Case studies in neuroscience: reversible signatures of edema following electric and piezoelectric craniotomy drilling in macaques

Abstract

In vivo electrophysiology requires direct access to brain tissue, necessitating the development and refinement of surgical procedures and techniques that promote the health and well-being of animal subjects. Here, we report a series of findings noted on structural magnetic resonance imaging (MRI) scans in monkeys with MRI-compatible implants following small craniotomies that provide access for intracranial electrophysiology. We found distinct brain regions exhibiting hyperintensities in T2-weighted scans that were prominent underneath the sites at which craniotomies had been performed. We interpreted these hyperintensities as edema of the neural tissue and found that they were predominantly present following electric and piezoelectric drilling, but not when manual, hand-operated drills were used. Furthermore, the anomalies subsided within 2–3 wk following surgery. Our report highlights the utility of MRI-compatible implants that promote clinical examination of the animal’s brain, sometimes revealing findings that may go unnoticed when incompatible implants are used. We show replicable differences in outcome when using electric versus mechanical devices, both ubiquitous in the field. If electric drills are used, our report cautions against electrophysiological recordings from tissue directly underneath the craniotomy for the first 2–3 wk following the procedure due to putative edema.

NEW & NOTEWORTHY Close examination of structural MRI in eight nonhuman primates following craniotomy surgeries for intracranial electrophysiology highlights a prevalence of hyperintensities on T2-weighted scans following surgeries conducted using electric and piezoelectric drills, but not when using mechanical, hand-operated drills. We interpret these anomalies as edema of neural tissue that resolved 2–3 wk postsurgery. This finding is especially of interest as electrophysiological recordings from compromised tissue may directly influence the integrity of collected data immediately following surgery.

INTRODUCTION

Research in nonhuman primates (NHPs) is instrumental for probing fundamental questions about the nature and mechanisms of brain function related to perception, action, and cognition. Electrophysiological investigations in NHPs require surgical procedures that enable direct access to the brain to target neuronal populations of interest. Techniques to streamline such surgical procedures are continuously developed to optimize the well-being and experimental outcomes of the animal subjects.

Drilling is a core component of the surgical procedures to provide access to neural tissue for electrode placement. Manually operated tools such as trephines and hand drills (Fig. 1A) are widely used and require specific technique and skill of the surgeon to ensure that only skull tissue is drilled, and that surgical tools do not compromise the integrity of the dura mater and the underlying brain tissue. In extreme cases, accidental lowering of the drill into tissue, also termed “plunging,” may have severe consequences such as hematomas or irreversible brain lesions (1–3), though these tools can often be equipped with safety stoppers to limit plunging (Fig. 1A). In addition, electric counterparts of the hand drill are commonly used, as they enable the surgeon to expedite the drilling process. In human surgery, perforator-style automatic drills have been designed to seize upon complete drilling of skull tissue, reducing the possibility of plunging events and clearing bone shavings from the site of drilling (2, 4) (cf. Ref. 3). In addition, nonperforator high-speed electric or pneumatic drills are also regularly used in clinical practice. Furthermore, manual alternatives, such as hand drills and trephination are in practice bedside or in emergency events (5). Electric drills typically used in NHP surgeries are of a different design (see Fig. 1B). Similar to dental drills, they utilize diamond-shaped burrs, and require a skilled surgeon to avoid plunging events. Finally, piezoelectric drills have been introduced as a new class of drills. They utilize ultrasonic vibrations that enable selective removal of hard surfaces (i.e., bone) without harming soft tissue (i.e., the dura or brain tissue). The piezoelectric drill, originally developed for oral surgeries (6, 7), has been adopted for pediatric cranial surgery and has been recommended for NHP craniotomies, as it affords the surgeon a procedure that is less prone to harm brain tissue (8–13). Although different drilling tools have been historically tested in animals and refined in clinical practice for humans, usage in NHP research is quite heterogeneous and may depend on specific laboratory practices or the surgeon’s preferences.

Figure 1.

Surgical tools. A: the mechanical drill is used to create 4.5-mm diameter holes in the skull. A sharp-tipped drill (left) is equipped with a safety stopper (right) of an appropriate length to reduce the possibility of plunging. Drilling is conducted in a twisting motion perpendicular to the skull surface. After the tip of the sharp drill reaches the dural surface (forming a conical hole in the skull of max width 4.5 mm), it is replaced with the blunt-tipped drill (middle). The blunt-tipped drill is used to remove all remaining skull within the 4.5 diameter hole. B: the electric drill is controlled using a foot pedal and is equipped with a diamond-shaped burr that is maximally efficient at drilling laterally. Drilling is conducted in a circular motion to cover the full area of the intended craniotomy, incrementally removing thin layers of skull either until dura is visible (electric-only drilling) or until the surgeon decides to switch to piezoelectric drilling (electric followed by piezoelectric drilling). C: the piezoelectric drill is operated using a foot pedal and includes an irrigation system to limit tissue heating. Inserts designed for osteotomy (PL1 or PL2) are used to drill in a circular manner that covers the full area of the craniotomy to incrementally remove skull tissue. The Endo setting is used when drilling bone close to the dural surface.

Here, we present a series of magnetic resonance imaging (MRI) scans following 10 craniotomy surgeries with different surgical drilling tools in NHPs implanted with MRI-compatible materials. We found an unexpected anomaly in scans that were collected within two to three weeks of craniotomies conducted with electric or piezoelectric drills. Specifically, hyperintensities in the vicinity of the craniotomies were noted on T2-weighted scans and were interpreted as edemas of neural tissue. These anomalies were not accompanied by behavioral abnormalities, or any other clinical signs. Our findings have potential implications for intracranial recordings from the tissue directly underlying a craniotomy immediately following a procedure that used an electric drill.

METHODS

Subjects, Inclusion Criteria, and Surgical Procedures

Data from eight male macaques (6 Macaca mulatta, 2 Macaca fascicularis) were included in the study. All procedures were approved by the Princeton University Animal Care and Use Committee and conformed to the National Institutes of Health guidelines for the humane care and use of laboratory animals. All monkeys were subjects in electrophysiological studies requiring cranial implants and craniotomies to provide access to regions-of-interest in the brain. Each animal had one or two surgeries and was included in the study if they participated in a craniotomy surgery followed by an MRI scanning session within 1 mo postsurgery. Each craniotomy surgery was denoted with a separate case number (Table 1).

Table 1.

Ten cases involving eight animals following craniotomy surgeries with one or more drilling tools, weight and age at time of surgery, the number of days following the surgery that a scan was conducted, scanner used, and whether T2 hyperintenstities were found

No.	M	Weight	Age	No. of Cran	Elec	Piezo	Mec	ΔScan	Scanner	Anomalies
1	P	12.2	7	3	+	+		7	Skyra	Yes
								14	Prisma	No
2	R	8.1	8	5		+		7	Prisma	Yes
3	V	8.1	6	3	+			6	Skyra	Yes
4	Ph	6.9	9	1		+		22	Skyra	No
5	Ph	6.1	7	3			+	12	Prisma	No
6	M	7.6	7	3			+	5	Prisma	No
7	M	7.7	8	3			+	3	Skyra	No
8	F	6.5	4	1			+	6	Skyra	No
9	Mc	7.6	14	2			+	8	Skyra	Yes
10	B	5.2	19	1			+	4	Skyra	No

Note that case 1 had two scans within 1 mo postsurgery. Elec, electric drill; M, monkey; Mec, mechanical drill; No. of Cran, number of craniotomies; Piezo, piezoelectric drill; ΔScan, number of days postsurgery that the scan was conducted.

All surgical procedures were performed under general anesthesia with isoflurane (induction 2%–5%, maintenance 0.5%–2.5%) under strictly aseptic conditions. Vital signs, including heart rate, $\text{S}{\text{p}}_{{\text{O}}_2}$, respiratory rate, blood pressure, end-tidal carbon dioxide, and body temperature, were continuously monitored throughout the procedure. Animals were first implanted with 12–16 ceramic screws (Thomas Recording, Gieseen, Germany; or titanium screws, Fine Science Tools; Foster City, CA in cases 9 and 10). To affix screws to the skull, small holes were drilled (using manual, hand-operated drills) ∼4 mm deep, after which the holes were tapped, and the screws were placed. The screws were then affixed with surrounding dental acrylic (Yates Motloid Crosslinked Flash Acrylic; Elmhurst, IL), and/or bone cement (Palacos bone cement, PLACE; or DJO Surgical Cobalt HV bone cement, Austin, TX) to create a small implant for head-fixation and chamber placement (14). Using additional dental acrylic or bone cement, we affixed customized cylindrical plastic recording chambers (15–28 mm inner diameter) to the animal’s implant to enable access to regions-of-interest. There were cases when chamber placement was conducted in a separate surgery before the craniotomies (cases 1 and 8), in combination with craniotomy drilling (cases 2, 3, 5, 7, 9, and 10), or as part of the initial implantation and screw placement (cases 4 and 6). In all assessed cases, screw placement and craniotomy drilling were not conducted during the same surgery and were separated by at least 3 mo.

Small circular craniotomies (4.5–8 mm diameter) were drilled inside implanted chambers. Drilling was conducted in one of four ways: 1) using a mechanical hand-drill (Synthes; Fig. 1A) and respective stoppers to penetrate the skull to the dural surface (6 surgeries, 13 craniotomies; Table 1); 2) using an electric drill (RAMPOWER; Ram products Inc.; maximum operation of 45,000 rotations per min; Fig. 1C) with burrs of sizes 0.5–2.1 mm (Fine Science Tools) to remove an entire column of skull tissue leading to the dural surface, removing any thin bone debris using fine forceps (1 surgery, 3 craniotomies; Table 1); 3) using an electric drill to remove excess cement and to thin out skull tissue before using a piezoelectric drill (Mectron; 24–36 kHz operating frequency; Fig. 1C) to remove the skull above the dura using the lowest power setting (Endo) and continuous irrigation (1 surgery, 3 craniotomies; Table 1); 4) strictly using a piezoelectric drill with continuous irrigation for the full craniotomy procedure, starting with an intermediate drilling setting, to mitigate the lengthy drilling process on low settings, followed up by the lowest setting once the drill is close to the dural surface (2 surgeries, 6 craniotomies; Table 1). For all four approaches, drilling was conducted to remove all tissue in the entire target circular area (i.e., burr hole), with only minor fragments removed by fine forceps. All craniotomy surgeries concluded uneventfully with no visible damage to the dura or plunging events. Following each surgery, animals were monitored closely by facility veterinarians, animal care staff, and researchers. Following prescription by the veterinarians, animals were also administered postoperative medicine, including analgesics and antibiotics.

Neuroimaging Protocols

For scanning sessions, animals were sedated with ketamine (1–10 mg/kg im) and xylazine (1–2 mg/kg im), and provided with atropine (0.04 mg/kg im). Sedation was maintained with tiletamine/zolazepam (1–5 mg/kg im). The animals were then placed in an MR-compatible stereotaxic frame (1530 M; David Kopf Instruments, Tujunga CA). Vital signs were monitored with wireless ECG, pulse, respiration sensors (Siemens AG, Berlin), and a fiber optic temperature probe (FOTS100; Biopac Systems Inc., Goleta CA). Body temperature was maintained with blankets and a warm water re-circulating pump (TP600; Stryker Corp, Kalamazoo MI).

All animals had whole brain structural MRI data collected either on a Siemens 3-T MAGNETOM Skyra or on a Siemens 3-T MAGNETOM Prisma using a Siemens 11-cm loop coil placed over the head (see Table 1). T2-weighted volumes were acquired with a three-dimensional (3-D) turbo spin echo with variable flip-angle echo trains (3-D T2-SPACE) sequence (voxel size: 0.5 mm, slice orientation: sagittal, slice thickness: 0.5 mm, field of view (FoV): 128 × 128 mm, FoV phase: 79.7%, repetition time (TR): 3,390 ms, echo time (TE): 386 ms (Skyra) or 387 (Prisma), base resolution: 256 × 256, and acquisition time (TA): 17 min 41 s (Skyra) or 15:39 (Prisma). T1-weighted structural images were collected using the 3-D Magnetization-Prepared Rapid-Acquisition Gradient Echo (MPRAGE) sequence, voxel size: 0.5 mm, slice orientation: sagittal, slice thickness: 0.5 mm, FoV: 128 × 128 mm, FoV phase: 100%, TR: 2,700 ms, TE: 3.27 ms (Skyra) or 2.78 ms (Prisma), inversion time (TI): 850 ms (Skyra) or 861 ms (Prisma), base resolution: 256 × 256, and TA: 11 min 311 s.

All animals had scanning sessions conducted before their craniotomy surgeries for accurate positioning of craniotomy locations. These scans were conducted no sooner than 2 mo following initial implantation in all animals included in the study. The scans were examined by experienced personnel to rule out long-lasting anomalous findings due to screw placements or other complications before surgery and served as reference images. Further scans were conducted following the animals’ craniotomy surgeries either for clinical purposes (case 1, monkey P) or to confirm electrode localizations in brain regions-of-interest.

For animals with detected T2 hyperintensities, we used FSL (15) (6.0.5.1; FSLeyes 1.3.0) to define areas on the T2 volumes corresponding to the anomalies, and then calculated the approximate hyperintensity volume using the fslstats command.

RESULTS

Behavioral Signs

Exactly one week following his craniotomy surgery, monkey P was reported for inappetence and reduced social interactions with humans. Precautionary T1- and T2-weighted scans were conducted the following day to ensure integrity of the brain postsurgery. The other animals whose data were included in this report did not show behavioral abnormalities postsurgery.

Neuroimaging Findings and Their Resolution

Monkey P (case 1) had undergone three craniotomies in anterior, central, and posterior chambers of the left hemisphere, targeting three regions of interest for electrophysiological recordings from the frontal eye fields, pulvinar, and superior colliculus, respectively. Drilling had been done initially using an electric drill for each craniotomy. Once a thin skull layer was separating the drill from the dural surface, the surgeon switched to a piezoelectric drill set with sufficient irrigation and slow speed (see methods; Table 1, case 1). The T2-weighted scan taken 1 wk postsurgery showed two white matter hyperintensities ∼151 and 104 mm³ in size and located underneath the central and anterior chambers, respectively (Fig. 2A, top). The locations of these anomalies were confirmed using stereotaxic coordinates and by reference to anatomical landmarks in comparison with scans obtained preoperatively to be directly underneath the anterior and central craniotomies. No anomalies were visible underneath the posterior craniotomy. The anomalies were less salient on the T1 scan, but could be detected post hoc as hypointensities (Fig. 2A, bottom). Follow-up scans were acquired 1 wk later, before any electrode penetration through the dura was conducted, and revealed a complete resolution of the anomalies associated with both the anterior and central locations (see Fig. 2B). T2 hyperintensities are typically observed in edematous neural tissue along with hypointensive findings on T1-weighted scans (16) (see Fig. 2A, bottom). Although the T1/T2 differential intensity profile would not rule out an acute hemorrhage (17) or infection (18), the delineated localized hyperintensities on the T2 scan, the localized hypointensities on the T1 scan, and the fact that the anomalies were directly underneath the craniotomies and confined to their extent led us to interpret the anomalies as edemas of neural tissue caused by cranial drilling. The accidental finding in monkey P motivated a retrospective examination of scans from seven other animals, who underwent craniotomy surgery and had scans acquired within 1 mo postsurgery (see methods and Table 1). We were particularly interested to investigate whether these anomalies were associated with a specific drilling method.

Figure 2.

Magnetic resonance imaging (MRI) findings in monkey P. White matter T2 hyperintensities (T2w) found 7 days after surgery using electric and piezoelectric drilling in the vicinity of anterior (left; red arrows) and central (right; green arrows) craniotomies in case 1 (A, top). The same areas show hypointensities on a T1-weighted (T1w) scan acquired in the same session (A, bottom). These anomalies resolved on a follow-up scan conducted one week after the first (B). No recordings were conducted between the two scans. Approximate locations of chamber walls are presented on coronal slices using dotted purple lines. Coronal slice is denoted in millimeters with respect to Ear Bar Zero (EBZ). A, anterior.

Similar T2 anomalies were readily found during our retrospective examination of scans following surgeries utilizing electric and piezoelectric drilling (three surgeries in three animals to drill nine craniotomies; excluding case 1). Case 2 denotes a surgery that targeted five craniotomies (two in close proximity within an anterior chamber; three in close proximity within a central chamber) in the right hemisphere, which were all performed using piezoelectric drilling. The animal was scanned 1 wk postsurgery. Examination of the scans shows two extensive hyperintensities ∼127 and 251 mm³ in size below the central and anterior chambers, respectively (Fig. 3A). Accurate distinction of whether all craniotomies induced lesioning was not possible. The next MR scanning for this animal occurred more than three months later and only included a T1-scan. No trace of characteristic T1 hypointensities under the craniotomies were noted (data not shown). In case 3, three craniotomies were performed (two in an anterior chamber; one in a central chamber). All drilling in the third case was done exclusively using the electric drill. The T2-weighted scans showed defined hyperintensities ∼659 and 139 mm³ in size below the anterior and central chambers, respectively (Fig. 3B). Similar to case 2, it was difficult to delineate whether all three craniotomies induced lesions due to the spatial proximity of two of the craniotomies. Case 4 represents the only instance where no anomalous findings on T2-weighted scans were present following craniotomy surgery that used an electric/piezoelectric drill. However, in this case, the animal was not scanned until 3 wk following surgery. Based on the profile of the anomalies in our animals and their progression timeline observed in case 1, it is possible that, if an edema was induced during the surgery, it may have resolved at this point.

Figure 3.

Magnetic resonance imaging (MRI) findings in three additional cases. Similar T2 findings were found in three other cases. Case 2 (A; using the piezoelectric drill) and 3 (B; using the electric drill) had two regions underneath anterior (red arrows) and central craniotomies (green arrows), and case 9 (C; using the mechanical drill) had one anomaly underneath the anterior craniotomy (red arrows). Note susceptibility-induced signal drops along the length of the electrodes used for localization (A–C) as well as artifacts caused by titanium skull screws (C). Conventions as in Fig. 2.

Anomalous findings were rare following craniotomy surgeries conducted strictly with mechanical drills. There were six scans (across 5 animals) conducted within a week following these craniotomy surgeries, only one of which showed T2 hyperintensities (86 mm³; see Fig. 3C). All others showed no anomalous findings despite electrode penetration at the time of the scans, indicating that the observed findings were unlikely to be caused by acute electrode placement for trajectory localizations (e.g., Fig. 4, A and B; cases 6 and 7).

Figure 4.

Negative magnetic resonance imaging (MRI) findings postsurgery. Exemplar negative findings underneath craniotomies (red arrows) in T2-weighted scans within 2 wk from conducting craniotomy surgeries in cases 6 (A) and 7 (B). Both craniotomies were conducted using mechanical twist drills (Table 1). Conventions as in Fig. 2.

In summary, six out of eight craniotomies done using piezoelectric drilling showed T2 hyperintensities (excluding the scan acquired more than 2 wk following surgery), and five out of six done using electric drilling showed T2 hyperintensities (note the overlap in the use of the two tools in monkey P). In contrast, only one of thirteen craniotomies exclusively conducted with mechanical drills had underlying T2 hyperintensities.

DISCUSSION

Utilizing MRI-compatible implants in NHP research permits detailed whole brain monitoring of animals following procedures and throughout the time that the animal participates in research (13, 19, 20). Our utilization of MRI-compatible implants enabled us to note a number of anomalies that may have otherwise gone undetected. T2 hyperintensities following craniotomies, particularly those performed with electric drills, were consistently found in our retrospective examination of craniotomy surgeries in eight animals. Inspecting T2-signatures in humans following craniotomy surgery, traumatic brain injury, and hemorrhages, the hyperintensities found in our study are best compatible with a clinical interpretation of edema of neural tissue (2, 16, 21–26). Our findings suggest that the mode of surgical drilling is responsible for the probability of edema occurrence at craniotomy sites. These anomalies resolve within a short period of time (e.g., two weeks in case 1). Thus, there does not appear to be long-lasting structural (as seen in our T1/T2 scans) or behavioral effects. However, our findings raise the question of surgical tool recommendations to optimize neural tissue integrity in our research animals, which may affect experimental performance (26), and especially as electrophysiological recordings may be conducted immediately postsurgery from tissue that may be physiologically compromised (27, 28).

Our results suggest that edema of neural tissue resolves within 2 wk postsurgery, as indicated by a complete recovery observed in structural scans. Although we do not have histological data from our animals, data from the human and rodent literature suggest that resolution of neural tissue edema is of a similar timeline (23, 26, 29, 30). One study showed that craniotomies using electric drills and trephines induced T2-hyperintensities and behavioral detriments to rats within 2 wk postsurgery (26). Hyperintensities in animals with electric drill craniotomies were larger and were significantly reduced 2 wk postsurgery. However, in trephine surgeries, the affected areas were smaller and the resolution trajectories were heterogeneous among subjects. The study suggested that these findings may result from the disruption of vasculature that innervates epidural tissue, in addition to trauma induced by vibrations or heat during surgery. Our contrasting findings in surgeries performed with mechanical and electric drills corroborate the timelines from this study, but suggest that trephines may induce effects on neural tissue dissimilar to the mechanical drills used in our study.

Postoperative MRI findings have been studied in humans (e.g., see Refs. 1 and 21). Clinical examination of different tools and techniques is key for safer surgeries to reduce unexpected neural lesioning to the patient. Although tools leveraged in animal work can be more diverse and less sophisticated than their counterparts utilized in humans, commonalities exist that may shed light on our findings here. It is difficult to identify the exact etiology of the edemas observed in our cases. Human studies have identified similar T2 hyperintensities following plunging incidents (1, 3, 21). However, it is unlikely that our findings were due to plunging given the characteristics of the piezoelectric drill, the consistency of the findings, and the lack of any detectable deep migration of the tool (or dural tear) reported by the surgical team. Based on the findings in our scans, our data point to the possibility that use of either electric tool may be sufficient to cause an edema of the underlying tissue. Combining both tools in the same surgery did not indicate an additive effect; however, given our small sample size this conclusion is tentative. Conversely, surgeries that exclusively used mechanical drills were not likely to show postsurgical anomalies.

Contrasting signatures on T1 and T2 scans (case 1) and recovery trajectory suggest that edemas underneath the craniotomies were induced either by mechanical forces imposed by an electric (piezo or otherwise) tool, or the heat generated by the tool. An earlier study showed that vibrations from the piezoelectric drill cause minimal damage even when directly applied to tissue (31). However, this study only utilized a brief exposure of 5 s compared with drilling procedures containing bursts of several minutes each. We suspect extended localized vibrations as a probable cause for our findings, as vibrations may reverberate through the skull and dura mater to affect neural tissue, applying mechanical forces in a manner that may promote proinflammatory cytokines, a characteristic of certain types of traumatic brain injury (26, 29, 32, 33). Furthermore, traumatic brain injury has been reported to cause reversible edema with similar T2/FLAIR profiles (22–25). However, given the disparate drilling mechanisms (and operating frequencies) of our two electric drills, their seemingly similar effect on the brain following surgery remains unclear and will require further investigation. Alternatively, thermal dissipation into brain tissue while drilling could also have caused the T2 anomaly; however, reports of lesions following thermal ablation by laser or ultrasound conveyed more focal, circular T2 findings that were sharply delineated from surrounding tissue (34–36) and were long-lasting [observable more than two months following thermal application (37)]. It is also unlikely that thermal effects would be equivalent across the two types of electric drills as the piezoelectric drill includes an irrigation system that alleviates heat build-up during operation. Although less frequent, the occurrence of T2 hyperintensities in surgeries done solely with mechanical drills, which are unlikely to generate heat, also supports the notion that the anomalies were not due to thermal lesions. However, it should be noted that thermal effects reported in the literature were caused by mechanisms of heat application that are quite different from the ones associated with the tools used in our studies, which confounds a direct comparison with our T2 anomalies.

In summary, our results suggest that a more elaborate examination of surgical tools is warranted, especially given recent recommendations that embrace piezoelectric drilling as an ideal tool in macaque cranial surgeries to minimize risk for brain damage (13). This class of drill may prevent serious forms of plunging events with severe consequences to the animal, but may induce nonpermanent lesioning despite an intact dural surface. Although our findings suggest that there is no critical long-term clinical effect on brain tissue, it may be advisable to investigate pharmacological options for postsurgical edema attenuation (16, 22, 30) or, more prudently, to delay neural recordings from cortical sites directly underneath craniotomies that were conducted using electric drilling for a few weeks. If this is not an option, mechanical devices appear to be a better choice to ensure the integrity of neural tissue underlying craniotomies.

GRANTS

This work was supported by National Eye Institute (NEI) Grant 2R01EY017699 (to S.K.), National Institute of Mental Health (NIMH) Grants 2R01MH064043 and P50MH109429 (to S.K.), and National Sciences and Engineering Research Council (NSERC) Grant PDF-557604-2021 (to R.B.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.B. and S.K. conceived and designed research; R.B., M.E., K.D., B.M.M., and M.A.P. performed experiments; R.B. analyzed data; R.B., M.E., M.A.P., and S.K. interpreted results of experiments; R.B., B.W., and M.H. prepared figures; R.B. drafted manuscript; R.B., M.E., K.D., B.W., B.M.M., M.H., M.A.P., and S.K. edited and revised manuscript; R.B., M.E., K.D., B.W., B.M.M., M.A.P., and S.K. approved final version of manuscript.