Detecting single-abasic residues within a DNA strand immobilized in a biological nanopore using an integrated CMOS sensor

Abstract

In this paper, we demonstrate the application of a novel current-measuring sensor (CMS) customized for nanopore applications. The low-noise CMS is fabricated in a 0.35μm CMOS process and is implemented in experiments involving DNA captured in an α-hemolysin (α-HL) nanopore. Specifically, the CMS is used to build a current amplitude map as a function of varying positions of a single-abasic residue within a homopolymer cytosine single-stranded DNA (ssDNA) that is captured and held in the pore. Each ssDNA is immobilized using a biotin-streptavidin linkage. Five different DNA templates are measured and compared: one all-cytosine ssDNA, and four with a single-abasic residue substitution that resides in or near the ~1.5nm aperture of the α-HL channel when the strand is immobilized. The CMOS CMS is shown to resolves the ~5Å displacements of the abasic residue within the varying templates. The demonstration represents an advance in application-specific circuitry that is optimized for small-footprint nanopore applications, including genomic sequencing.

1. Introduction

The original Sanger dideoxy enzymatic sequencing technique was presented in 1977 [1]. Sequencing methods developed since Sanger sequencing include pyrosequencing [2], oligonucleotide hybridization [3] and optical mapping [4]. Following the completion of the human genome assembly in 2001, advances in genomics research and the promise of personalized medicine have motivated researchers and companies to pursue more cost and time-efficient genome sequencing methods. Nanopores offer great promise as a next-generation sequencing method [5], with the potential for long read lengths (up to 100 kilo-bases) and without the need for optics, fluorescent labeling or amplification of the molecular sample. Nanopore sequencing aims to identify DNA bases by detecting base-specific electrical current signatures that are generated when ssDNA passes through a nanopore channel, driven electrophoretically by an applied trans-channel voltage. Here, the nanopore channel is a nanometer-scale aperture that forms naturally a protein channel in a lipid membrane (the biological pore), or is engineered by drilling or etching the opening in a solid-state substrate, such as silicon nitride, aluminum oxide or graphene (the solid-state pore) [6]. While solid-state pores offer superior scalability and stability, biological pores enable high precision measurements owing to their atomically precise geometries. Most research in nanopore-based DNA analysis has been performed using the α-homylsin (α-HL) nanopore [7–9], which has a limiting aperture of ~1.5nm, just larger than the width of ssDNA. For sequencing DNA, α-HL [9, 10] and an alternative biological nanopore MspA [11] are being pursued.

Advances in some sequencing technologies have been driven by the integration of new and more efficient sensors for measuring the sequence. For instance, the Ion Torrent sequencing device eliminated the need for optics to track sequencing progress; instead, a voltage-measuring sensor detects the potential change caused by pH variation (≈58mV/pH) during enzyme-catalyzed synthesis, using an ion-sensitive field-effect transistor [12]. By contrast, nanopore sequencing requires a current-measuring sensor (CMS) to monitor the ionic current change in the picoampere range through the nanopore. Toward identifying the current sensitivity required for sequencing in a nanopore, researchers have examined current amplitude maps while varying the nucleotide identities and positions within individual DNA substrates held fixed in the α-HL nanopore [13, 14]. One study utilized DNA polymerase bound to each DNA, to hold DNA in a known position to within single nucleotide (5Å) precision, and built an amplitude map while varying the position of a three-abasic DNA segment by single nucleotide steps [15]. The three-abasic DNA placement inside the α-HL nanopore generated current amplitudes that spanned the range of 24 to 36pA in a 0.3M KCl buffer solution. In any nanopore sequencing platform, the CMS must exhibit ultra-low noise to discern base-specific current signatures that can register for each 5Å-displacement of DNA during enzyme-catalyzed motion [9, 11]. In addition to low-noise performance, miniaturization of the CMS is a critical issue toward the development of portable genetic analysis devices. Although commercially available instruments such as the Axopatch 200B [16] and the Apollo [17] provide the nanopore CMS capability while achieving close to the theoretical limit in low-noise performance, their bulky bench-top size would not be useable in a commercial nanopore sequencing device. It is reported that we could sequence an entire human genome (≈3 billion bases pairs) with 50-fold coverage in one hour if the nanopore can be scaled up to an array of 100,000 individually addressed nanopores operating in parallel [6]. Therefore, it is imperative to shrink the CMS volume in order to increase the number of CMSs in a multichannel nanopore device for high-throughput DNA acquisition and analysis.

Submicron CMOS process technology makes it possible to drastically miniaturize the CMS. Due to the space minimization, we can reduce connection cablings and other parasitic capacitances, both of which lower the measurement bandwidth and cause noise and interference during nanopore sensing, thereby improving overall electrical performance of the CMS. For this reason, many research groups have developed low-noise CMOS CMSs toward the advancement of nanopore sequencing devices [18–21]. Using these platforms, researchers demonstrated detection of DNA translocation blockades in the hundreds of picoamperes to nanoamperes range. In this work, we explore a much smaller amplitude range, toward achieving the sensitivity required for sequencing. Specifically, we design and test a low-noise CMOS CMS for discerning small picoampere current changes that are caused by distinct DNA captured in the α-HL nanopore. In testing, we generate a current amplitude map while varying a single-abasic residue position within otherwise homopolymer DNA. The resulting map shows that the CMS is sensitive to 5Å displacements of DNA, which register in modern nanopore sequencing methods [9, 11].

2. Material and methods

2.1 Electrical configuration of nanopore device

Fig. 1a shows an electrical configuration of the nanopore device used in this experiment. A single α-HL nanopore is inserted into a lipid bilayer that separates two compartments (cis and trans chambers) containing electrolyte solution (1M KCl buffer at pH 8). The cis-opening of the pore has a head with a diameter of 2.6nm and a vestibule with a diameter of 3.6nm, which can accommodate a double-stranded DNA [22]. The pore opening tapers to a limiting aperture with a 1.5nm diameter, through which single-stranded polynucleotides can pass [15]. Once a command voltage V_CMD is applied between the cis and tran chambers (Fig. 1a), the potassium K⁺ and chloride Cl⁻ ionic current passes through the nanopore. The ionic current flow I_ION is mainly impeded by the pore size, and thus the nanopore resistance R_N is defined by V_CMD/I_ION. For instance, a V_CMD of 160mV applied across the membrane generates an open channel ionic current of 60pA in a 0.3M KCl buffer solution, resulting in R_N of 3GΩ [15]. The open channel current and resulting channel resistance value varies with ionic solution concentration, with higher current and lower resistance as the ion concentration is increased [23]. As ssDNA is captured into the α-HL pore, the ion channel path gets narrow, leading to the attenuated ionic current by |ΔI_ION|. At this moment, the nanopore resistance is increased by ΔR_N=(V_CMD×|ΔI_ION|)/[(I_ION−|ΔI_ION|)×I_ION]. As shown in this work, the ΔI_ION value varies with displacements of a single-abasic DNA positions in the α-HL nanopore. The ionic current shift in the picoampere range is amplified by a CMOS CMS that we designed. The signal is sampled through a digitizer (Axon Digidata 1440A) [16] or a field-programmable gate array (FPGA) then stored and analyzed on a desktop PC.

Fig. 1.

(a) Electrical configuration of the nanopore device used in this experiment, where the digitizer receives a control command from the computer and then provides a command voltage to the CMOS CMS. (b) Schematic of the CMOS CMS. The output buffer and 5^th-order Bessel LPF are not illustrated. (c) Schematic of self-biased differential amplifier used in the CMS.

2.2 CMOS current-measuring sensor design

In order to accurately monitor a DNA translocation event through the nanopore channel and convert its minute current to a readable voltage range suitable for digitization, a high-sensitivity transimpedance amplifier (TIA) using a capacitive feedback [18, 21] or a resistive feedback [19] should be adopted as a headstage of the CMS. The capacitive-feedback TIA, usually known as integrator, has better low-noise performance than the resistive-feedback TIA because a capacitor doesn’t have a thermal noise source. However, the integrator (discrete time system) requires a periodic reset pulse to prevent an output common-mode level from saturating. The full-scale reset pulse on a CMOS switch in the headstage often induces a channel charge injection and a clock feedthrough to the input node connected with a nanopore. As a result, this residual charge injected to the nanopore can result in a command voltage fluctuation during measuring. This integrator also needs a low phase-drift clock source and a reset pulse generator, leading to hardware complexity. Therefore, the resistive-feedback TIA enabling continuous time recording and hardware simplicity is preferred as a headstage of the CMS [19] and in this work we adopt the resistive-feedback TIA for the headstage.

The CMOS CMS schematic is shown in Fig. 1b, where an instrumentation-amplifier topology is chosen by virtue of its high common-mode rejection and consequent immunity to interference [24]. This CMS is comprised of three stages: 1) a headstage that has a transimpedance gain of 146dBΩ, 2) a programmable-gain difference amplifier (PDA) that can select a voltage gain of 32dB, 23.5dB, 14dB or 0dB, and 3) a unity-gain difference amplifier which changes differential inputs to a single-ended output. Owing to the instrumentation-amplifier architecture, the headstage and the PDA, composed of the core amplifiers Amp1, Amp2, Amp3 and Amp4, have the symmetric structure, which helps to reduce an input-referred offset voltage that often restricts output-voltage dynamic ranges [24]. If the core amplifier Amp1 has a high gain, its inverting input, connected to the cis chamber, would follow the V_CMD variation applied to the non-inverting input. The potential difference across the membrane is maintained constant, which gives rise to an electric field. For this reason the CMS designed in this work can be referred as a potentiostat. Due to the electric field, a negatively charged ssDNA molecule in the cis chamber is driven toward the positive trans-side and captured into the nanopore.

To achieve the high-gain core amplifier, we adopted a self-biased differential amplifier employing a current reuse technique [25], illustrated in Fig. 1c. This architecture enables a high gain because both of the NMOS and PMOS input transistors contribute to the effective transconductance (g_m). Thus, the core amplifier has an open-loop gain of (g_m2+g_m4)×(R_O3||R_O5), where R_O is an output resistance, and an input-referred noise voltage of

$$\overline{V_{n,\mathit{AMP}1}^2(f)}=\underset{1/f-\mathit{noise}}{\underbrace{\left(\frac{1}{W_2{L}_2}+\frac{1}{W_4{L}_4}\right)·\frac{2K}{f·C_{ox}}}}+\underset{\mathit{Thermal}\mathit{noise}}{\underbrace{\frac{8kT\gamma }{g_{m2}+g_{m4}}}}.$$

Here, k, T, K and γ are Botzmann’s constant, absolute temperature, and process-dependent 1/f noise constant and coefficient, respectively. By employing large-width PMOS and NMOS transistors for input pairs, we can not only increase g_m, but also suppress the 1/f noise in the above equation.

During nanopore sensing, a RMS input-referred noise current of the CMOS CMS shown in Fig. 1b can be approximately expressed as [26]

$$I_{n,\mathit{RMS}}\approx \sqrt{\frac{1}{BW}{\int }_{0.1}^{BW}{\left[\frac{1}{R_F}+\frac{1}{R_N}+2\pi f(C_P+C_N)\right]}^2\times \overline{V_{n,\mathit{AMP}1}^2(f)}df+\frac{4kT}{R_F}}.$$

Here, R_F is the feedback resistor of the headstage, R_N is the nanopore resistance, C_N is the nanopore capacitance, C_P is an input parasitic capacitance, and BW is the bandwidth. Since R_N (≥1GΩ) is much higher than R_F (≈20MΩ) used in this design, the 1/R_N is negligible. This noise equation shows one zero term of 1/[2π×(C_P+C_N)×R_F], which induces a high-frequency noise injection along with the core amplifier input-referred noise V²_n,AMP1. To restrict the bandwidth and reduce the high-frequency noise, we thus used an off-chip 5^th–order Bessel low pass filter (LPF) with a bandwidth of 10KHz after the CMOS CMS output buffer.

2.3 Single-abasic DNA molecule templates

In this work, we examine five different 40-mer ssDNA oligonucleotides. All are homopolymer cytosine, with one all-nucleotide ssDNA and four with a single-abasic (1′, 2′-dideoxy) residue at a prescribed position. The 5′ end of each ssDNA is biotinylated. Each oligonucleotide is mixed in a one-toone ratio with Streptavidin that is a protein secreted by Streptomyces avidinii bacteria, to form a 5′ biotin-streptavidin linkage. Capture by the 3′ end results in an immobilized complex, which can be used to examine the impeded current amplitude for each for the five oligonucleotides. The same approach was used in [13, 14] with homopolymers using varying nucleobase substitutions, instead of abasic substitutions. The five oligonucleotides are aligned and displayed in Fig. 2a. In each ssDNA, X indicates the position of the abasic residues. We reference each abasic configuration with the notation 1_ab(x), where x is the distance (in nucleotides) of the abasic residue measured from the biotin (n=0) position. In this study, we create a current map at 5Å resolution (i.e., single nucleotide resolution) using the four ssDNA defined by 1_ab(7), 1_ab(8), 1_ab(9) and 1_ab(10), and compare it to the all-cytosine ssDNA. Fig. 2b illustrates the chemical structures of standard DNA nucleobases and abasic residue. Since the abasic residue occupies less space than a nucleobase, abasic residues positioned near the pore lumen are known to decrease resistance and thus boost the current amplitude [15].

Fig. 2.

(a) DNA oligonucleotides used in this experiment. In each sequence, X indicates the position of the single-abasic (1′, 2′-dideoxy) residue positioned within the otherwise cytosine homopolymer. (b) Chemical structures of abasic and standard DNA residues, where the standard residues show a nucleobase at the 1′ position represented here as a purine. (c) Illustration of Streptavidin-Biotin-DNA complex captured and immobilized in the nanopore. Oligonucleotides 1_ab(8) and 1_ab(9) shown in (a) position the abasic residue in the α-HL nanopore aperture, the most sensitive location of the channel.

Fig. 2c displays an illustration of a streptavidin-biotin-DNA complex captured in the α-HL nanopore. The distance n from the 1^st to the 15^th template base along the strand is indicated. Depending on the position of the single-abasic residue inside the nanopore, the ionic current variation may be different than the all-nucleotide ssDNA. Single-abasic residues are used to create a more precision position map than is possible with three-abasic residues, though larger current variations can be detected (i.e., it is easier to detect the current differences) with more abasic residues [15]. We also consider the low noise level required to discern 5Å (single nucleotide) displacements of the single-abasic residue in the nanopore. These experiments were done in 1M KCl buffer with 10mM HEPES, 1mM EDTA, a pH of 8.00±0.05 and at a temperature of 23C. The 40-mer 5′-biotinylated oligonucleotides mixed in a 1:1 ratio with Streptavidin were concentrated to 1μM in the cis chamber.

3. Results and discussion

The CMS illustrated in Fig. 1b was fabricated in a 0.35μm CMOS process and packaged using a Quad LPP plastic that offers a minimum space (5mm width×5mm length×0.85mm height), as shown in Fig. 3. All experiments in this work were performed in a Faraday cage to shield 60Hz-noise induced from building ground.

Fig. 3.

Photo of the CMOS CMS package which has a volume of 21.25mm³, and compared with a US quarter dollar coin (Micrograph of the chip is also shown in the right side, where the dashed line indicates the CMOS CMS used in this work).

We first measured the CMOS CMS gain, which has an average gain of 261438064.9Ω (≈168.34dBΩ) in the V_CMD range of −200mV to +200mV, and then estimated standard deviations from the input current signals, displayed in Figs. S1 (supporting information), for the CMOS CMS input-referred noise and the background noise including the CMS, recording electrode, nanopore and ionic solution. As a result, the CMOS sensor and the background noises have input-referred noises of 4.07pA_RMS and 5.82pA_RMS in a bandwidth of 10KHz, respectively. Using an off-line 8^th-order Bessel LPF, we further restricted the bandwidth to 1KHz, so that the background noise was reduced to 2.11pA_RMS.

Electrical performance of the CMOS CMS designed in this work is summarized in Table 1 and also compared with the prior studies presented in [18, 19, 21]. As mentioned in the section 2.2, CMSs using the capacitive feedback as a headstage [18, 21] have relatively better noise performance than our CMS using the resistive feedback. Interestingly, the CMS described in [19] shows the best noise performance even though they employ the resistive feedback. This is principally due to advanced fabrication process (0.13μm CMOS technology) they used and high power dissipation (5mW). On the other hand, our CMS targeting for a future portable analytic device, in which low power consumption would be critical, dissipates the lowest power (0.5mW), which is ten times lower than [19], owing to the current reuse technique adopted in core amplifiers. Instead it raises the input noise up to 4.07pA_RMS that is four times higher than [19]. In the future, we will be able to further decrease the input noise by compromising with power consumption and using an advanced CMOS technology. In addition to the low-power feature, the proposed CMOS CMS has 2.6 times higher gain than [19].

Employing the CMOS CMS, we measured the open-channel currents on the α-HL nanopore, which is used to estimate the nanopore conductance. Fig. S2 (in the supporting information) shows the conductance graph with 1.08nS in the positive V_CMD and 0.72nS in the negative. Here, this conductance deviation in the positive and negative V_CMD arises from the asymmetric structure of the α-HL nanopore [27]. In the case where a V_CMD of +120mV is applied to α-HL pore, the measured conductance is approximately 1.04nS, which results in a R_N of 961MΩ.

Next, to create a mapping of current level as a function of single-abasic position in α-HL, 1μM of streptavidin-bound homopolymer (without a single-abasic residue) was added to the cis-side chamber. Current amplitudes acquired when the all-DNA template is captured in the α-HL served as the baseline all-DNA current amplitude, which is compared with current shifts due to the four different abasic-bearing DNA templates. As soon as a complex was captured at a V_CMD of 120mV, our finite-state machine logic executed on an FPGA holds the captured complex for 2.5s, and then ejects it back into the cis chamber by reversing the V_CMD polarity to −50mV, as illustrated in Fig. 4. This is regarded as one event and was repeated to obtain a distribution of current levels for each strand tested. In this work, we conducted four experiment sets to measure current amplitudes corresponding four abasic-bearing DNA molecules (1_ab(7), 1_ab(8), 1_ab(9) and 1_ab(10)). First, we added 1μM of all-DNA and after detecting tens of events added 1μM of abasic-bearing DNA under hundreds of events were recorded. The cis chamber was then perfused, and the steps were repeated but using a different abasic-bearing DNA. This was repeated until all DNA were measured.

Fig. 4.

Illustration of capturing oligonucleotides with the CMOS CMS, (i) pore insertion, (ii) open-channel current, (iii) Streptavidin-Biotin-DNA complex capture event, (iv) held complex, (v) ejection of complex back into the cis chamber, and (vi) return to capture voltage and open channel current.

Fig. 5a displays the populations corresponding to each position in the nanopore. The length of α-HL is ~20 nucleotides of ssDNA and our mappings considered the first 15 of those positions inside α-HL, starting from the cis opening of the pore to the trans opening. As explained previously, the cis opening of α-HL nanopore has a diameter of 2.6nm that continues into larger vestibule approximately 1–7 nucleotides in length. As shown in previous studies [15], abasic residues in the wider vestibule do not measurably affect the current. We confirmed that templates 1_ab(x), for x =1–7, showed no statistically significant difference in current when compared to the baseline amplitude level (gray dots in Fig. 5a). Populations of standard DNA and 1_ab(7) templates have a current level of 21.86±0.84pA on average. The value for R_N is increased from 961MΩ to 5.49GΩ when the templates are captured in the pore. Templates of 1_ab(x) with x = 11–15 have the abasic residue positioned in the β-barrel area, a cylindrical domain with a diameter of ~2.2nm, and generate a relatively uniform amplitude that is slightly higher than that of the all DNA template (gray dots in Fig. 5a). The β-barrel space has higher sensitivity than the vestibule area, and like the limiting aperture of the pore can accommodate ssDNA but not duplex DNA. However, it does not enough sensitivity to discern single-abasic residue displacements within the immobilized DNA strand. At positions corresponding to 1_ab(8), 1_ab(9) and 1_ab(10), we span the most limiting region within α-HL, referred to as the aperture. The aperture has a diameter of ~1.5nm and can only accommodate ssDNA[22]. Consistent with other studies [13, 14], our experimental results show that the highest average current levels are shown for those positions. At position 8, the average current amplitude of 1_ab(8) templates is approximately 25.44±0.92pA, which is 3.58pA higher than the baseline DNA template and also has a R_N of 4.72GΩ. At the position 9, template of 1_ab(9) has the highest current level on average measured as 26.69±0.867pA (≈4.50GΩ). From position 10, the current amplitude starts decreasing. The average current of 24.73±1.06pA (≈4.85GΩ) for 1_ab(10) is lower than the current levels at positions 8 and 9, and slightly higher than the baseline current. This is because position 10 is close to the β-barrel area.

Fig. 5.

(a) Data show the mean current of each captured DNA held for 2.5s in experiments with the all polyC ssDNA (gray dots) and one of the abasic-residue bearing ssDNA molecules, measured using the CMOS CMS. Dashed lines mark steps of perfusion of DNA from the cis chamber followed by addition of the next set of DNA. (b) Mean and standard deviation for populations for DNA 1_ab(7), 1_ab(8), 1_ab(9) and 1_ab(10), mean values 21.86pA, 25.44pA, 26.69pA and 24.73pA, respectively. The mean of 1_ab(7) doesn’t show a difference from the all nucleotide DNA, and is here used as a baseline and compared with other current levels of 1_ab(8), 1_ab(9) and 1_ab(10). (c) Histogram of measured current populations.

Fig. 5b illustrates a current mapping at the position 7, 8, 9 and 10, respectively. From these results, we can conclude that it would be possible to detect single-nucleotide translocation steps reported by motion of a single-abasic residue within ssDNA through the α-HL channel using our CMOS CMS. To confirm and compare the CMS generated map, we also generated a current map measured with an Axopatch 200B bench-top amplifier. Fig. S3 (in the supporting information) shows the measurement results in an experiment in which the all-DNA template was first added, followed by 1 μM addition of each of the abasic-bearing templates in stages. Each new template generates a distinct level that has a tighter amplitude distribution than is possible with our CMOS CMS, due to the higher input noise (4.07pA_RMS). For a direction comparison, the Axopatch 200B generated the following: 29.33±0.84pA for template 1_ab(7), 32.91±0.93pA for template 1_ab(8), 34.12±0.87pA for template 1_ab(9) and 32.12±1.04pA for template 1_ab(10). Both amplifiers generate a consistent mapping for positions 8 and 9 which span the most sensitive region in the α-HL nanopore.

In future work, we will reduce the CMS input-referred noise current, with the objective of achieving at most ~1pA_RMS, so that tracking motion of the abasic residue through positions 6–10 would be possible, as with the Axopatch 200B (Fig. S3b). According to Eq. 2, to lessen the overall input noise, we have to reduce the nanopore and input parasitic capacitance, increase the feedback resistor R_F and design an ultra-low noise core amplifier, such as the design described in [28]. Although the Axopatch presently exhibits lower noise at recording bandwidths, only integrated systems such as our CMS can scale for multi-channel nanopore implementations, which will be necessary for efficient genomic sequencing [29].

4. Conclusion

A low-noise CMOS current-measuring sensor has been designed and its sensitivity tested using different individual ssDNA held in an α-hemolysin nanopore channel. Each ssDNA is immobilized in the α-hemolysin channel by a biotin-streptavidin linkage such that an ionic current amplitude map is generated for different position of a single-abasic residue within an otherwise homopolymer cytosine ssDNA. The CMOS sensor achieves detection of 5Å displacements of the residue for the three nucleotide positions closest to and within the ~1.5 nm aperture of the pore. The potentiostat sensor was fabricated in a 0.35μm 4M2P CMOS process, and operating in ±1.5V has a gain of 168.34dBΩ, an input-referred noise of 4.07pA_RMS and consumes a power of 502μW. Using the CMOS sensor, an average nanopore resistance of 1.1GΩ and a background noise of 5.82pA_RMS were measured in experiments at 10 kHz bandwidth. Future designs will aim to reduce noise to 1pA_RMS at 10 kHz, toward achieving the required amplitude sensitivity for sequencing while still preserving sub-millisecond detection of DNA motion under enzyme control [11, 15]. A recent significant work [19] reduced high-frequency noise in a solid-state nanopore setup by reducing the input parasitic capacitance, making it to possible to detect capture events for very short (25 bp) dsDNA through the pore at 500 kHz bandwidth. It is plausible that by reducing noise of our CMS to 1pA_RMS at 10 kHz, and using the novel setup in [19], our platform can advance biological nanopore sequencing by permitting low (kHz) and high (MHz) bandwidth recording options that achieve signal-to-noise ratios that are not possible with commercially available amplifiers.

Table I.

Electrical performance summary and comparison of CMSs

	[18]	[19]	[21]	This work
Headstage Type	Capacitive	Resistive	Capacitive	Resistive
Power dissipation	1.5mW@3.3V	5mW@1.5V	23mW@3.3V	0.5mW@3V
Input noise in a bandwidth	1.35pARMS in 10KHz	1.00pARMS in 10KHz	0.15pARMS in 1KHz	4.07pARMS in 10KHz
Transimpedance gain	N/A	100MΩ	N/A	261MΩ (Measured)
DC input offset cancellation	No	No	Yes	Yes
Active area	0.139mm2	0.2mm2	0.5mm2	0.103mm2
Process	0.5μm 3M2P CMOS	0.13μm CMOS	0.35μm 4M2P CMOS	0.35μm 4M2P CMOS

Biographies

Jungsuk Kim received the B.S. degree (magna cum laude) in electrical engineering from Sogang University, Seoul, Korea, in 2003, the M.S. degree in electrical engineering from University of Southern California, Los Angeles, in 2006, and the Ph.D from the University of California, Santa Cruz, in 2011. After graduating in 2003, he worked at DRAM Design Lab, Samsung Electronics as an engineer. He is currently working as a postdoctoral scholar in computer engineering at the University of California, Santa Cruz. His research interests include integrated circuit and system designs for nanopore sequencing technology and other biomedical applications.

Raj Maitra received the B.S. degree in Biomolecular Engineering from University of California, Santa Cruz in 2010. In 2011, he worked at the UCSC Biocontrol Laboratory as a research staff. He is currently working towards a MS degree in computer engineering at the University of California, Santa Cruz. His research interests include bioMEMS device design, control algorithms for single-molecule biophysics.

Kenneth Pedrotti received the B.S. degree in engineering physics in 1978 and the M.S. degree in quantum electronics in 1979 from the University of California at Berkeley, and the Ph.D. degree in electrical engineering in 1985 from Stanford University, Stanford, CA. Currently, he is a Professor and Chairman of the Department of Electrical Engineering at the University of California at Santa Cruz. His interests include devices and circuits for optical communication networks, RF, biomedical applications, energy scavenging and imaging. From 1998 to 2000, he was with the Rockwell Science Center, Thousand Oaks, CA, working on mixed-signal VLSI for visible and IR imaging. Prior to that, he worked from 1997 to 1998 for Rockwell Semiconductor Systems, Newbury Park, CA, on integrated circuits for optical communications. From 1985 to 1997, he was with Rockwell Science Center.

William B. Dunbar received the B.S. degree in engineering science and mechanics from the Virginia Polytechnic Institute and State University in 1997, the M.S. degree in applied mechanics and engineering science from the University of California, San Diego, in 1999, and the Ph.D. in control and dynamical systems from the California Institute of Technology in 2004. He is currently an Associate Professor in the Department of Computer Engineering at the University of California, Santa Cruz. His research is focused on applications of feedback control to problems in biophysics and sequencing with nanopores.