- Email: [email protected]
Acyclovir-resistant, pathogenic herpesviruses
REVIEWS
Acyclovir-resistant, pathogenic herpesviruses Donald M. Coen hut I
‘erpesviruscs are in]portant human pathotheir in.gerls and fections can be particularly scvcrc in immutlocompromised patients, such as those with AIDS. I’crhaps the most striking successes in antiviral chemotherapy have ._come in treating herpesvIrus Intcctlons, cspccially with the development of acyclovir against herpes simplex virus (HSV). However, as is typical of successful antimicrobial agents, drug resistance has reared its head’m4. Although its composers may not have had this in mind, the song ‘Friend of the Devil’ is pertinent to two intertwined aspects of drug resistance and pathogenicity of HSV. ‘l’he first is that the resistance of any pathogen to antimicrobial drugs is a trade-off between drug rcsistancc and pathogenicity. ‘I‘his is particularly true for I ISV, where drug resistance requires mutation in viral genes that have evolved to assure success as a pathogen. Thus, to cause clinically significant drug resistance, HSV must evade the drug (‘set out running’), but must also retain sufficient function in gent products important for pathogcncsis (‘take its time’) to cause discasc. Intertwined with this trade-off is probably the most important aspect of HSV pathogencsis - latency - in which I ISV stops replicating (‘takes its time’) and enters a noninfectious form with limited gene cxprcssion in sensory neurons, allowing it to avoid immune surveillance. HSV can then reactivate from latency to cause disease and allow transmission to a new host. 1Iowevcr, the simplest route to resistance to acyclovir - thymidinekinasc-negative (TK-) mutation - not only attenuates pathogcncsis generally, but also prevents reactivation from latency in animal models. Even so, ‘I’K- mutants have been reported to be associated with cases of disease caused by IISV that is resistant to therapy with acyclovir. This review addresses this paradox.
to the drug is summarized in Fig. 1. Acyclovir is first activated by phosphorylation, which is performed much more efficiently by the HSV TK than it is by cellular enzymes. Cellular enzymes then convert acyclovir monophosphate to its triphosphate, which is a more potent inhibitor of HSV DNA polymcracc than it is of cellular enzymes (rcvicwed in Ref. 5). Accordingly, resistance to acyclovir (rev’iewed in Kcf. 6) can arise by mutation in either the viral tk gene or the D~A-polyrnerase-encoding (~~1) gene (Fig. 1). TK is not csscntial for virus rcplication in cultured cells or certain tissues. Therefore, three different kinds of acyclovir-resistant tk mutations can arise. The simplest type, which confers the greatest resistance of all the single mutations, renders the virus totally devoid of ‘I‘K activity (TK.). SUCII mutations include missense point mutations in active sites, nonsense mutations and single-base deletions or insertions that shift the translational reading frame. A second kind of mutant has an impaired TK, so that the virus is acyclovir resistant, hut nevertheless retains some TK activity (TK partial). As discussed below, TK-partial mutants are not always easy to distinguish from TK-- mutants, and can arise via unusual mechanisms. ‘I’hc third kind of mutant can still phosphorylatc natural substrates, such as thymidine, with fair efficiency, hut its phosphorylation of acyclovir is highly impaired (TK &cd). There are prohabl) fewer sites for this kind of mutation than for those that cause the TK phenotype. Sonic mutants are 1~0th ‘I‘K partial and TK altered; that is, they have reduced activity, but arc more impaired for phosphorylation of acyclovir than for phosphorylation of thymidinc. The HSV polymcrase is essential for viral replication, and so acyclovir-resistant pr)l mutants express an enzyme in infected cells that is less susceptible to inhibition by acyclovir triphosphatc than is the wild-type enzyme (Pol altered) (Fig. I ). A wide variety of sites of mutation in the /xJ gcnc occur (reviewed in Kcf. 7).
Mechanisms of acyclovir action and resistance Acyclovir is a nuclcosidc analog in which guaninc is linked to an acyclic moiety that substitutes for the normal dcoxyribose. Our current understanding of the mechanisms of action of acyclovir and of resistance
Pathogenicity of drug-resistant HSV mutants in immunocompetent mice HSV is readily amenable to studies of pathogenesis in animal models. Numerous animal spccics and routes of infection have been invcstigatcd over the years, and
I set otlt running, take my time ‘Friend of the Devil’ Grateful Dcacl
H
In herpes simplex virus, the simplest path to resistance to the drug acyclovir is a mutation that knocks out the enzyme thymidinc kinasc. Such mutants arc highly attenuated in mouse models of viral pathogen&, but have been reported to be associated with severe disease in immunocompromised patients. This review discusses possible resolutions of this paradox.
REVIEWS
nitely be ascribed to a tk mutation, rather than to an adventitious mutation in another gene”. Various Mechanisms ACV . ACv_PReplication AC’/-P, reports of viruses that arc described -+Iblocked of action as TK-, but that replicate in ganglia enzymes Cell Thymidine DNA and/or reactivate (for example, Kcf. kinase (TK) polymerase 12), arc probably due to the viruses not being truly l’K-, but expressing some TK because of leaky or reverting mutations (reviewed in Ref. 11). Mechanisms TK negative Pol altered As implied above, although of resistance TK partial acyclovir-resistant mutants that TK altered express very low levels of TK (TK Fig. 1. Current understanding of the mechanisms of action of acyclovir (ACV) against herpes partial) are usually attenuated for simplex virus (HSV) and the mechanisms of resistance of HSV towards ACV. ACV-P, acyclovir certain parameters of pathogenicity, monophosphate; ACV-P,, acyclovir triphosphate. See text for details. they nevertheless replicate can acutely in ganglia and/or reactivate The levels of TK activity and pathofrom latency”-“. each has its champions. ‘I’hc most popular species has genesis appear to be correlated’“. TK-altered and pal been the mouse, and immunocompetent strains have mutants arc generally the least attenuated for repligenerally been used. Two assays of pathogenesis are cation at peripheral sites. They can also reactivate most popular. One more-or-less mimics the rare human and have variable degrees of attenuation for lethal condition of herpes encephalitis and involves lethal CNS infection”-“. These two classes of mutants ininfection of the central nervous system (CNS). ‘l’his is clude the most pathogenic of the acyclovir-resistant frequently assayed by measuring the dose of virus that mutants. kills 50%, of inoculated animals after intracranial inoculation (III,,,). The relevance of this kind of assay In summary, TK- mutants arc the most attenuated of the acyclovir-resistant mutants. They are the most to most disease caused by I ISV in humans (which is impaired for replication in certain peripheral tissues, usually not lethal and which more commonly involves such as skin. The major, qualitative difference in pathothe peripheral nervous system and mucocutancous gcnicity between TK- viruses and the other classes of sites) is not certain. acyclovir-resistant mutants is that TK- viruses cannot The second popular assay of pathogenesis more-orless mimics the natural course of infection of humans replicate acutely in or reactivate from murine sensory by HSV, and can be used to study latency. It involves ganglia. How these quantitative and qualitative differences relate to the course of disease in humans is inoculation at a peripheral site, such as the footpad, snout or cornea. The virus replicates at the peripheral still an unanswered question. site and enters nerve terminals. It is then transported Acyclovir-resistant HSV in immunocompetent back to neuronal cell bodies in sensory ganglia where, at lcast in animal models, the virus replicates, producing humans Until recently, there were very few cases of acyclovirmore infectious virus (acute ganglionic replication). ‘Nevertheless, the virus eventually establishes a latent resistant mutants associated with hcrpetic discasc in infection in neurons, such that no infectious virus can immunocompetent patients, and these patients could be treated successfully with acyclovir (for example, be detected in ganglion homogenates. The latent virus Ref. 22). Recently, however, an immunocompctcnt can, however, be reactivated, most commonly by cxpatient with genital herpes that could not be supplanting the ganglia. In certain animal models, rcactipressed by acyclovir was identified”. The disease revation can be achieved in viuo, and sometimes leads curred frequently and the virus that was associated to recrudescent disease. It is, nevertheless, difficult to with these rccurrcnccs was predominantly ‘I’K altered. know whether the behavior of HSV mutants in this Interestingly, the mutant virus retained more phoskind of assay would mimic their behavior during the phorylation activity towards thymidinc and acyclovir course of human disease. than did a TK-altered virus from a patient who was sucWhen acyclovir-resistant HSV mutants have been cessfully trcatcd with acyclovir, and whose acyclovirexamined with either of thcsc kinds of assays, nearly resistant virus did not recurL2. Although there are all arc attenuated to some extent (Table 1). A crucial insufficient data for a strong correlation, this recent point is that the TK- mutation-the simplest mutation cast is consistent with the results obtained in mouse that gives rise to the greatest resistance to acyclovir models of HSV pathogcncsis. results in the greatest attenuation. Truly TK- mutants (such as those containing deletions of critical residues) Acyclovir-resistant HSV in immunocompromised are somewhat deficient for lethal infection of the CNS patients and for replication at certain peripheral sites. They can The vast majority of cases of acyclovir-resistant HSV establish latent infections after peripheral inoculation, infection has come from immunocompromised patients; although with lower efficiency8 .‘I. Most strikingly, indeed, there is a correlation between the dcgrcc of they fail to replicate acutely or reactivate from latency immunosupprcssion and the frequency with which in murinc ganglia”-“. These kinds of defects can defi-
Activation
Inhibition
REVIEWS
resistant viruses arc isolated2’. Although there have been examples of acyclovir-resistant I’K-partial, ‘I’K-altered and pal mutants in immunocompromised patients?’ “, the vast majority of isolates have been dcscribcd as l‘K-, with some reported to be not cxprcssing fulllength TK polypeptidcs 2q-Jx. Certain of these mutants have been isolated from patients who suffered severe, progressive disease. This is puzzling, given the quantitative and qualitative impairments in pathogcnicity of TK- mutants compared with other acyclovir-resistant mutants. There arc two possible resolutions to this paradox. (1) The mouse models of IISV pathogcncsis do not predict the course of human disease, and ‘I’K mutants arc highly pathogenic in immL~nocomprt,miscd humans. (2) ‘I’he mutants described as ‘I’K- are not, in fact, TK--, but rather cxprcss low lcvcls of TK that were not dctcctcd (TK partial) and retain more pathogcnicity. Evidence related to these altcrnativcs has appeared recently. Animal models other than mice
deficient (SCID) strains of mice. Thcsc studies did not examine latency, but rather studied the course of disease after peripheral inoculation4J--~‘h. In general, the mutants were attcnuatcd, but still caused discasc. The most rcccnt study used a well-defined ‘IX mutant inoculated on the cornea of SC111 micP. The mutant caused slowly progrcssivc disease, leading to death. The cause ot death appeared to be related to viral replication in the superficial and deep facial tissues, rather than to CNS lesions, which wcrc not detected. The study found very limited cvidencc of productive infection of ganglionic neurons, and even whcrc this occurred, gene expression only, rather than infectious virus, was observed. ‘I’hcse results suggest that, in the absence of normal immune surveillance, ‘I’K- viruses can replicate slowly in peripheral tissues without the need for latency or reactivation in neurons of sensory ganglia. This very slow course does rescmblc certain cases of acyclovirresistant herpetic discasc in iinmunocomproiniscd patients. I-Iowcvcr, it dots not rcscmble the more rapidly progressing course of acyclovir-resistant discase in other immunocompromiscd patients.
‘I’he first possible resolution to the paradox might come from results in animal models other than mice that suggest that TIC mutants may not differ qualitatively from other acyclovir-resistant mutants. In guinea-pig Expression of TK from apparently TK- mutants and rabbit, truly TK- HSV mutants arc quantitatively An alternative explanation for why viruses described attenuated for pathogcnicity, but ncverthclcss can as TK- are associated with severe, progressive human disease is that such viruses actually cxprcss low levels establish and rcactivatc from latent infections in the ab“),““. These animal models of ‘I’K. A potential source of low levels of ‘I’K in a clinisencc of inimunosuppression cal isolate is virus heterogeneity. TK mutants that are of IISV might be considered to mimic human disease not dclctions can revert, and the presence of a low better than does the mouse model, in that the virus reactivates spontaneously; however, this also raises level of rcvertants can lcad to replication in and/or reactivation from gangliaJ1. Expcrinicntally constructed the question of whether the virus dctcctcd is due to mixtures of TK-proficient and ‘1%deficient viruses true reactivation or to smoldering infection. Whether latency and reactivation of the TK- mutants in these complement each other for pathogcnicity and/or animals occurs in neurons or in non-neuronal cells resistance to acyclovir in micc’x~45.4~‘~‘q. As acyclovirresistance mutations in the pal gene allow the growth has also not been established. Few stud& have been performed in primatc models. In one cast, a TK virus of othcrwisc susceptible viruses in the prcscncc of that also contained other lesions failed to rcactivatc acyclovir”‘, drug-resistant ~xA mutants and viral from immunocompctcnt owl monkeys, while a derivative of this mutant in which Table 1. Pathogenicity of acyclovir-resistant mutants in micea expression of ‘I’K was restored did rcactivate at low frequcncy4’. A requirement for Reactivation Relative Replication at TK in an otherwise wild-type virus for Type of virus virulence (o/b) b from latency ( %)d periphery (%)’ reactivation in this model can neither bc _ excluded nor confirmed by these data. It Wild type 100 100 70-100 might be intcrcsting to explore systcmatiTK0.01-3 10-100 0 tally the behaviors of different classes of TK partial 3-100 20-100 10-100 acyclovir-resistant mutants in these animal TK altered 10 100 50-100 models other than mice. Polymerase I-IO 20-100 50-100 Studies
in immunocompromised
mice
Other studies have investigated the behavior of ‘I‘K mutants in immL~nocompromised mice and other species. In an early study, a mutant that was defined as TKby cnzymc assay failed to replicate in or reactivate from murine ganglia after immunosuppression induced by cyclophosphalnidc~~. Similar results have lxwi obtained using a 'I‘Kdeletion mutant”‘. Other studies have used athymic or severe combined irnmuno-
*‘Compiled from various references, including 6,8-11,14,16-20. The ranges reflect different behaviors of different mutants in each class due to molecular differences among the mutants, different wild-type parental strains or differing routes of inoculation. “Relative virulence after intracerebral inoculation. The wild type is set at 100%. Mutant viruses with, for example, tenfold higher LD,, values are expressed as 10%; with 100.fold higher, as 1%. ‘Replication at the peripheral site of inoculation, expressed as a percentage of the wild-type titer. “Proportion of sensory ganglia that reactivate virus after inoculation at the peripheral site innervated by those ganglia. _~_~ -
REVIEWS
heterogeneity may be important in pathogencsis. There arc several examples of clinical isolntcs that show substantial hcterogeneity’“,“,‘i’,~‘. Such hctcrogcneity has not been observed in certain cnscs in which TKaltcrcd or pal mutants have been found2’,?‘, but has not in general been ruled out in clinical isolates reported to be TK -. Even in 3 pure population of acyclovir-resistant virus, it can hc difficult to detect low lcvcls of I‘K, in part because assays of TK activity can bc insensitive. In certain of the studies of viruses recovered from patients, the limit of detection of TK activity was k5’%, (Refs 3 l-33,35). Moreover, if a mutation not only alters the level of activity, but also the characteristics of the enzyme (for example, the salt optimum), then it might be even more difficult to detect the activity in vitro. l’his is critical because, in mouse models, viruses retaining 4% TK activity can retain the ability to rcactivatc from latency”-“. Translatlonal frame-shifting to express TK Nevertheless, certain isolates from patients with acyclovir-resistant discasc would seem, at first glance, to be clearly 'IX because they do not express full-length ‘TK polypcptides and/or contain frame-shift mutations in their tk gcncslxl-“.‘$. I Iowcvcr, the relevance of failurc to detect full-length TK depends on the sensitivity of the assays used, which was not documcntcd. Moreover, at least some truncated ‘I‘K polypeptidcs retain it would still IX presumed TK activity 16. Nevertheless, that mutants with frame-shift mutations”.‘” would IX TK . This presumption has been challcngcd by the study of an acyclovir-rcsistnnt clinical isolate that contains a sing+-base insertion in its tk gcnc’ ‘. Despite the insertion, the mutant not only synthcsizcs the predictcd truncated polypeptidc, but also low lcvcls of full-length ‘I‘K polypeptide, due to frame-shifting during translation. I’hc mutant also retains sufficient TK activity to reactivate from latent infections of murinc +nsory ganglia. This suggests that, during the cvolution of acyclovir-resistant virus in patients, there may IX sclcction for unusual mechanisms to allow sufficient cxprcssion of ‘I’K to lead to pathogenesis. A potential unifying hypothesis The indolent course of infection of WI) mice by TK HSV mimics the kind of ncyclovir-resistant disease that is found in certain immunocompromiscd patients, while other patients have much more rapid and scvcrc courses. A hypothesis that might unify the data available is that a pure population of truly TK HSV might cause indolent infections in humans, while the more rapidly progressing infections might bc caused by other classes of acyclovir-resistant mutants or by ‘Thorwgh and scnsitivc heterogcncous mixtures. assays of acyclovir-resistant clinical isolates, correlated with clinical outcomes, might address this hypothesis, and the study of drug-resistant virus associated with recurrences may be particularly informative. Finally, studies of pathogenic, acyclovir-resistant isolates from patients might uncover further unusual mechanisms, such as translational frame-shifting, that
allow this virus to run away from drug, but not so fast as to lose pathogcnicity. Acknowledgements I rhank rrsearchcr\ in the ii&l oi 1ISVlntcnq nnd drug rcsistnm \vho make it w srimulatq. (kmt support irom the SII I (KOl r\12612h. UOI Al26077 ,~nd I’0 I N24010) k ncknowlcdgcd gratefully. References 1 Richman, D.1). (lYY4i Trends Microhro/. 2: 401-407 2 Hirsch, ;1I.S. d Schoolcy. R. I’. [lYXY] N. Eug/. 1. Mcrl. 320. 313-314
6 I.nrdcr, B.A. nnd Darby, G. (1984) Adid Rcs.4, l-42 7 Cocn, D..LI. ( 1992) Semru. l:ilo/. 3, 3- 12 8 Mclkrmott, MR. et al. ( 1984)j. Vird. 51, 747-753 9 Coen, I)..\l. et al. (1989) I’roc. Nut/ h7d. Ser. USA 86, 4735-4739 10 Ebtathiou, S. rf a/. ( IYXY) /. Gerz. \‘rro/. 70. X69-879 11 Jacobson, J.G. et 01.( I YY.3) /. V’/YO/. 67. 6YO3-6908 12 Wilcox. CL., Crnic, I..S. and Pi/u, L.I. (1992) Virology 187. 34X-352 13 Gordon, Y. cl a/. (19X3) Arch. bum/. 76, 39-4’) 14 Darb):, G., Churcher, M.J. nnd Larder, K.A. (19X4)/. \‘irr~/. 50, 83X-846 15 Scnrb, .\.I:., hlupnicr. B. and Koi7man. B. (19X.5) 1. \~A. 55, 410-416 16 COCII, D.M. et a/. (1989) \~~~o/og~ 168, 221-231 17 f Iwang, C.H.C. rl 671.(1994) Proc. Gt/ Acud. SCI. MA 91. 456 I-4565 18 T’enrrr, R.R., Rcscl, S. and Dunstan, MI<. ( IYXl) \‘~ro/og)’ 112, 32X-341 19 Darly, G., I~ield. I1.J. nnd Snli,burl;, 5..4. (198I)N~twe LXY. Xl-X3 20 I.‘udcr, B.A. and I);uhy, G. (1 YXT) \:rro/qq 146,262-271 21 Ficld. 1r.J. and Cocn, D.M. (lYX611. \‘IYo/. 60. 2X6-288 22 Ellis, N.11. et ,I/. ( I Y 87) Antimr~roh. Agcwts Chemofher. 3 I, 1117-1125 23 Englund, J. et a/. :1990)hn. Irzterrz. Med. 112, 416-422 24 Sack$, S.L.. el al. (lYX9) i\wz. Itztc~r7. Med. I1 1, X93-XYY 2.5 I%rkcr, A.(:. et al. ( 1987) Larcet il. 146 I 26 Collins. I’. of a/. ( I YXY) /. CA. b’rrol. 70. V-382 27 Sugicr, F. et u/. (1 Y Y 1) htiur~al Chm. Cbemother. 2, 295-302 2X I hvang, (XC., Rufinrr. K.I.. and Cocn, I).M. ( IYY~J 1. VI,O/. 66. 17’4-1776
30 31 32 33 34 35 36 37 38 39 40 41 42
(SuppI. IO, 127-134 Youle, .Il.M. et a/. ( IYXX) Lawet ii, 341-342 Erlich, K.S. ct a/. i 19X9) ;V. I:rq/. /. .blcd. 320, 293-296 Chntis, PA. et a/. (1 YXY) N. I,q/. /. Med. 320, 297-100 GateIcy, A. et (I/. ( 1990) 1. IIZ~~CI Dis. I6 I, 71 l-71.7 I.pngman, P. et al. (199Oj /. I+-f. Ih. 162: 244-24X Birch, C.J. et a/. (IYYO) /. Infect. DrS. 162. 73 l-734 Snirin. S. et a/. (IYYl) g. Eng/.j. h,Jed. 325. jjl-SST Chata PA. 2nd Crumpckcr. C.S. (1991 j Virology I X0, 79 1-797 Pali], C. et al. (lYY2i l’irr~s Kes. 2.5: 133-144 St.mbcrry. I .I<.,Kit, 5. and Myers, .\I.(;. :‘lYX.<) j. Viral. SS, 322-128 Alcigmcr. B. et a/. ! 19881 \‘iro/ogy 162, 25 l-254 .Clcigmer, I\. et ,I/. i,I990)/. /rz{rcl. /)rs. 162, 3 13-32 I I’ricc. R.W. aid Khu, A. I I YSl) I&t /numuz.34, .$‘I-580
0
43 Rleignier, K., lmgnecker,
It. ant1 Itoizman,
B.
(1988)/, /tr/ect.
Dis. 1.58,602-614 44 Sihrack, CD. et al. (1982) 1. Infect. Dis. 146. 673-682 45 Elhs, b1.N. ct J/. i 19891 Ar~tmc~o6. Agem Chetmther.
3.3,
304-310
46 Valyi-SSlgy3 ‘1’.et II/. (1994) \‘iro/ogy 199,4X4-490 47 IGcld,I I.J. ( lYX2)r\~ztimrcroh.Agc~ls Che~?lr~ther. 2 I, 744-752
1’ I N
48 Field, H.J. and l.~y, E. (19843A~~tivrrnlKcs. 4, 4.3-52 49 Coen, D.M. and Schaffcr, PA. (19X0) I’M. Nut/ Acmf. SC;. MA 77.226.5-2269 50 Christophers? 1. and Sutton, R.S.1’. (1987) /. Arztiwzicrob. Cbemotber. 20, 38Y-3Y8
51 Mnrtin, j.1,. it al. 181-187
(198.5) ArrtunicroI~ &c~zts
Chemothcr.
28.
Computer processing of microscopic images of bacteria: morphometry and fluorimetry Michael H.F. Wilkinson, Gijsbert 1. Jansen and Dirk van der Waaij
M
icroscopic
computerized
image processing has proved to bc a valuable tool in studying complex microbial ecosystems, whether tcrrcstrial’, planktonic?.’ or bound to various surfaces’-“. Considering planktonic bacteria, Hjcrrnset? and Sicracki el al.’ (among others) have done much important work in detcrmining numbers of organisms, their size distribution and total biomass, and most methods have focused on thcsc variables. These parameters yield important clues to the state of an ccosystcm, and any change in any of thcsc parameters indicates a change in the ecosystem. Our group is intcrcstcd in the ccosysteni of the human intestinal microflora. The human intcstine contains over 400 spccics of predominantly anaerobic bacteria, which live in apparent harmony with the host and arc tolerated by the immune system. This microflora is thought to be responsible for the resistance to colonization of the human intestine; it forms a first line of defence against intruding pathogcns-. If this microflora is wiped out, for example, by antimicrobial therapy, overgrowth by drug-resistant microbes can result. IJsing computer image analysis, we have attempted to study both the microflora itself and its interaction with the immune system of the host”.‘.
Several techniques that USC computer analysis of microscopic images have been developed to study the complicated microbial flora in the human intestine, including measuring the shape and fluorescence intensity of bacteria. These tcchniqucs allow rapid assessment of changes in the intestinal flora and could apply equally to other complex microbial ecosystems.
We have focused on shapex.“‘~” and fluorescence”.“,’ 1nicasurements on microscopic slides of bacteria isolated directly from facces; however, the methods that WC have wised could probably be applied to many other microbial ecosystems. In this article, WC describe these methods and the results that we have obtained using them, and discuss some future prospects for microscopic image analysis in microbiology. The GRID system Our microscopic image analysis system has been named Groningcn reduction of image data (GRII))S~‘~. The system consists of an industrial video camera that has been modificd to provide the long exposure
times that arc needed for fluorcsccnce measurements’2, mounted on top of a microscope and linked to a computer, which is equipped with electronic circuits to allow video images to be captured in computcr memory. Once images have been stored, they can subsequently be processed using standard digital image analysis tcchniqucs’j and personal computers and software that we have developed. Morphometry A human observer can readily distinguish bacteria on an image such as that in Fig. la, but computers find the task more difficult. Bcforc any mcasurcmcnt can bc made, the image must bc divided into objects (the bacteria) and background. We use an automated segmentation technique’” similar to those that have been dcscribcd prcviously’5-‘7. ‘[‘he image is segmented and convcrtcd into a binary image (Fig. I b), revealing objects in the form of white connected regions. The surface area, perimctcr, width, length and other morphological paramctcrs of each connected region can be obtained readily”. Obtaining the true size or volume of each bacterium is slightly more complicated. No imaging system is completely free of distortion, and many staining and slide-l”eparatioli tcchniques distort the original shape of
I 0
N
Acyclovir-resistant, pathogenic herpesviruses Donald M. Coen hut I
‘erpesviruscs are in]portant human pathotheir in.gerls and fections can be particularly scvcrc in immutlocompromised patients, such as those with AIDS. I’crhaps the most striking successes in antiviral chemotherapy have ._come in treating herpesvIrus Intcctlons, cspccially with the development of acyclovir against herpes simplex virus (HSV). However, as is typical of successful antimicrobial agents, drug resistance has reared its head’m4. Although its composers may not have had this in mind, the song ‘Friend of the Devil’ is pertinent to two intertwined aspects of drug resistance and pathogenicity of HSV. ‘l’he first is that the resistance of any pathogen to antimicrobial drugs is a trade-off between drug rcsistancc and pathogenicity. ‘I‘his is particularly true for I ISV, where drug resistance requires mutation in viral genes that have evolved to assure success as a pathogen. Thus, to cause clinically significant drug resistance, HSV must evade the drug (‘set out running’), but must also retain sufficient function in gent products important for pathogcncsis (‘take its time’) to cause discasc. Intertwined with this trade-off is probably the most important aspect of HSV pathogencsis - latency - in which I ISV stops replicating (‘takes its time’) and enters a noninfectious form with limited gene cxprcssion in sensory neurons, allowing it to avoid immune surveillance. HSV can then reactivate from latency to cause disease and allow transmission to a new host. 1Iowevcr, the simplest route to resistance to acyclovir - thymidinekinasc-negative (TK-) mutation - not only attenuates pathogcncsis generally, but also prevents reactivation from latency in animal models. Even so, ‘I’K- mutants have been reported to be associated with cases of disease caused by IISV that is resistant to therapy with acyclovir. This review addresses this paradox.
to the drug is summarized in Fig. 1. Acyclovir is first activated by phosphorylation, which is performed much more efficiently by the HSV TK than it is by cellular enzymes. Cellular enzymes then convert acyclovir monophosphate to its triphosphate, which is a more potent inhibitor of HSV DNA polymcracc than it is of cellular enzymes (rcvicwed in Ref. 5). Accordingly, resistance to acyclovir (rev’iewed in Kcf. 6) can arise by mutation in either the viral tk gene or the D~A-polyrnerase-encoding (~~1) gene (Fig. 1). TK is not csscntial for virus rcplication in cultured cells or certain tissues. Therefore, three different kinds of acyclovir-resistant tk mutations can arise. The simplest type, which confers the greatest resistance of all the single mutations, renders the virus totally devoid of ‘I‘K activity (TK.). SUCII mutations include missense point mutations in active sites, nonsense mutations and single-base deletions or insertions that shift the translational reading frame. A second kind of mutant has an impaired TK, so that the virus is acyclovir resistant, hut nevertheless retains some TK activity (TK partial). As discussed below, TK-partial mutants are not always easy to distinguish from TK-- mutants, and can arise via unusual mechanisms. ‘I’hc third kind of mutant can still phosphorylatc natural substrates, such as thymidine, with fair efficiency, hut its phosphorylation of acyclovir is highly impaired (TK &cd). There are prohabl) fewer sites for this kind of mutation than for those that cause the TK phenotype. Sonic mutants are 1~0th ‘I‘K partial and TK altered; that is, they have reduced activity, but arc more impaired for phosphorylation of acyclovir than for phosphorylation of thymidinc. The HSV polymcrase is essential for viral replication, and so acyclovir-resistant pr)l mutants express an enzyme in infected cells that is less susceptible to inhibition by acyclovir triphosphatc than is the wild-type enzyme (Pol altered) (Fig. I ). A wide variety of sites of mutation in the /xJ gcnc occur (reviewed in Kcf. 7).
Mechanisms of acyclovir action and resistance Acyclovir is a nuclcosidc analog in which guaninc is linked to an acyclic moiety that substitutes for the normal dcoxyribose. Our current understanding of the mechanisms of action of acyclovir and of resistance
Pathogenicity of drug-resistant HSV mutants in immunocompetent mice HSV is readily amenable to studies of pathogenesis in animal models. Numerous animal spccics and routes of infection have been invcstigatcd over the years, and
I set otlt running, take my time ‘Friend of the Devil’ Grateful Dcacl
H
In herpes simplex virus, the simplest path to resistance to the drug acyclovir is a mutation that knocks out the enzyme thymidinc kinasc. Such mutants arc highly attenuated in mouse models of viral pathogen&, but have been reported to be associated with severe disease in immunocompromised patients. This review discusses possible resolutions of this paradox.
REVIEWS
nitely be ascribed to a tk mutation, rather than to an adventitious mutation in another gene”. Various Mechanisms ACV . ACv_PReplication AC’/-P, reports of viruses that arc described -+Iblocked of action as TK-, but that replicate in ganglia enzymes Cell Thymidine DNA and/or reactivate (for example, Kcf. kinase (TK) polymerase 12), arc probably due to the viruses not being truly l’K-, but expressing some TK because of leaky or reverting mutations (reviewed in Ref. 11). Mechanisms TK negative Pol altered As implied above, although of resistance TK partial acyclovir-resistant mutants that TK altered express very low levels of TK (TK Fig. 1. Current understanding of the mechanisms of action of acyclovir (ACV) against herpes partial) are usually attenuated for simplex virus (HSV) and the mechanisms of resistance of HSV towards ACV. ACV-P, acyclovir certain parameters of pathogenicity, monophosphate; ACV-P,, acyclovir triphosphate. See text for details. they nevertheless replicate can acutely in ganglia and/or reactivate The levels of TK activity and pathofrom latency”-“. each has its champions. ‘I’hc most popular species has genesis appear to be correlated’“. TK-altered and pal been the mouse, and immunocompetent strains have mutants arc generally the least attenuated for repligenerally been used. Two assays of pathogenesis are cation at peripheral sites. They can also reactivate most popular. One more-or-less mimics the rare human and have variable degrees of attenuation for lethal condition of herpes encephalitis and involves lethal CNS infection”-“. These two classes of mutants ininfection of the central nervous system (CNS). ‘l’his is clude the most pathogenic of the acyclovir-resistant frequently assayed by measuring the dose of virus that mutants. kills 50%, of inoculated animals after intracranial inoculation (III,,,). The relevance of this kind of assay In summary, TK- mutants arc the most attenuated of the acyclovir-resistant mutants. They are the most to most disease caused by I ISV in humans (which is impaired for replication in certain peripheral tissues, usually not lethal and which more commonly involves such as skin. The major, qualitative difference in pathothe peripheral nervous system and mucocutancous gcnicity between TK- viruses and the other classes of sites) is not certain. acyclovir-resistant mutants is that TK- viruses cannot The second popular assay of pathogenesis more-orless mimics the natural course of infection of humans replicate acutely in or reactivate from murine sensory by HSV, and can be used to study latency. It involves ganglia. How these quantitative and qualitative differences relate to the course of disease in humans is inoculation at a peripheral site, such as the footpad, snout or cornea. The virus replicates at the peripheral still an unanswered question. site and enters nerve terminals. It is then transported Acyclovir-resistant HSV in immunocompetent back to neuronal cell bodies in sensory ganglia where, at lcast in animal models, the virus replicates, producing humans Until recently, there were very few cases of acyclovirmore infectious virus (acute ganglionic replication). ‘Nevertheless, the virus eventually establishes a latent resistant mutants associated with hcrpetic discasc in infection in neurons, such that no infectious virus can immunocompetent patients, and these patients could be treated successfully with acyclovir (for example, be detected in ganglion homogenates. The latent virus Ref. 22). Recently, however, an immunocompctcnt can, however, be reactivated, most commonly by cxpatient with genital herpes that could not be supplanting the ganglia. In certain animal models, rcactipressed by acyclovir was identified”. The disease revation can be achieved in viuo, and sometimes leads curred frequently and the virus that was associated to recrudescent disease. It is, nevertheless, difficult to with these rccurrcnccs was predominantly ‘I’K altered. know whether the behavior of HSV mutants in this Interestingly, the mutant virus retained more phoskind of assay would mimic their behavior during the phorylation activity towards thymidinc and acyclovir course of human disease. than did a TK-altered virus from a patient who was sucWhen acyclovir-resistant HSV mutants have been cessfully trcatcd with acyclovir, and whose acyclovirexamined with either of thcsc kinds of assays, nearly resistant virus did not recurL2. Although there are all arc attenuated to some extent (Table 1). A crucial insufficient data for a strong correlation, this recent point is that the TK- mutation-the simplest mutation cast is consistent with the results obtained in mouse that gives rise to the greatest resistance to acyclovir models of HSV pathogcncsis. results in the greatest attenuation. Truly TK- mutants (such as those containing deletions of critical residues) Acyclovir-resistant HSV in immunocompromised are somewhat deficient for lethal infection of the CNS patients and for replication at certain peripheral sites. They can The vast majority of cases of acyclovir-resistant HSV establish latent infections after peripheral inoculation, infection has come from immunocompromised patients; although with lower efficiency8 .‘I. Most strikingly, indeed, there is a correlation between the dcgrcc of they fail to replicate acutely or reactivate from latency immunosupprcssion and the frequency with which in murinc ganglia”-“. These kinds of defects can defi-
Activation
Inhibition
REVIEWS
resistant viruses arc isolated2’. Although there have been examples of acyclovir-resistant I’K-partial, ‘I’K-altered and pal mutants in immunocompromised patients?’ “, the vast majority of isolates have been dcscribcd as l‘K-, with some reported to be not cxprcssing fulllength TK polypeptidcs 2q-Jx. Certain of these mutants have been isolated from patients who suffered severe, progressive disease. This is puzzling, given the quantitative and qualitative impairments in pathogcnicity of TK- mutants compared with other acyclovir-resistant mutants. There arc two possible resolutions to this paradox. (1) The mouse models of IISV pathogcncsis do not predict the course of human disease, and ‘I’K mutants arc highly pathogenic in immL~nocomprt,miscd humans. (2) ‘I’he mutants described as ‘I’K- are not, in fact, TK--, but rather cxprcss low lcvcls of TK that were not dctcctcd (TK partial) and retain more pathogcnicity. Evidence related to these altcrnativcs has appeared recently. Animal models other than mice
deficient (SCID) strains of mice. Thcsc studies did not examine latency, but rather studied the course of disease after peripheral inoculation4J--~‘h. In general, the mutants were attcnuatcd, but still caused discasc. The most rcccnt study used a well-defined ‘IX mutant inoculated on the cornea of SC111 micP. The mutant caused slowly progrcssivc disease, leading to death. The cause ot death appeared to be related to viral replication in the superficial and deep facial tissues, rather than to CNS lesions, which wcrc not detected. The study found very limited cvidencc of productive infection of ganglionic neurons, and even whcrc this occurred, gene expression only, rather than infectious virus, was observed. ‘I’hcse results suggest that, in the absence of normal immune surveillance, ‘I’K- viruses can replicate slowly in peripheral tissues without the need for latency or reactivation in neurons of sensory ganglia. This very slow course does rescmblc certain cases of acyclovirresistant herpetic discasc in iinmunocomproiniscd patients. I-Iowcvcr, it dots not rcscmble the more rapidly progressing course of acyclovir-resistant discase in other immunocompromiscd patients.
‘I’he first possible resolution to the paradox might come from results in animal models other than mice that suggest that TIC mutants may not differ qualitatively from other acyclovir-resistant mutants. In guinea-pig Expression of TK from apparently TK- mutants and rabbit, truly TK- HSV mutants arc quantitatively An alternative explanation for why viruses described attenuated for pathogcnicity, but ncverthclcss can as TK- are associated with severe, progressive human disease is that such viruses actually cxprcss low levels establish and rcactivatc from latent infections in the ab“),““. These animal models of ‘I’K. A potential source of low levels of ‘I’K in a clinisencc of inimunosuppression cal isolate is virus heterogeneity. TK mutants that are of IISV might be considered to mimic human disease not dclctions can revert, and the presence of a low better than does the mouse model, in that the virus reactivates spontaneously; however, this also raises level of rcvertants can lcad to replication in and/or reactivation from gangliaJ1. Expcrinicntally constructed the question of whether the virus dctcctcd is due to mixtures of TK-proficient and ‘1%deficient viruses true reactivation or to smoldering infection. Whether latency and reactivation of the TK- mutants in these complement each other for pathogcnicity and/or animals occurs in neurons or in non-neuronal cells resistance to acyclovir in micc’x~45.4~‘~‘q. As acyclovirresistance mutations in the pal gene allow the growth has also not been established. Few stud& have been performed in primatc models. In one cast, a TK virus of othcrwisc susceptible viruses in the prcscncc of that also contained other lesions failed to rcactivatc acyclovir”‘, drug-resistant ~xA mutants and viral from immunocompctcnt owl monkeys, while a derivative of this mutant in which Table 1. Pathogenicity of acyclovir-resistant mutants in micea expression of ‘I’K was restored did rcactivate at low frequcncy4’. A requirement for Reactivation Relative Replication at TK in an otherwise wild-type virus for Type of virus virulence (o/b) b from latency ( %)d periphery (%)’ reactivation in this model can neither bc _ excluded nor confirmed by these data. It Wild type 100 100 70-100 might be intcrcsting to explore systcmatiTK0.01-3 10-100 0 tally the behaviors of different classes of TK partial 3-100 20-100 10-100 acyclovir-resistant mutants in these animal TK altered 10 100 50-100 models other than mice. Polymerase I-IO 20-100 50-100 Studies
in immunocompromised
mice
Other studies have investigated the behavior of ‘I‘K mutants in immL~nocompromised mice and other species. In an early study, a mutant that was defined as TKby cnzymc assay failed to replicate in or reactivate from murine ganglia after immunosuppression induced by cyclophosphalnidc~~. Similar results have lxwi obtained using a 'I‘Kdeletion mutant”‘. Other studies have used athymic or severe combined irnmuno-
*‘Compiled from various references, including 6,8-11,14,16-20. The ranges reflect different behaviors of different mutants in each class due to molecular differences among the mutants, different wild-type parental strains or differing routes of inoculation. “Relative virulence after intracerebral inoculation. The wild type is set at 100%. Mutant viruses with, for example, tenfold higher LD,, values are expressed as 10%; with 100.fold higher, as 1%. ‘Replication at the peripheral site of inoculation, expressed as a percentage of the wild-type titer. “Proportion of sensory ganglia that reactivate virus after inoculation at the peripheral site innervated by those ganglia. _~_~ -
REVIEWS
heterogeneity may be important in pathogencsis. There arc several examples of clinical isolntcs that show substantial hcterogeneity’“,“,‘i’,~‘. Such hctcrogcneity has not been observed in certain cnscs in which TKaltcrcd or pal mutants have been found2’,?‘, but has not in general been ruled out in clinical isolates reported to be TK -. Even in 3 pure population of acyclovir-resistant virus, it can hc difficult to detect low lcvcls of I‘K, in part because assays of TK activity can bc insensitive. In certain of the studies of viruses recovered from patients, the limit of detection of TK activity was k5’%, (Refs 3 l-33,35). Moreover, if a mutation not only alters the level of activity, but also the characteristics of the enzyme (for example, the salt optimum), then it might be even more difficult to detect the activity in vitro. l’his is critical because, in mouse models, viruses retaining 4% TK activity can retain the ability to rcactivatc from latency”-“. Translatlonal frame-shifting to express TK Nevertheless, certain isolates from patients with acyclovir-resistant discasc would seem, at first glance, to be clearly 'IX because they do not express full-length ‘TK polypcptides and/or contain frame-shift mutations in their tk gcncslxl-“.‘$. I Iowcvcr, the relevance of failurc to detect full-length TK depends on the sensitivity of the assays used, which was not documcntcd. Moreover, at least some truncated ‘I‘K polypeptidcs retain it would still IX presumed TK activity 16. Nevertheless, that mutants with frame-shift mutations”.‘” would IX TK . This presumption has been challcngcd by the study of an acyclovir-rcsistnnt clinical isolate that contains a sing+-base insertion in its tk gcnc’ ‘. Despite the insertion, the mutant not only synthcsizcs the predictcd truncated polypeptidc, but also low lcvcls of full-length ‘I‘K polypeptide, due to frame-shifting during translation. I’hc mutant also retains sufficient TK activity to reactivate from latent infections of murinc +nsory ganglia. This suggests that, during the cvolution of acyclovir-resistant virus in patients, there may IX sclcction for unusual mechanisms to allow sufficient cxprcssion of ‘I’K to lead to pathogenesis. A potential unifying hypothesis The indolent course of infection of WI) mice by TK HSV mimics the kind of ncyclovir-resistant disease that is found in certain immunocompromiscd patients, while other patients have much more rapid and scvcrc courses. A hypothesis that might unify the data available is that a pure population of truly TK HSV might cause indolent infections in humans, while the more rapidly progressing infections might bc caused by other classes of acyclovir-resistant mutants or by ‘Thorwgh and scnsitivc heterogcncous mixtures. assays of acyclovir-resistant clinical isolates, correlated with clinical outcomes, might address this hypothesis, and the study of drug-resistant virus associated with recurrences may be particularly informative. Finally, studies of pathogenic, acyclovir-resistant isolates from patients might uncover further unusual mechanisms, such as translational frame-shifting, that
allow this virus to run away from drug, but not so fast as to lose pathogcnicity. Acknowledgements I rhank rrsearchcr\ in the ii&l oi 1ISVlntcnq nnd drug rcsistnm \vho make it w srimulatq. (kmt support irom the SII I (KOl r\12612h. UOI Al26077 ,~nd I’0 I N24010) k ncknowlcdgcd gratefully. References 1 Richman, D.1). (lYY4i Trends Microhro/. 2: 401-407 2 Hirsch, ;1I.S. d Schoolcy. R. I’. [lYXY] N. Eug/. 1. Mcrl. 320. 313-314
6 I.nrdcr, B.A. nnd Darby, G. (1984) Adid Rcs.4, l-42 7 Cocn, D..LI. ( 1992) Semru. l:ilo/. 3, 3- 12 8 Mclkrmott, MR. et al. ( 1984)j. Vird. 51, 747-753 9 Coen, I)..\l. et al. (1989) I’roc. Nut/ h7d. Ser. USA 86, 4735-4739 10 Ebtathiou, S. rf a/. ( IYXY) /. Gerz. \‘rro/. 70. X69-879 11 Jacobson, J.G. et 01.( I YY.3) /. V’/YO/. 67. 6YO3-6908 12 Wilcox. CL., Crnic, I..S. and Pi/u, L.I. (1992) Virology 187. 34X-352 13 Gordon, Y. cl a/. (19X3) Arch. bum/. 76, 39-4’) 14 Darb):, G., Churcher, M.J. nnd Larder, K.A. (19X4)/. \‘irr~/. 50, 83X-846 15 Scnrb, .\.I:., hlupnicr. B. and Koi7man. B. (19X.5) 1. \~A. 55, 410-416 16 COCII, D.M. et a/. (1989) \~~~o/og~ 168, 221-231 17 f Iwang, C.H.C. rl 671.(1994) Proc. Gt/ Acud. SCI. MA 91. 456 I-4565 18 T’enrrr, R.R., Rcscl, S. and Dunstan, MI<. ( IYXl) \‘~ro/og)’ 112, 32X-341 19 Darly, G., I~ield. I1.J. nnd Snli,burl;, 5..4. (198I)N~twe LXY. Xl-X3 20 I.‘udcr, B.A. and I);uhy, G. (1 YXT) \:rro/qq 146,262-271 21 Ficld. 1r.J. and Cocn, D.M. (lYX611. \‘IYo/. 60. 2X6-288 22 Ellis, N.11. et ,I/. ( I Y 87) Antimr~roh. Agcwts Chemofher. 3 I, 1117-1125 23 Englund, J. et a/. :1990)hn. Irzterrz. Med. 112, 416-422 24 Sack$, S.L.. el al. (lYX9) i\wz. Itztc~r7. Med. I1 1, X93-XYY 2.5 I%rkcr, A.(:. et al. ( 1987) Larcet il. 146 I 26 Collins. I’. of a/. ( I YXY) /. CA. b’rrol. 70. V-382 27 Sugicr, F. et u/. (1 Y Y 1) htiur~al Chm. Cbemother. 2, 295-302 2X I hvang, (XC., Rufinrr. K.I.. and Cocn, I).M. ( IYY~J 1. VI,O/. 66. 17’4-1776
30 31 32 33 34 35 36 37 38 39 40 41 42
(SuppI. IO, 127-134 Youle, .Il.M. et a/. ( IYXX) Lawet ii, 341-342 Erlich, K.S. ct a/. i 19X9) ;V. I:rq/. /. .blcd. 320, 293-296 Chntis, PA. et a/. (1 YXY) N. I,q/. /. Med. 320, 297-100 GateIcy, A. et (I/. ( 1990) 1. IIZ~~CI Dis. I6 I, 71 l-71.7 I.pngman, P. et al. (199Oj /. I+-f. Ih. 162: 244-24X Birch, C.J. et a/. (IYYO) /. Infect. DrS. 162. 73 l-734 Snirin. S. et a/. (IYYl) g. Eng/.j. h,Jed. 325. jjl-SST Chata PA. 2nd Crumpckcr. C.S. (1991 j Virology I X0, 79 1-797 Pali], C. et al. (lYY2i l’irr~s Kes. 2.5: 133-144 St.mbcrry. I .I<.,Kit, 5. and Myers, .\I.(;. :‘lYX.<) j. Viral. SS, 322-128 Alcigmcr. B. et a/. ! 19881 \‘iro/ogy 162, 25 l-254 .Clcigmer, I\. et ,I/. i,I990)/. /rz{rcl. /)rs. 162, 3 13-32 I I’ricc. R.W. aid Khu, A. I I YSl) I&t /numuz.34, .$‘I-580
0
43 Rleignier, K., lmgnecker,
It. ant1 Itoizman,
B.
(1988)/, /tr/ect.
Dis. 1.58,602-614 44 Sihrack, CD. et al. (1982) 1. Infect. Dis. 146. 673-682 45 Elhs, b1.N. ct J/. i 19891 Ar~tmc~o6. Agem Chetmther.
3.3,
304-310
46 Valyi-SSlgy3 ‘1’.et II/. (1994) \‘iro/ogy 199,4X4-490 47 IGcld,I I.J. ( lYX2)r\~ztimrcroh.Agc~ls Che~?lr~ther. 2 I, 744-752
1’ I N
48 Field, H.J. and l.~y, E. (19843A~~tivrrnlKcs. 4, 4.3-52 49 Coen, D.M. and Schaffcr, PA. (19X0) I’M. Nut/ Acmf. SC;. MA 77.226.5-2269 50 Christophers? 1. and Sutton, R.S.1’. (1987) /. Arztiwzicrob. Cbemotber. 20, 38Y-3Y8
51 Mnrtin, j.1,. it al. 181-187
(198.5) ArrtunicroI~ &c~zts
Chemothcr.
28.
Computer processing of microscopic images of bacteria: morphometry and fluorimetry Michael H.F. Wilkinson, Gijsbert 1. Jansen and Dirk van der Waaij
M
icroscopic
computerized
image processing has proved to bc a valuable tool in studying complex microbial ecosystems, whether tcrrcstrial’, planktonic?.’ or bound to various surfaces’-“. Considering planktonic bacteria, Hjcrrnset? and Sicracki el al.’ (among others) have done much important work in detcrmining numbers of organisms, their size distribution and total biomass, and most methods have focused on thcsc variables. These parameters yield important clues to the state of an ccosystcm, and any change in any of thcsc parameters indicates a change in the ecosystem. Our group is intcrcstcd in the ccosysteni of the human intestinal microflora. The human intcstine contains over 400 spccics of predominantly anaerobic bacteria, which live in apparent harmony with the host and arc tolerated by the immune system. This microflora is thought to be responsible for the resistance to colonization of the human intestine; it forms a first line of defence against intruding pathogcns-. If this microflora is wiped out, for example, by antimicrobial therapy, overgrowth by drug-resistant microbes can result. IJsing computer image analysis, we have attempted to study both the microflora itself and its interaction with the immune system of the host”.‘.
Several techniques that USC computer analysis of microscopic images have been developed to study the complicated microbial flora in the human intestine, including measuring the shape and fluorescence intensity of bacteria. These tcchniqucs allow rapid assessment of changes in the intestinal flora and could apply equally to other complex microbial ecosystems.
We have focused on shapex.“‘~” and fluorescence”.“,’ 1nicasurements on microscopic slides of bacteria isolated directly from facces; however, the methods that WC have wised could probably be applied to many other microbial ecosystems. In this article, WC describe these methods and the results that we have obtained using them, and discuss some future prospects for microscopic image analysis in microbiology. The GRID system Our microscopic image analysis system has been named Groningcn reduction of image data (GRII))S~‘~. The system consists of an industrial video camera that has been modificd to provide the long exposure
times that arc needed for fluorcsccnce measurements’2, mounted on top of a microscope and linked to a computer, which is equipped with electronic circuits to allow video images to be captured in computcr memory. Once images have been stored, they can subsequently be processed using standard digital image analysis tcchniqucs’j and personal computers and software that we have developed. Morphometry A human observer can readily distinguish bacteria on an image such as that in Fig. la, but computers find the task more difficult. Bcforc any mcasurcmcnt can bc made, the image must bc divided into objects (the bacteria) and background. We use an automated segmentation technique’” similar to those that have been dcscribcd prcviously’5-‘7. ‘[‘he image is segmented and convcrtcd into a binary image (Fig. I b), revealing objects in the form of white connected regions. The surface area, perimctcr, width, length and other morphological paramctcrs of each connected region can be obtained readily”. Obtaining the true size or volume of each bacterium is slightly more complicated. No imaging system is completely free of distortion, and many staining and slide-l”eparatioli tcchniques distort the original shape of
I 0
N