A Drosophila model of closed head traumaticbrain injuryRebeccah J. Katzenbergera, Carin A. Loewenb, Douglas R. Wassarmana, Andrew J. Petersena, Barry Ganetzkyb,1,and David A. Wassarmana,1

aDepartment of Cell and Regenerative Biology, School of Medicine and Public Health, and bLaboratory of Genetics, University of Wisconsin–Madison,Madison, WI 53706

Contributed by Barry Ganetzky, September 6, 2013 (sent for review August 6, 2013)

Traumatic brain injury (TBI) is a substantial health issue world-wide, yet the mechanisms responsible for its complex spectrum ofpathologies remains largely unknown. To investigate the mecha-nisms underlying TBI pathologies, we developed a model of TBI inDrosophila melanogaster. The model allows us to take advantageof the wealth of experimental tools available in flies. Closed headTBI was inflicted with a mechanical device that subjects flies torapid acceleration and deceleration. Similar to humans with TBI,flies with TBI exhibited temporary incapacitation, ataxia, activa-tion of the innate immune response, neurodegeneration, anddeath. Our data indicate that TBI results in death shortly aftera primary injury only if the injury exceeds a certain thresholdand that age and genetic background, but not sex, substantiallyaffect this threshold. Furthermore, this threshold also appears tobe dependent on the same cellular and molecular mechanisms thatcontrol normal longevity. This study demonstrates the potential offlies for providing key insights into human TBI that may ultimatelyprovide unique opportunities for therapeutic intervention.

concussion | insect | chronic traumatic encephalopathy

Traumatic brain injury (TBI) is a leading cause of neurologicaldeficits and mortality worldwide (1, 2). TBI outcomes result

from primary and secondary injuries that cause cell damage anddeath in the brain. Primary injuries occur during the initial im-pact and are triggered by external mechanical forces that deformthe brain, whereas secondary injuries are triggered by cellularand molecular responses that occur over time in reaction to theprimary injuries. TBI outcomes are heterogeneous in the humanpopulation owing to variation in the location and strength ofprimary injuries as well as genetic and environmental factors thataffect the severity of primary and secondary injuries. This het-erogeneity is one of the most significant barriers to the de-velopment of therapeutic interventions (3–5). Therefore, researchaimed at determining the genetic and environmental factors thataffect the severity of primary and secondary injuries is essential fordeveloping treatments for TBI.To investigate the underlying cellular and molecular basis of

TBI, we developed a Drosophila melanogaster model. Key advan-tages of flies are that (i) large numbers of animals can be rapidlyand inexpensively analyzed to establish causality between injuriesand outcomes, (ii) many molecular and genetic tools are availableto investigate themolecules and pathways that underlie injuries, and(iii) outcomes can readily be evaluated over the entire animal life-span. Thus, flies provide unique opportunities to understand TBI.It is reasonable to expect that TBI can be modeled in flies be-

cause Drosophila has already proved to be an extremely usefulmodel for studying other human neurodegenerative disorders (6).In fact, research in flies has already provided novel insights intoneurodegeneration, memory, and sleep, all of which are affectedin human TBI (6–8). Additionally, fly and human brains havecommon architectural, cellular, and molecular features. The flybrain is bilaterally symmetrical and is joined to the ventral gan-glion that innervates the body, similar to the way that the spinalcord innervates the human body (9). The fly brain is organized into

three regions, the protocerebrum, deutocerebrum, and trito-cerebrum, which are homologous to the forebrain, midbrain, andhindbrain, respectively, of humans (10). The fly brain is composedof functionally diverse neurons with all of the structural features,neurotransmitters, and ion channels found in human neurons(11). Glial cell types and functions in flies are also similar to thosein humans (12). For instance, surface glial cells cover brain neu-rons to form the blood–brain barrier, and glial cells are part of theinnate immune system (13, 14). The fly brain is encapsulated by anexoskeleton, also known as the cuticle (15). Like the human cra-nium, the cuticle is relatively inelastic. It provides protection fromthe environment, and it defines the structure of the head. Thebrain is further separated from the cuticle by fluid hemolymph,a blood-like carrier system for macrophages and nutrients (16).The topology of the fly brain relative to the cuticle is such that animpact to the head or body most likely causes the brain to ricochetand deform against the cuticle resulting in TBI.Here, we describe the development and initial characteriza-

tion of a fly model of TBI. Using the model, we found thatfundamental characteristics of human TBI also occur in flies. Wealso found that primary injuries exacerbate the normal age-related decline in flies. This may explain why human TBI is as-sociated with cognitive and neurodegenerative disorders that aretypical of older individuals and why TBI outcomes are worse inolder individuals.

ResultsThe HIT Device Reproducibly Inflicts Closed Head TBI in Flies. To in-flict TBI in flies, we built the “high-impact trauma” (HIT) device,which consists of a metal spring clamped at one end to a wooden

Significance

Traumatic brain injury (TBI) occurs when a strong jolt to thehead causes damage to brain cells, resulting in immediate andlong-term consequences including physical, behavioral, andcognitive problems. Despite the importance of TBI as a majorhealth issue, our understanding of the underlying cellular andmolecular mechanisms is limited. To unravel these mechanisms,we have developed a model of TBI in the fruit fly, Drosophilamelanogaster, where we can apply many powerful experi-mental tools. The main features of human TBI also occur in flies,suggesting that the underlying mechanisms are conserved. Ourstudies demonstrate the value of a fly model for understandingthe consequences of TBI andmay ultimately enable developmentof therapies for their prevention and treatment.

Author contributions: R.J.K., A.J.P., B.G., and D.A.W. designed research; R.J.K., C.A.L.,D.R.W., A.J.P., and D.A.W. performed research; D.A.W. contributed new reagents/analytictools; R.J.K., C.A.L., B.G., and D.A.W. analyzed data; and R.J.K., C.A.L., B.G., and D.A.W.wrote the paper.

The authors declare no conflict of interest.1To whom correspondence may be addressed. E-mail: ganetzky@wisc.edu or dawassarman@wisc.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1316895110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1316895110 PNAS Early Edition | 1 of 8

NEU

ROSC

IENCE

PNASPL

US

board with the free end positioned over a polyurethane pad (Fig.1A and Fig. S1). A standard plastic vial containing unanesthetizedflies that are confined to the bottom quarter of the vial by a sta-tionary cotton ball is connected to the free end of the spring.When the spring is deflected and released, the vial rapidly contactsthe polyurethane pad, and a mechanical force is delivered tothe flies as they contact the vial wall and rebound (Fig. 1B). In-dividual flies presumably contact the wall with different regions oftheir head and/or body and with different forces, so primaryinjuries will vary among flies in the same vial. The lack of pene-trating injuries and the randomness of impact location andstrength are features of closed head TBI in the human population.Since closed head TBI is the most common form of TBI occurringin falls, sports collisions, and automobile crashes, a fly model is ofsignificant potential value (17). As shown in Fig. 2, the phenotypiceffects on flies subjected to HIT device-induced injury are highlyreproducible.The strength of the primary injuries inflicted by the HIT de-

vice can be adjusted, either by varying the extent of spring de-flection or by varying the number of strikes. Deflection of thespring to 90° resulted in an impact velocity of ∼3.0 m/s (6.7 miles/h)and an average force of 2.5 N. Some flies subjected to a singlestrike with the spring deflected to 90° became temporarily in-capacitated and fell to the bottom of the vial; however, they did notincur obvious external damage to the head, body, or appendages(Fig. 1 C and D). During the first minute after a strike, 8.8 ± 3.8%of flies were incapacitated, lying on their back or side on the bot-tom of the vial. Most of the incapacitated flies recovered locomotoractivity within 5 min (Fig. S2A). Subsequently, their mobility, asmeasured by climbing ability, was reduced but gradually returnedto normal over a 2-d period (Fig. S2B and Table S1). The imme-diate loss of motor ability, followed by ataxia and gradual recoveryof mobility are reminiscent of concussion in humans and areconsistent with the idea that the HIT device inflicts brain injuryin flies.

Primary Injuries That Exceed a Threshold Cause Death Within 24 h. Todetermine the effect of subjecting flies to multiple TBI incidents,we varied the number of strikes that flies received and measured

the percentage of flies that died within 24 h. We define thepercentage of flies that died within 24 h after injury as themortality index at 24 h (MI24). To minimize any variation inoutcome associated with differences in age, genotype, or sex, weused 0- to 3-d-old white (w1118) flies in every experiment, with anapproximately equal number of males and females. Flies re-ceived 0–10 strikes with the spring deflected to 90° and with 5min recovery periods between strikes. After a single strike, theMI24 was 4.5 ± 1.2 (Fig. 2A). Additional strikes resulted in anincrease in the MI24; however, additional strikes did not signifi-cantly increase the MI24 per strike (Fig. 2B). The fact that not allflies died after a single strike indicates that primary injuries causedeath within 24 h only if the injuries exceed a specific threshold,where threshold is presumably a composite measure of impactlocation and strength. Furthermore, the fact that the number ofstrikes did not affect the MI24 per strike indicates that fliessubjected to primary injuries below the threshold do not haveincreased susceptibility for mortality from subsequent primaryinjuries.We also examined whether varying the recovery time between

repeated strikes had any effect on the MI24. The expectation wasthat if the time between primary injuries was increased, thensecondary injuries or injury repair mechanisms would have moretime to occur, thereby altering the threshold for death caused bysubsequent primary injuries. Thus, the MI24 would change as thetime between strikes changed. To test this hypothesis, flies re-ceived four strikes with time intervals from 1 to 60 min betweensuccessive strikes. The temporal spacing of strikes did not sig-nificantly affect the MI24 (Fig. S3A). Therefore, over the timestested, the time between successive strikes neither exacerbatesnor ameliorates the consequences of multiple primary injuries.The fact that the time between strikes did not affect the MI24indicates that secondary injuries and injury repair mechanismseither neutralize one another or do not contribute to the primaryinjury threshold for death within 24 h.Another parameter that could affect the MI24 is the number of

flies per vial. To investigate this possibility, we placed 10–60 fliesper vial in increments of 10 and subjected the vials to four strikeswith 5-min recovery periods between strikes. The number of fliesin a vial did not significantly affect the MI24 (Fig. S3B). Thus, invials containing 10–60 flies, contact among flies does not appearto affect the primary injury threshold for death within 24 h.

TBI Reduces Lifespan. To investigate whether multiple TBI inci-dents have long-term effects on survival, we determined thelifespan of flies that survived beyond 24 h after receiving 0–10strikes with 5 min recovery periods between strikes. We assumedthat death from primary injuries was complete by 24 h becausethe percent survival at 24 h was not substantially different fromthe percent survival at 48 h. Relative to untreated flies, flies thatreceived one strike had a substantially reduced median lifespan(48.3 ± 1.2 d vs. 38.3 ± 3.5 d) and significantly reduced maximumlifespan (85.7 ± 3.6 d vs. 73.0 ± 0.6 d) (Fig. 2C and Table S2).Additional strikes caused further reductions in the median andmaximum lifespan, but an upper limit for median lifespan wasreached at four strikes; additional strikes did not further reducethe median lifespan. These results demonstrate that primaryinjuries that are below the threshold for death within 24 h none-theless cause a reduction in lifespan that likely results from thelong-term consequences of secondary injuries.

Age-Associated Processes Lower the Primary Injury Threshold. Forsubsequent experiments, we developed a standard TBI protocolconsisting of four strikes separated by 5 min intervals. Thisprotocol resulted in an MI24 of 19.0 ± 4.0 for 0- to 3-d-old w1118

flies (Fig. 2A). This moderate MI24 is convenient for identifyinggenetic and environmental factors that enhance or suppress theeffect of primary injuries.

Fig. 1. The HIT device was used to inflict TBI in 0- to 4-d-old w1118flies.

Images show the HIT device and flies before and after a strike. Shown are (A)the HIT device deflected to 90° before a strike, (B) the HIT device immedi-ately after a strike, (C) 60 flies in a vial before a strike, and (D) after a strike.

2 of 8 | www.pnas.org/cgi/doi/10.1073/pnas.1316895110 Katzenberger et al.

We used the standard TBI protocol to determine the effect ofsex and age on the MI24. w

1118 male and female flies at 0–4 and20–22 d old were assayed separately. Sex did not have a signifi-cant effect on the MI24 at either age (Fig. 3A). In contrast, agedid have a significant effect; the MI24 was approximately twofoldhigher for older flies compared with younger flies.To more systematically examine the effect of age on the MI24,

we examined flies in 12 age groups ranging from 0–3 to 28–29 dold that were treated with the standard TBI protocol. The datarevealed an increase in the MI24 as age increased (Fig. 3B).These results suggest that cellular and molecular changes occurduring aging that lower the primary injury threshold for deathwithin 24 h.

Age-Associated Processes Enhance Neurodegeneration Due to BrainInjuries. TBI in rodents and humans increases the occurrence ofchronic traumatic encephalopathy (CTE), a form of neuro-degeneration that generally appears long after (years to decadesin the case of humans) the primary injury (18). To determine ifneurodegeneration is also a long-term outcome of TBI in fliesand whether it is affected by age at the time of primary injury, we

performed histological analysis on brains of younger (0- to 4-d-old) and older (20- to 21-d-old) flies 14 d after they were sub-jected to the standard TBI protocol. Neurodegeneration in fliesis commonly manifested by the appearance of vacuolar lesions inthe brain neuropil, a region enriched for axons, synaptic termi-nals, and glial cells (6). In contrast with untreated flies whosebrains had a smooth and uniform appearance throughout theneuropil (Fig. 4A), treated flies exhibited neuropathology man-ifested by the appearance of numerous small vacuolar lesions(∼1.0 μm in diameter) (Fig. 4B). We also observed large vacuolarlesions (5.0–10.0 μm in diameter) in injured flies. The incidenceof large vacuolar lesions in the central region of the brain in-creased with the number of times flies were subjected to thestandard TBI protocol (Fig. 4C). In addition, the incidence oflarge vacuolar lesions was higher in older flies subjected to TBIcompared with similarly treated younger flies. These data indicatethat brain injuries in flies elicit neurodegeneration as a long-termconsequence and that the extent of neurodegeneration is de-termined by the severity of primary injuries and the age at the timeof primary injuries.

Fig. 2. The effect of the number of strikes on the MI24 and the lifespan of w1118flies. (A) The MI24 is graphed vs. the number of strikes. MI24 values were

normalized to those of untreated flies. (B) Data from A are graphed as the MI24 per strike vs. the number of strikes. The MI24 per strike was not significantlyaffected by the number of strikes (P = 0.82, one-way ANOVA). (C) The percent survival is graphed vs. age for flies that received the indicated number ofstrikes. Only flies that survived 24 h after treatment were included in the analysis. The line at 50% indicates the median lifespan. Error bars indicate the SD forat least three independent trials of 60 flies each. Exact values and SD for median and maximum lifespans are provided in Table S2.

Katzenberger et al. PNAS Early Edition | 3 of 8

NEU

ROSC

IENCE

PNASPL

US

Primary Injuries Activate the Innate Immune Response. One impor-tant cause of secondary injuries in TBI is activation of the innateimmune response (19). The innate immune response is triggeredby pathogen-derived molecules and by endogenous moleculesgenerated by stressed and injured cells (20). A component of thisresponse is the production of proinflammatory cytokines, such astumor necrosis factor (TNF), by microglial cells and astrocytes.In some studies, elevated levels of TNF in the cerebrospinal fluidof TBI patients are correlated with unfavorable clinical out-comes, providing evidence of the injurious role of cytokines (21).Likewise, inhibition of TNF shortly after primary injuries inrodents reduces the severity of some deleterious outcomes, suchas tissue loss (22). On the other hand, long-term recovery ofneurological damage in TBI is impaired in TNF-deficient mice,providing evidence of a beneficial role of cytokines (23). Therefore,the innate immune response plays a critical but not yet fully un-derstood role in TBI. Flies provide an opportunity to advance ourunderstanding of this role because innate immune responsepathways are highly conserved between flies and humans (24).The Toll pathway in flies is analogous with mammalian Toll-likereceptor (TLR) pathways, and the Immune deficiency (Imd)pathway is analogous with the mammalian TNF pathway. Amongthe key functions of both the Toll and Imd pathways is the tran-scriptional activation of antimicrobial peptide (AMP) genes (24).To assess activation of the innate immune response in flies

following primary injuries, we used quantitative real-time reversetranscription–PCR (qRT-PCR) to determine AMP gene mRNAlevels in heads of w1118

flies subjected to the standard TBIprotocol. To control for the usual increase in expression of innateimmunity genes that occurs as a function of age, treated flieswere normalized to age-matched untreated flies (25). Flies thatsurvived the standard TBI protocol exhibited an increase inAMP gene expression within 24 h after the primary injuries (Fig.5). Similar results were observed in 0- to 4-d-old and 20- to 21-d-old flies. Some AMP genes were activated within 1 h after theprimary injury and all AMP genes were activated within 24 h.Activation of innate immunity genes following primary injuries is

at least somewhat specific because the expression of other genessuch as TAF1 was not similarly increased. Thus, TBI in flies, asin mammals, elicits activation of an innate immune responsepathway.

Genetic Background Affects the Primary Injury Threshold. An in-creasing body of evidence implicates genetic factors in the vari-able clinical outcomes of TBI in humans (26). To investigate theeffect of genetic background on the primary injury threshold inflies, we determined the MI24 for 42 fly lines (aged 0–7 d) thatwere subjected to the standard TBI protocol. Five of these linesare commonly used as wild-type controls in various Drosophilaexperiments; the remaining 37 lines contain mutations in genesthat encode components of the Imd or Toll pathways (24).The MI24 showed wide variation among the lines, ranging

from 14.9 ± 1.2 to 82.5 ± 5.0 (Fig. 6A). Variability was evenobserved among wild-type controls, for which the MI24 rangedfrom 20.3 ± 1.4 to 39.9 ± 3.8. Among the mutant lines there werestriking differences for different alleles of the same gene. Forexample, the MI24 for ImdSDK and Imd10191 was 21.0 ± 3.3 and40.9 ± 4.8, respectively. Additionally, the MI24 did not correlatewith either the Imd or the Toll pathway. For example, mutationsin Relish (Rel), which encodes the NF-κB transcription factor inthe Imd pathway, had a significantly different effect on the MI24than mutations in kenny (key), which encodes an activator of Rel(RelE20, 47.2 ± 2.3; RelE38, 41.7 ± 3.3; keyf05097, 8.5 ± 1.2; andkeyc02831, 7.9 ± 1.9) (24). The wide range of effects for wild-typelines and the lack of coherent effects for Imd and Toll pathwaylines suggests that in young flies many genes in the geneticbackground affect the primary injury threshold for death within24 h. We suspect that, in some cases, the genetic backgroundmasks the effect of mutations in Imd or Toll pathway genesmaking it impossible to determine from these data whetherthe innate immune response specifically influences the MI24.

Genetic Background Affects the Age-Dependent Reduction of thePrimary Injury Threshold. To investigate the effect of geneticbackground on the primary injury threshold in older flies, we

Fig. 3. The MI24 of w1118

flies subjected to the standard TBI protocol is not affected by sex but is affected by age at the time of injury. (A) For flies treatedwith the standard TBI protocol, the MI24 is graphed vs. sex. The experiment was performed with 0- to 4-d-old and 20- to 22-d-old flies. MI24 values werenormalized to those of untreated flies. Sex did not have a significant effect on the MI24 for either 0- to 4-d-old flies (P = 0.27, one-tailed t test) or 20- to 22-d-old flies (P = 0.34, one-tailed t test). Age at the time of injury did have a significant effect on the MI24 for both males (P = 0.004, one-tailed t test) and females(P = 0.002, one-tailed t test). (B) The MI24 is graphed vs. age at the time of treatment with the standard TBI protocol. MI24 values were normalized to those ofuntreated flies. Error bars indicate the SD for at least three independent trials of 60 flies each.

4 of 8 | www.pnas.org/cgi/doi/10.1073/pnas.1316895110 Katzenberger et al.

reexamined the 42 fly lines at 20–27 d old. The results largelyparalleled those observed in young flies in that wide variabilitywas observed in the MI24, and the variability did not correlatewith the Imd or the Toll pathway (Fig. 6B). In addition, for everyline, the MI24 for 20- to 27-d-old flies was higher than the MI24for 0- to 7-d-old flies (Fig. S4). MIs24 of younger and older flieshad a correlation coefficient (r) of 0.77. Thus, in all geneticbackgrounds, age appears to be a critical determinant of theprimary injury threshold for death within 24 h. However, theMI24 in younger flies was not entirely predictive of the extentof increase in the MI24 in older flies. For example, 0- to 7-d-oldkeyf05097 and spatzle (spz3) flies had similar MIs24 (14.9 ± 1.2and 15.9 ± 3.2, respectively) but 20- to 27-d-old keyf05097 and

spz3 flies had substantially different MIs24 (22.5 ± 4.2 and 55.1 ±9.4, respectively) (Fig. S4). Conversely, 0- to 7-d-old necrotic (nec2)and TAK1-associated binding protein 2 (Tab2EY00380) flies hadsubstantially different MIs24 (17.6 ± 4.1 and 44.2 ± 4.6, re-spectively) but 20- to 27-d-old nec2 and Tab2EY00380 flies hadsimilar MIs24 (62.4 ± 5.9 and 65.1 ± 6.3, respectively). These dataindicate that the genes that affect the primary injury threshold fordeath within 24 h in young flies are different from those that affectthe age-dependent reduction in the primary injury threshold.

Variation in Primary Injury Threshold Correlates with Variation inLongevity. Because the primary injury threshold varies withage, we examined the correlation between MI24 and longevity

Fig. 4. Treatment ofw1118flies with the standard TBI protocol causes neurodegeneration. Images of sections of fly brains from 14- to 18-d-old (A, Upper and

Lower) untreated and (B, Upper and Lower) treated flies are shown. The treated fly received the standard TBI protocol at 0–4 d old. The large vacuole in-dicated by the arrow in the magnified image (B, Lower) is 7.75 μm in diameter. (Scale bars in A, Upper and B, Upper: 100 μm. Images in Lower panels aremagnified an additional 5×.) (C) The number of large, i.e., 5.0- to 10.0-μm-diameter vacuoles in the central region of the brain is graphed vs. the number oftimes flies were subjected to the standard TBI protocol. Brain sections chosen for analysis were at equivalent depths in the brain, as exemplified by A and B.Multiple treatments with the standard TBI protocol occurred over successive days. All flies were analyzed 14 d after the time of the first treatment, i.e., flieslabeled 14–18 d were treated beginning at 0–4 d old, and flies labeled 34–35 d were treated beginning at 20–21 d old. Error bars indicate the SD for at leastthree heads. When treated with one standard TBI protocol, young flies (14–18 d old) developed significantly fewer large vacuoles that older flies (34–35 d old)(P = 0.003, one-tailed t test), but the difference between young and older flies was not significant for treatment with two or three standard TBI protocols (P =0.07 and P = 0.07, one-tailed t test, respectively). Both young and older flies developed significantly more large vacuoles when treated with three vs. onestandard TBI protocol (P = 0.029 for young flies and P = 0.03 for older flies, one-tailed t test).

Katzenberger et al. PNAS Early Edition | 5 of 8

NEU

ROSC

IENCE

PNASPL

US

in the absence of TBI. We analyzed 14 fly lines, including lineswith low, average, or high MI24 after treatment with the stan-dard TBI protocol. We found a negative linear relationshipbetween the MI24 and the median lifespan for both 0- to 7-d-oldand 20- to 27-d-old flies (Fig. 7A). The correlation coefficient r

between the MI24 and the median lifespan was −0.67 for 0-to 7d-old flies and −0.84 for 20- to 27-d-old flies. Thus, thelongevity of a particular fly line and its primary injury thresh-old for death within 24 h are largely determined by the samegenetic factors.

Fig. 5. Treatment of w1118flies with the standard

TBI protocol activates the innate immune response.The histogram shows the fold-increase in mRNAlevels in treated vs. untreated flies at the indicatedtimepoints after treatment of 0- to 4- or 20- to 21-d-old flies. The AMP genes examined were Attacin-C(AttC), Diptericin B (DiptB), and Metchnikowin(Mtk), and the control gene examined was TBP-as-sociated factor 1 (TAF1). Error bars indicate the SEMfor at least three independent trials.

Fig. 6. The MI24 is strongly affected by geneticbackground. Histograms show the MI24 for 42 dif-ferent fly lines treated with the standard TBI pro-tocol and tested at (A) 0–7 d old and (B) 20–27 d old.The MI24 vs. fly genotype is graphed. MI24 valueswere normalized to those of untreated flies. Geno-types are listed in Table S3. White bars indicate flylines containing mutations in genes implicated in theImd pathway. Gray bars indicate fly lines containingmutations in genes implicated in the Toll pathway.Black bars indicate fly lines commonly used as wild-type controls in Drosophila experiments. For a refer-ence, fly line number 7 is w1118. Error bars indicatethe SD for at least three independent trials of 60flies each.

6 of 8 | www.pnas.org/cgi/doi/10.1073/pnas.1316895110 Katzenberger et al.

To test this proposal, we extended the lifespan of flies andexamined the effect on the MI24. To extend the lifespan, weraised flies at 18 °C rather than at 25 °C, the temperature atwhich flies were raised for all of the prior experiments. Un-treated w1118

flies raised at 18 °C had a significantly longer medianlifespan relative to flies raised at 25 °C (68.5± 0.8 d vs. 48.3± 1.2 d,P < 0.0001, one-tailed t test). After treatment with the standardTBI protocol, w1118

flies raised at 18 °C had a significantly lowerMI24 than equivalent-age flies raised at 25 °C (Fig. 7B). The neg-ative correlation between natural longevity and MI24 was observedfor both 0- to 3-d-old and 21- to 22-d-old flies. Thus, flies of thesame genotype and chronological age but different median lifespandiffered in their primary injury threshold for death within 24 h. Thisresult indicates that environmental factors such as temperaturedetermine both the longevity of a particular line in the absence ofinjury and its primary injury threshold for death within 24 h.

DiscussionFlies Can Model Human TBI. Here, we have described the de-velopment and initial characterization of a fly model of TBI. Wefound that inflicting mechanical injury on flies by rapid acceler-ation and deceleration produces outcomes that are similar tooutcomes characteristic of closed head TBI in humans (1, 2).These outcomes include temporary incapacitation (Fig. S2A),ataxia (Fig. S2B and Table S1), activation of the innate immuneresponse (Fig. 5), neurodegeneration (Fig. 4), and death (Fig. 2).In addition, we found that risk factors for mortality in flies areshared with humans. Our data indicate that the MI24 followinginjury does not vary with sex (Fig. 3A). Similarly, a study of 914women and 916 men by Coimbra et al. concluded that sex is nota risk factor for mortality in TBI patients (27). The MI24, how-ever, is dependent on the age of flies at the time of injury (Figs.3B and 6). Likewise, in a study of 5,612 TBI patients, Hukkelhovenet al. found that the percent mortality within 6 mo of primaryinjuries increased with age, 21% at <35 y and 52% at >55 y (28).Also, in a study of 244 TBI patients, Dhandapani et al. foundthat the percent mortality within 1 mo of primary injuries was

significantly associated with increasing age, 15% at <18 y, 44% at18–59 y, and 52% at >59 y (29). These similarities between fliesin our TBI model and human TBI patients strongly indicate thatflies and humans incur TBI through similar cellular andmolecularmechanisms.

The Fly Model of TBI Can Provide Unique Insights into Human TBI.Because the fly TBI model enables us to analyze many animalsthroughout their lifespan, we have found several properties ofTBI that may be of clinical relevance: age at the time of primaryinjury as well as the number of primary TBI incidents are riskfactors for neurodegeneration (Fig. 4C); the number of primaryTBI incidents affects longevity (Fig. 2C and Table S2); the MI24following TBI varies with genetic background (Fig. 6); and fi-nally, the genes that affect the susceptibility to mortality aredifferent at different ages (Fig. 6 and Fig. S4). Although it isdifficult to obtain exactly comparable information from humanTBI patients, available data are consistent with our conclusions.For example, as in flies, human studies suggest that age and theseverity of primary injury are important factors in CTE de-velopment (30, 31). In addition, studies in different rodentstrains as well as genetic association studies in TBI patientsindicate that, as in flies, TBI outcomes depend on genetic back-ground (26, 32).Finally, we found that the genetic and environmental factors

that affect longevity in the absence of injury also affect the sus-ceptibility to mortality following TBI (Fig. 7). Two lines of evi-dence suggest that fractional age rather than chronological ageper se is a risk factor for TBI-induced mortality in humans as itis in flies, where fractional age is defined as chronological agerelative to the maximum longevity of flies of that genotype. First,drugs such as rapamycin and resveratrol that alleviate TBI out-comes in mammals also extend normal longevity (33–36). Sec-ond, the gene apolipoprotein E (apoE), implicated in controllinglongevity, is also found to be a risk factor for TBI outcomes inmammals (37, 38). Interestingly, rapamycin, resveratrol, andapoE also appear to affect the severity of neurodegeneration inAlzheimer’s disease, which is phenotypically similar to CTE (18,39). Collectively, these findings suggest that secondary injuries inTBI accelerate some normal aging processes in the brain. Thisidea may explain why the average age of onset (AAO) for CTE issubstantially earlier than for Alzheimer’s disease (CTE: AAO =42.8 ± 12.7 y, range = 25–76 y; Alzheimer’s disease: AAO = 72.8 ±6.8 y, range = 49–97 y) (40, 41).In summary, our studies demonstrate that a fly TBI model

offers considerable potential for understanding the cellularand molecular mechanisms that underlie the pathological con-sequences of human TBI and may ultimately facilitate developmentof unique therapeutic strategies.

Materials and MethodsFly Lines and Culturing. Flies were maintained on standard molasses mediumat 25 °C unless otherwise stated. All fly lines were obtained from the Bloo-mington Stock Center except for ImdSDK and Imd10191, which were obtainedfrom Mimi Shirasu-Hiza (Columbia University, New York). Genotypes of fliesused in Figs. 6 and 7A and Fig. S4 are described in Table S3.

Assays for TBI Outcomes. Construction and use of the HIT device is described inFig. S1. The impact velocity of the HIT device was determined by analyzingimages from a high-speed camera. MIs24 were determined by analyzing atleast 180 flies (three experiments of 60 flies). The percentage of flies thatwere incapacitated was determined by analyzing movie recordings of fliesafter a single strike. Flies that did not show obvious locomotor activity wereconsidered incapacitated. Before each timepoint, hitting the benchtop uponwhich the fly vial sat was used to stimulate locomotion. Climbing, lifespan,histology, and qRT-PCR assays were performed as described in Petersen et al.(14, 42).

Fig. 7. The susceptibility of different fly lines to TBI-induced mortality isinversely correlated with their respective longevity. (A) The MI24 of 20- to 27-d-old flies is graphed vs. the median lifespan for 14 of the fly lines that wereanalyzed in Fig. 6. The lines that were analyzed are listed in Table S3. RelE20

and RelE38 flies had the same MI24 and median lifespan, so they appear asa single open box on the graph. (B) The MI24 is graphed for w1118

flies of theindicated age that were raised at either 18 °C or 25 °C. MI24 values werenormalized to those of untreated flies. Temperature had a significant effecton the MI24 for both 0- to 4-d-old flies (P = 0.012, one-tailed t test) and 21- to22-d-old flies (P = 0.001, one-tailed t test). Error bars indicate the SD for atleast three independent trials of 60 flies each.

Katzenberger et al. PNAS Early Edition | 7 of 8

NEU

ROSC

IENCE

PNASPL

US

ACKNOWLEDGMENTS. We thank Nicole Bertram, Grace Boekhoff-Falk, mem-bers of the D.A.W. and B.G. laboratories, and Ron Kalil for their insights thatgreatly improved this research; Russell Little for help in determining the impact

velocity; and Satoshi Kinoshita for performing the histology. This work wassupported by National Institutes of Health Grants R01 NS059001 (to D.A.W.)and R01 AG033620 (to B.G.).

1. Masel BE, DeWitt DS (2010) Traumatic brain injury: A disease process, not an event.J Neurotrauma 27(8):1529–1540.

2. Blennow K, Hardy J, Zetterberg H (2012) The neuropathology and neurobiology oftraumatic brain injury. Neuron 76(5):886–899.

3. Maas AI, et al. (2012) Re-orientation of clinical research in traumatic brain injury:Report of an international workshop on comparative effectiveness research. J Neurotrauma29(1):32–46.

4. Xiong Y, Mahmood A, Chopp M (2013) Animal models of traumatic brain injury. NatRev Neurosci 14(2):128–142.

5. O’Connor WT, Smyth A, Gilchrist MD (2011) Animal models of traumatic brain injury:A critical evaluation. Pharmacol Ther 130(2):106–113.

6. Lessing D, Bonini NM (2009) Maintaining the brain: Insight into human neuro-degeneration from Drosophila melanogaster mutants. Nat Rev Genet 10(6):359–370.

7. Kahsai L, Zars T (2011) Learning and memory in Drosophila: Behavior, genetics, andneural systems. Int Rev Neurobiol 99:139–167.

8. Bushey D, Cirelli C (2011) From genetics to structure to function: Exploring sleep inDrosophila. Int Rev Neurobiol 99:213–244.

9. Boyan GS, Reichert H (2011) Mechanisms for complexity in the brain: Generating theinsect central complex. Trends Neurosci 34(5):247–257.

10. Reichert H (2005) A tripartite organization of the urbilaterian brain: Developmentalgenetic evidence from Drosophila. Brain Res Bull 66(4-6):491–494.

11. Littleton JT, Ganetzky B (2000) Ion channels and synaptic organization: Analysis of theDrosophila genome. Neuron 26(1):35–43.

12. Stork T, Bernardos R, Freeman MR (2012) Analysis of glial cell development andfunction in Drosophila. Cold Spring Harb Protoc 2012(1):1–17.

13. DeSalvo MK, Mayer N, Mayer F, Bainton RJ (2011) Physiologic and anatomic charac-terization of the brain surface glia barrier of Drosophila. Glia 59(9):1322–1340.

14. Petersen AJ, Rimkus SA, Wassarman DA (2012) ATM kinase inhibition in glial cellsactivates the innate immune response and causes neurodegeneration in Drosophila.Proc Natl Acad Sci USA 109(11):E656–E664.

15. Fristrom D, Fristrom JW (1993) The metamorphic development of the adult epidermis.The Development of Drosophila melanogaster, eds Bate M, Martinez-Arias A (ColdSpring Harbor Laboratory Press, Cold Spring Harbor, NY), pp 843–897.

16. Piyankarage SC, Featherstone DE, Shippy SA (2012) Nanoliter hemolymph sampling andanalysis of individual adult Drosophila melanogaster. Anal Chem 84(10):4460–4466.

17. Faul M, Xu L, Wald MM, Coronado VG (2010) Traumatic Brain Injury in the UnitedStates: Emergency Department Visits, Hospitalizations, and Deaths (Centers forDisease Control and Prevention, National Center for Injury Prevention and Control,Atlanta).

18. Baugh CM, et al. (2012) Chronic traumatic encephalopathy: Neurodegeneration fol-lowing repetitive concussive and subconcussive brain trauma. Brain Imaging Behav6(2):244–254.

19. Hellewell SC, Morganti-Kossmann MC (2012) Guilty molecules, guilty minds? Theconflicting roles of the innate immune response to traumatic brain injury. MediatorsInflamm 2012:356494.

20. Ransohoff RM, BrownMA (2012) Innate immunity in the central nervous system. J ClinInvest 122(4):1164–1171.

21. Csuka E, et al. (1999) IL-10 levels in cerebrospinal fluid and serum of patients withsevere traumatic brain injury: Relationship to IL-6, TNF-alpha, TGF-beta1 and blood-brain barrier function. J Neuroimmunol 101(2):211–221.

22. Shohami E, Bass R, Wallach D, Yamin A, Gallily R (1996) Inhibition of tumor necrosisfactor alpha (TNFalpha) activity in rat brain is associated with cerebroprotection afterclosed head injury. J Cereb Blood Flow Metab 16(3):378–384.

23. Sullivan PG, et al. (1999) Exacerbation of damage and altered NF-kappaB activation inmice lacking tumor necrosis factor receptors after traumatic brain injury. J Neurosci19(15):6248–6256.

24. Lemaitre B, Hoffmann J (2007) The host defense of Drosophila melanogaster. AnnuRev Immunol 25:697–743.

25. Ramsden S, Cheung YY, Seroude L (2008) Functional analysis of the Drosophila im-mune response during aging. Aging Cell 7(2):225–236.

26. Dardiotis E, et al. (2010) Genetic association studies in patients with traumatic braininjury. Neurosurg Focus 28(1):E9.

27. Coimbra R, Hoyt DB, Potenza BM, Fortlage D, Hollingsworth-Fridlund P (2003) Doessexual dimorphism influence outcome of traumatic brain injury patients? The answeris no!. J Trauma 54(4):689–700.

28. Hukkelhoven CW, et al. (2003) Patient age and outcome following severe traumaticbrain injury: An analysis of 5600 patients. J Neurosurg 99(4):666–673.

29. Dhandapani S, Manju D, Sharma B, Mahapatra A (2012) Prognostic significance of agein traumatic brain injury. J Neurosci Rural Pract 3(2):131–135.

30. Gavett BE, Stern RA, McKee AC (2011) Chronic traumatic encephalopathy: A potentiallate effect of sport-related concussive and subconcussive head trauma. Clin SportsMed 30(1):179–188, xi.

31. Gavett BE, Stern RA, Cantu RC, Nowinski CJ, McKee AC (2010) Mild traumatic braininjury: A risk factor for neurodegeneration. Alzheimers Res Ther 2(3):18.

32. Al Nimer F, et al. (2013) Strain influences on inflammatory pathway activation, cellinfiltration and complement cascade after traumatic brain injury in the rat. BrainBehav Immun 27(1):109–122.

33. Singleton RH, Yan HQ, Fellows-Mayle W, Dixon CE (2010) Resveratrol attenuatesbehavioral impairments and reduces cortical and hippocampal loss in a rat controlledcortical impact model of traumatic brain injury. J Neurotrauma 27(6):1091–1099.

34. Timmers S, Auwerx J, Schrauwen P (2012) The journey of resveratrol from yeast tohuman. Aging (Albany, NY Online) 4(3):146–158.

35. Guo D, Zeng L, Brody DL, Wong M (2013) Rapamycin attenuates the development ofposttraumatic epilepsy in a mouse model of traumatic brain injury. PLoS ONE 8(5):e64078.

36. Johnson SC, Rabinovitch PS, Kaeberlein M (2013) mTOR is a key modulator of ageingand age-related disease. Nature 493(7432):338–345.

37. Mahley RW, Huang Y (2012) Apolipoprotein e sets the stage: Response to injurytriggers neuropathology. Neuron 76(5):871–885.

38. Ang LS, Cruz RP, Hendel A, Granville DJ (2008) Apolipoprotein E, an important playerin longevity and age-related diseases. Exp Gerontol 43(7):615–622.

39. Tanzi RE (2012) The genetics of Alzheimer disease. Cold Spring Harb Perspect Med2(10):a006296, 10.1101/cshperspect.a006296.

40. McKee AC, et al. (2009) Chronic traumatic encephalopathy in athletes: Progressivetauopathy after repetitive head injury. J Neuropathol Exp Neurol 68(7):709–735.

41. Li YJ, et al. (2002) Age at onset in two common neurodegenerative diseases is ge-netically controlled. Am J Hum Genet 70(4):985–993.

42. Petersen AJ, Katzenberger RJ, Wassarman DA (2013) The innate immune responsetranscription factor relish is necessary for neurodegeneration in a Drosophila modelof ataxia-telangiectasia. Genetics 194(1):133–142.

8 of 8 | www.pnas.org/cgi/doi/10.1073/pnas.1316895110 Katzenberger et al.