The second optical fibre 613 is positioned such that light provided by a substance at the test site 112, particularly fluorescence light emitted by a fluorophore positioned at the test site 112, travels through the second optical fibre 613 to the photo detector 614, which is adapted to generate a signal in response to light, particularly the light guided by the second optical fibre 613.

(58) FIG. 18A shows a cross-section of an inlet part 2 according to the invention comprising inlet channels 211, which comprise a reservoir section 212 and a connecting section 213, wherein each reservoir section 212 is connected to a corresponding connecting section 213 by means of a conical transition section 221. Therein, the reservoir sections 212 and the connecting sections 213 are incorporated in a single inlet part 2. Five inlet channels 211 are depicted, wherein the connecting sections 213 of the outer four inlet channels 211 are angled sections 220, comprising an angle of less than 90 with respect to a plane p by the at least one sample layer 111 depicted in FIG. 18B. The angle is smallest in the outer inlet channels 211 and increases towards the center of the inlet part 2, wherein the connecting section 213 of the center inlet channel 211 is arranged at an angle of 90 and is therefore not an angled section 220. The reservoir sections 212 are arranged at an angle of 90. Furthermore, the reservoir sections 212 comprise a cross-sectional first diameter d.sub.1, and the connecting sections 213 comprise a cross-sectional second diameter d.sub.2, wherein the first diameter d.sub.1 is larger than the second diameter d.sub.2, and wherein the diameter decreases in the conical transition sections 221. That is, the first diameter d.sub.1 of the respective transition section 221 at the connection to the respective reservoir section 212 is larger than the second diameter d.sub.2 of the transition section 221 at the connection to the respective connecting section 213.

(59) Each reservoir section 212 comprises a respective first opening 218 arranged in a first plane p.sub.1 parallel to the at least one sample layer 111 at the distal side of the inlet part 2 with respect to the at least one sample layer 111, and each connecting section 213 comprises a respective second opening 219 arranged in a second plane p.sub.2 parallel to the at least one sample layer 111 at the proximal side with respect to the at least one sample layer 111, when the inlet part 2 is assembled with the at least one sample layer 111 as depicted in FIG. 18B. As a result of the arrangement of the angled sections 220 and the ratio between the first diameter d.sub.1 and the second diameter d.sub.2, the centre-to-centre distance D.sub.2 of the second openings 219 is smaller than the centre-to-centre distance D.sub.1 of the first openings 218. This allows to load large sample volumes, i.e. for diluted samples, into the reservoir sections 212, and to apply the samples to small sample layers 111 comprising a densely spaced arrangement of test sites 112.

(60) FIG. 18B shows a sectional view of a device 1 for analysing liquid samples comprising the inlet part 2 depicted in FIG. 18A as well as further parts to those depicted in FIG. 12 to FIG. 16. The device 1 is assembled in an analogous manner to the devices 1 shown in FIG. 12 to FIG. 16.

(61) The setup shown in FIGS. 18A and B advantageously allows the use of the method for analysing liquid samples according to the invention, wherein essentially the total amount of a component of a diluted complex liquid sample can be captured by means of capture compounds, i.e. antibodies bound to the test sites 112. Furthermore, a small sample layer, particularly having dimensions of 55 mm or less advantageously allows to completely scan an entire sample layer at high resolution for optical signal analysis.

(62) Materials and Methods

(63) Materials.

(64) Alexa Fluor 488 anti-mouse IgG (H+L, produced in goat, highly cross-adsorbed), Alexa Fluor 488 anti-rabbit IgG (H+L, produced in goat, highly cross-adsorbed), streptavidin Alexa Fluor 488 conjugate and the TNF- human antibody pair kit, including anti-TNF-, biotinylated anti-TNF-, and recombinant human TNF- standard (Novex) were purchased from Invitrogen, Switzerland. The following antibodies were purchased from Sigma-Aldrich, Switzerland: IgG from mouse serum, IgG from rabbit serum, IgG from goat serum, anti-mouse IgG (produced in goat) and anti-rabbit IgG (produced in goat). The 3D array layers were Amersham Premium 0.45 m nitrocellulose membranes from VWR International, Switzerland. The membranes were functionalised with antibodies prepared in protein arraying buffer from Maine Manufacturing (Kerafast Inc., Boston, USA) and blocked with albumin from bovine serum (98%; Sigma, Switzerland). All other protein solutions were prepared in Tris buffered saline (TBS, Sigma, Switzerland), expect those for the TNF- assays, which were prepared in GIBCO phosphate buffered saline (pH 7,4; Invitrogen, Switzerland). TBS buffer was purchased either 10 concentrated or as tablets and used after diluting in ultrapure water (Milli-Q gradient A 10 system, Millipore Corporation, Switzerland) and filtrating (0,2 m). The polydimethylsiloxane (Sylgard 184, Dow Corning) for micro-moulding inlet reservoirs was prepared at a 10:1 ratio with its crosslinker. EDTA-stabilised blood was purchased from Blutspende Zrich (Zurich, Switzerland) and stored at room temperature for up to 1 week from when it was drawn.

(65) FoRe Microarray Device Assembly.

(66) The FoRe array was prepared as described previously (de Lange & Vrs, 2014, Anal Chem 86(9), 4209-4216), with the exception of the new inlet design. Briefly, the multiplexed affinity columns are formed by stacking wax-patterned and biofunctionalised nitrocellulose membranes. Hydrophobic wax barriers surround the antibody-loaded spots on each layer, allowing liquid to pass through vertically while isolating samples from each other laterally (FIG. 1A). The wax is printed with a solid ink printer (ColorQube 8570, Xerox, Switzerland) and quickly melted in an oven (125 C., 2 min) to extend the liquid barrier through the thickness of the porous nitrocellulose (Lu et al., 2010, Anal Chem 82(1), 329-335). Please note that nitrocellulose is highly flammable and has a flash point of 200 C. The microarrays consist of 25 spots, arranged in a 55 square. Each spot is approximately 400 m in diameter with 1.2 mm centre-to-centre spacing.

(67) After wax patterning, the nitrocellulose layers are functionalised by passively adsorbing the capture probes. A capture antibody solution of 100 g/ml was prepared in protein arraying buffer. We added 150 l of the capture antibody solution to a 6-mm polydimethylsiloxane (PDMS) reservoir above the array and incubated the slices for 1 h on a rotary shaker. The slices were rinsed briefly with arraying buffer (150 l, 5 min, gentle shaking) and dried under a stream of nitrogen. To improve protein adhesion, the slices were left at 37 C. for 1 h. The remaining binding sites were blocked with 1% (w/v) bovine serum albumin (BSA) to prevent nonspecific adsorption to the nitrocellulose (1 ml of BSA, 30 min, gentle shaking). The layers were then rinsed twice with TBS (1 ml, 10 min) and once with Millipore water (1 ml, 5 min). The slices were dried with nitrogen and stored for short term at room temperature and for longer at 4 C.

(68) We investigated two other functionalisation approaches to reduce the required amount of capture antibody (see FIG. 7 for more details). We functionalised the slices by manually spotting 0,2 l (400 g/ml) of the capture antibody on each spot; however, this resulted in highly variable amounts on each test site. Slices were also functionalised by assembling a one layer stack and flowing 10 l (20 g/ml) of the capture antibody solution through the channels (12 min, 201g) before the standard 1 h drying and blocking steps. The capture probe distribution was more uniform and comparable to the bulk approach presented above.

(69) To align the slices, four holes are punched out of the nitrocellulose with a biopsy punch (KAI biopsy punch, Medical-Impex, Germany) and the layers are stacked with the aid of four, 1 mm-diameter pins (FIG. 1A). The stack of nitrocellulose is clamped between micromachined poly(methyl methacrylate) (PMMA) inlet and outlet pieces. A PDMS layer, with an array of angled 500 m-diameter channels connects the wax pattern on the nitrocellulose with the larger wells (1.3 mm diameter) in the PMMA inlet (FIG. 1B). This novel design makes it possible to increase the sample volume without making the device impractically tall or compromising the high spot density (FIG. 8 for device pictures). We also fabricated an inlet array with vertical 500-m channels, which was used for testing volumes in the range of 1 to 6 l (FIG. 9). In the first version, inlet channels were 18 mm tall and injection was done in two steps with a GELoader pipette tip (used in FIG. 2C). The second version increased the channel height to 31 mm, assembled in several parts, to inject up to 6 l (used in FIG. 2B).

(70) Immunoassays.

(71) The device tests 25 independent samples for a variable number of proteins. We used four-layer stacks for the experiments in this publication, but have previously assembled stacks with up to ten layers and additional slices could be included if needed. The 3D arrays were secured to the top of a 6-well plate and after manually injecting the samples the device was centrifuged to pull the liquid through the channels. The speed and duration were adjusted for the different inlet designs to ensure that the entire sample passed through the nitrocellulose layers. Experiments performed with the 31 mm vertical channels were spun at 129g for 12 min and with the angled channels at 201g for 12 min. In the 18 mm vertical channels samples were either spun at 129g for 6 min (FIG. 2C, constant concentration curve) or 453g for 3 min (FIG. 2C, dilution series curve) after each injection step. After centrifugation, the layers were separated with tweezers and rinsed three times in TBS (1 ml, 10 min, gentle shaking). The microarrays were incubated in 150 L of the detection antibody (5 g/ml, 1 h, gentle shaking) and then rinsed three times with buffer (1 ml, 10 min) before imaging. The detection antibody was spiked into 1 mg/ml BSA to reduce nonspecific adsorption. For some experiments 0.5 mg/ml of goat IgG was additionally added, but this did not appear to improve the signal-to-noise and was removed from later experiments. The detection antibodies for TNF- were biotinylated and needed an additional incubation in streptavidin Alexa Flour 488 (5 g/ml, 30 min, gentle shaking) and rinsing before imaging.

(72) Blood samples were prepared by diluting 5 l of whole blood with 10 l of PBS in an Eppendorf tube. The mixture was spun at 14 100g for 3 min to sediment the red blood cells and any larger fragments which might clog the nitrocellulose. We removed 10 l of the supernatant and injected it into the FoRe microarray channels. To simplify the experimental protocol some replicates were prepared by diluting 15 l of blood with 30 l of PBS and injecting 10 l of supernatant into three different channels. Both approaches were employed to produce the dose response curve in FIG. 4 (i.e. the 3 replicates diluted individually or together), and there was no noticeable difference. The assembled stack was centrifuged at 201g for 12 min to pull the diluted plasma through the layers. This was followed by rinsing and incubation in the detection antibody, as described above. The slices were then clamped between two microscopy slides to flatten them for automated imaging with a confocal laser scanning microscope (please see ESI for details on imaging and data analysis).

(73) Imaging and Data Analysis

(74) Fluorescence images were taken with a Zeiss LSM 510 confocal laser scanning microscope. The nitrocellulose layers were imaged individually in TBS; the slices were clamped between two microscopy slides to flatten them for automated imaging. Individual images were taken of each spot using a 10EC Plan Neofluar objective (N.A. 0,3, open pinhole). The microscope settings were kept constant to image all spots in a given array. The fluorescence images were analyzed with MATLAB (The Mathworks Inc.) and ImageJ (Rasband, W., National Institute of Health).

(75) The signal was calculated from the mean intensity of a circular area, 200 m in diameter, centered over the fluorescent spot. The background was the average signal from at least three negative control spots (0 pM of the antigen), where the intensity of each spot is the mean of the circular area. The signal-to-background for the volume dependency experiments was calculated by dividing the average signal from three replicates by the average of the negative controls. For all other experiments we additionally performed unity-based normalisation; we subtracted the average intensity of the negative control from the signal and divided by the difference between the average maximum for that experiment and the average negative control. For the dose response curves all spots from the experimental repeats were averaged before performing normalisation.

(76) TABLE-US-00001 List of reference numerals 1 Device for analysing liquid samples 111 Sample layer 112 Test site 113 Barrier region 114 Sample channel 115 Top sample layer 116 Second sample layer 116a Bottom sample layer 117a Upper sealing part 117b Lower sealing part 118 Bottom plate 119 Stack of sample layers 120 Frame 121 Clamp or spring-loaded tension lock 122a Upper sealing part opening 122b Lower sealing part opening 123 Outlet 2 Inlet part 211 Inlet channel 212 Reservoir section 213 Connecting section 214 Top plate 215 Reservoir part 216 Connecting part 217 Top plate opening 218 First opening 219 Second opening 220 Angled section 221 Transition section 3 Separation membrane 4 Hydrophobic membrane 411 Hole 5 Air passage 511 First air passage 512 Second air passage 6 Optical unit 611 Light source 612 First optical fibre 613 Second optical fibre 614 Photo detector w Width h Height Angle d.sub.1 First diameter d.sub.2 Second diameter p Plane p.sub.1 First plane p.sub.2 Second plane D.sub.1 First centre-to-centre distance D.sub.2 Second centre-to-centre distance