방어관련 역전사 효소(DRT) --＞ DNA합성 원리를 뒤흔드는 결과 2026 사이언스!! 정리중!!

[문형철]

박테리아의 방어관련 역전사 효소 작용

발효 과정에서 어떻게 

bacterial biotransformation ?

Defense-associated reverse transcriptase(DRTs)

1. 역전사효소란 무엇일까?

보통 세포 안에서 유전정보는 이렇게 흐른다고 배웠어:

DNA → RNA → 단백질

이걸 센트럴 도그마(Central Dogma)라고 해. 하지만 역전사효소는 이 흐름을 거꾸로 만든 특별한 효소야.

→ RNA를 주형(설계도)으로 삼아서 DNA를 합성하는 효소!

그래서 이름도 “역(逆)전사” 효소야. (전사는 DNA → RNA인데, 역전사는 RNA → DNA)

2. 어디에서 발견됐을까?

• 바이러스(특히 레트로바이러스): HIV(에이즈 바이러스)가 대표적! HIV는 자신의 유전자가 RNA로 되어 있어. 사람 세포에 들어가서 역전사효소를 이용해 RNA를 DNA로 바꾼 다음, 그 DNA를 사람 DNA에 끼워 넣고 자기 복제를 해.
• 세균(박테리아): 최근 발견된 DRT(Defense-associated Reverse Transcriptase) 시스템
  • DRT9: ncRNA를 보고 A만 잔뜩 붙인 긴 poly(dA) DNA를 만들어 파지를 막음.
  • DRT3: Drt3a는 보통처럼 RNA를 보고 DNA 만들고, Drt3b는 주형 없이 자신의 단백질 일부가 설계도 역할을 해서 DNA를 만듦. (기존 규칙을 완전히 깨는 방식!)

3. 역전사효소는 어떻게 일할까? (간단 순서)

1. RNA 가닥을 읽음.
2. RNA에 맞춰서 보완 DNA (cDNA, complementary DNA)를 하나씩 붙여가며 합성.
3. RNA 부분을 잘라내고, DNA를 완성시켜 이중 가닥 DNA를 만듦.

HIV 역전사효소는 오류가 많아서 (정확도가 낮아서) 바이러스가 쉽게 변이되고, 약에 잘 안 듣는 이유 중 하나가 되기도 해.

4. 우리 생활에서 왜 중요한가?

• 의학: HIV 치료제 중 많은 약이 역전사효소 억제제야. (AZT 등) 이 효소를 막아서 HIV가 DNA를 못 만들게 함.
• 연구/진단: 실험실에서 RNA를 DNA로 바꿔서 PCR 검사나 유전자 연구를 함 (RT-PCR).
• 생명공학: CRISPR처럼 세균 방어 시스템에서 나온 기술. DRT3처럼 새로운 역전사효소를 이용하면 원하는 DNA를 쉽게 만들 수 있는 도구가 될 수 있음.

DRT 시리즈와 비교해서 한 줄 요약

• 일반 역전사효소 (HIV 등): RNA를 주형으로 DNA를 만듦 (보통 방식).
• DRT9 역전사효소: 파지 감염 신호(dATP 증가)를 받고, A만 잔뜩 붙인 DNA를 만들어 파지 복제를 막음.
• DRT3 역전사효소: 하나는 보통 방식, 다른 하나(Drt3b)는 단백질 자신이 설계도가 되어 DNA를 만듦. (생물학 교과서 개념을 바꾼 획기적 발견!)

한 문장으로 정리하면: 역전사효소는 “RNA → DNA”로 거꾸로 유전정보를 복사하는 특별한 효소로, HIV 같은 바이러스가 번식할 때 쓰이고, 세균은 이를 이용해 자신을 지키는 똑똑한 방어 무기로 사용하며, 우리에게는 치료제와 연구 도구로도 중요한 효소야.

1. 센트럴 도그마란 무엇일까?

분자생물학의 중심원리라는 뜻이야. 우리 몸(또는 모든 생물)의 유전정보가 어떻게 흘러가는지를 설명하는 가장 기본적인 규칙이야.

간단히 말하면:

DNA → RNA → 단백질

이 흐름을 센트럴 도그마라고 불러!

• DNA: 생물의 설계도 (모든 유전정보가 저장되어 있음)
• RNA (특히 mRNA): DNA에서 필요한 부분만 복사한 전령(메신저)
• 단백질: 실제로 몸을 만들고, 일을 하는 일꾼

2. 세 단계로 쉽게 이해하기

1. 복제 (Replication) DNA가 자기 자신을 똑같이 복사함. 세포가 나눌 때 딸세포에게 똑같은 설계도를 물려주기 위해 필요해. (DNA → DNA)
2. 전사 (Transcription) DNA의 한 부분을 주형(설계도)으로 삼아 RNA를 만듦. DNA는 핵 안에 있어서 밖으로 못 나가니까, RNA가 대신 정보를 가지고 세포질로 나감. (DNA → RNA)
3. 번역 (Translation) RNA(특히 mRNA)를 읽어서 아미노산을 연결해 단백질을 만듦. 리보솜(단백질 공장)에서 일어나. (RNA → 단백질)

이렇게 DNA의 정보 → RNA → 단백질로 흘러서 우리 몸이 만들어지고 기능하는 거야!

3. 왜 “도그마(Dogma)”라고 불렀을까?

원래 프랜시스 크릭(DNA 이중나선 발견한 과학자)이 1958년에 제안했어.

그는 “유전정보는 단백질에서 DNA나 RNA로 절대 거꾸로 돌아가지 않는다”고 강하게 주장했기 때문에

“도그마(절대적인 교리)”라는 강한 단어를 썼어.

4. 그런데 예외가 있다! (우리가 전에 배운 부분과 연결)

대부분의 경우는 DNA → RNA → 단백질이 맞지만, 완벽한 규칙은 아니야.

• 역전사효소가 발견되면서 큰 예외가 생김! RNA → DNA로 거꾸로 정보가 흐를 수 있다는 게 밝혀졌어. → HIV(에이즈 바이러스) 같은 레트로바이러스가 이걸 이용함.
• DRT9 (이전 설명): 세균이 파지 방어할 때, RNA를 보고 A만 잔뜩 붙인 DNA(poly(dA))를 만듦. (RNA → DNA)
• DRT3 (최근 발견):
  • Drt3a: 보통처럼 RNA를 주형으로 DNA 만듦.
  • Drt3b: 주형 없이 자신의 단백질 일부가 설계도 역할을 해서 DNA를 만듦. → “단백질이 DNA 합성의 정보를 제공한다”는 기존 생각을 완전히 뒤집는 획기적인 발견!

이 예외들 때문에 지금은 센트럴 도그마를 “대부분의 경우에 맞는 일반적인 원리”로 이해하고 있어.

한 줄 요약

센트럴 도그마는 “DNA 설계도 → RNA 복사본 → 단백질 일꾼”으로 유전정보가 흘러가는 생물학의 기본 규칙이야.

하지만 바이러스나 세균의 DRT 시스템처럼 RNA → DNA로 거꾸로 가는 특별한 경우도 존재해!

이전 대화에서 배운

역전사효소, DRT9, DRT3가 바로

이 센트럴 도그마의 예외를 보여주는 멋진 예시들이야

• 평소 (감염 전): DRT9 단백질 4개가 모여 테트라머(90° 각도)로 조용히 기다림. ncRNA(작은 RNA 조각)와 함께 있음.
• 파지(바이러스)가 침투하면: 파지가 자신의 DNA를 복제하려고 dATP(A 재료)를 많이 만들어냄. 세균 안 dATP 농도가 급격히 올라감.
• 신호 감지 → 형태 변화: DRT9이 4개 → 6개로 모여 헥사머(120° 각도)로 바뀜. (이 변화가 아주 중요!)
• 활성화: 바뀐 DRT9 헥사머가 역전사효소 역할을 해서, A만 잔뜩 붙인 긴 poly(dA) DNA (또는 poly-A-rich cDNA)를 길게 만듦.
• 결과: 이 긴 poly(dA) DNA가 파지가 꼭 필요한 SSB 단백질(파지 DNA 복제에 필수적인 단백질)을 가로채서 묶어버림. → 파지가 더 이상 복제하지 못하고, 감염된 세균은 성장 멈춤(abortive infection) → 주변 세균(형제들)은 살아남음.

DRT3는 어떤 시스템일까?

세균 안에는 DRT3라는 특별한 방어 팀이 있어. 이 팀은 세 가지로 구성돼:

• Drt3a (하나의 역전사효소)
• Drt3b (또 다른 역전사효소, 아주 특별함)
• ncRNA (작은 RNA 조각, 도와주는 스캐폴드 역할)

이 세 가지가 함께 6:6:6 복합체(총 18개가 모인 큰 팀)로 모여서 일함.

두 효소가 각각 다른 방식으로 DNA를 만든다

1. Drt3a (보통 방식)
   • 평범한 역전사효소처럼 ncRNA를 주형(설계도)으로 사용해.
   • RNA에 있는 ACACAC 부분을 보고, GTGTGT가 반복되는 poly(GT) DNA 한 가닥을 만듦.
2. Drt3b (기존 규칙을 완전히 깨는 방식 ← 이번 발견의 핵심!)
   • 아무 DNA나 RNA 주형도 없이 DNA를 만듦.
   • 대신 자신의 단백질 몸통(아미노산) 일부가 RNA처럼 행동해.
   • 단백질 활성 부위의 특정 아미노산들이 “템플릿(설계도)” 역할을 대신해서, AC가 반복되는 poly(AC) DNA 가닥을 정확하게 만듦.

→ 결과적으로 GT/AC가 번갈아 나오는 이중 가닥 DNA가 만들어져!

이걸 “단백질-템플릿 DNA 합성”이라고 해. 지금까지 과학자들은 “DNA는 항상 DNA나 RNA 설계도를 보고 만든다”고 생각했는데, DRT3b는 단백질 자신이 설계도 역할을 한다는 완전히 새로운 규칙을 보여줬어.

이 DNA를 만들어서 파지를 어떻게 막을까?

아직 정확히 밝혀지지는 않았지만, 연구팀은 두 가지 가능성을 생각하고 있어:

• 만든 DNA가 “분자 스펀지”처럼 파지의 중요한 부분을 붙잡아서 활동을 막음.
• 또는 이 DNA가 세균의 다른 면역 반응을 깨워서 파지를 죽임.

(이전 DRT9처럼 세균이 스스로 성장 멈추는 ‘포기 감염’ 방식일 수도 있지만, DRT3는 아직 연구 중이야.)

왜 이 발견이 중요할까? (중학생도 이해하기 쉽게)

• 생물학 기본 개념 변화: “생명정보는 DNA → RNA → 단백질” 방향으로만 흐른다고 배웠지? 그런데 DRT3b는 단백질이 DNA를 만드는 정보(템플릿)를 제공할 수 있다는 걸 보여줘서, 기존 교과서 개념에 큰 충격을 줬어.
• 새로운 생명공학 도구 가능성: CRISPR(유전자 가위)도 원래 세균 방어 시스템에서 나왔어. DRT3는 한 복합체 안에 모든 기능이 들어있어서, 추가 재료 없이 원하는 특정 DNA 서열(예: GT/AC 반복)을 쉽게 만들 수 있음. → 앞으로 원하는 유전자를 빠르고 정확하게 만드는 기술로 발전할 수 있어!

한 줄 요약

DRT3는 세균이 파지로부터 자신을 지키기 위해, Drt3a는 RNA를 보고 DNA를 만들고, Drt3b는 자신의 단백질 몸을 설계도로 써서 완전히 새로운 방식으로 반복 DNA를 만드는 똑똑한 방어 시스템이야. 이 발견은 “DNA는 어떻게 만들어지나?”에 대한 우리 생각을 바꿀 만큼 획기적이라고 과학자들이 평가하고 있어!

이전 설명한 DRT9 (dATP 신호를 받고 poly(dA) 긴 DNA 만드는 시스템)과 비교하면:

• DRT9: 한 종류 효소 + ncRNA → A만 잔뜩 (단일 가닥)
• DRT3: 두 종류 효소 + ncRNA → GT/AC 번갈아 (이중 가닥, 단백질 템플릿 사용)

1. DRT4는 어떤 시스템일까?

DRT4는 세균이 파지(바이러스)를 막기 위해 가진 단순한 방어 시스템이야. 다른 DRT 시스템과 달리 단 하나의 유전자로만 이루어져 있어 (DRT4 단백질 하나만!). 이 단백질이 역전사효소(Reverse Transcriptase) 역할을 해.

2. 그림으로 보는 작동 원리 (왼쪽: 정상 상태 / 오른쪽: 파지 감염 상태)

평소 (Uninfected, 세균이 건강할 때)

• DRT4 단백질이 항상 조금씩 활동하면서 임의의 ssDNA (단일 가닥 DNA, 랜덤한 서열)를 조금씩 만듦.
• 세균 자신의 3'→5' exonuclease (가위 같은 효소)가 이 ssDNA를 빠르게 잘라서 dNMPs (작은 DNA 조각)로 분해함.
• 그래서 세균은 아무 문제 없이 정상적으로 성장함.

파지가 침투하면 (Infected)

• 파지가 세균 안에 들어오면서 자신이 만든 단백질 ORF55 (파지 유전자에서 나온 DNA 결합 단백질)를 생산함.
• 이 ORF55가 DRT4가 만든 ssDNA의 3' 끝을 보호해줌.
• 세균의 exonuclease(가위)가 ssDNA를 자르지 못하게 됨.
• 결과적으로 ssDNA가 세균 안에 계속 쌓임 (Toxic ssDNA accumulation).
• 이 쌓인 ssDNA가 독처럼 작용해서 감염된 세균이 죽음 (Cell death).

→ 이걸 Abortive Infection (포기 감염)이라고 해. 감염된 세균 한 마리가 스스로 죽으면서, 파지가 더 이상 번식하지 못하게 막고 주변 세균 형제들은 살아남음.

3. DRT4의 특별한 점 (다른 DRT와 비교)

• DRT9: 파지가 dATP를 많이 만들면 신호를 받고, A만 잔뜩 붙인 긴 poly(dA) DNA를 만들어 파지의 SSB 단백질을 가로챔. (DNA로 파지 도구 빼앗기)
• DRT3: 두 종류 역전사효소가 협력. 하나는 RNA를 보고, 다른 하나는 단백질 자신이 설계도가 되어 반복 DNA를 만듦.
• DRT4: 가장 단순! 주형 없이 랜덤한 ssDNA를 만듦. 파지가 만든 ORF55 단백질이 트리거가 되어 ssDNA를 보호 → ssDNA 독 쌓임 → 세균 자살.

DRT4는 파지가 직접 자신의 방어 시스템을 활성화시키는 재미있는 전략을 써. 파지가 ORF55를 만들지 않으면 DRT4는 거의 작동하지 않아서, 세균이 불필요하게 죽는 일도 적음.

한 줄 요약

DRT4는 세균이 평소에 조금씩 만드는 랜덤 ssDNA를, 파지가 들어오면 파지 자신이 만든 ORF55 단백질로 보호해서 ssDNA가 쌓이게 만들고, 결국 감염된 세균을 죽여서 파지 번식을 막는 ‘자기희생형’ 방어 시스템이야.

이 시스템은 2025년 Nature Communications에 발표된 최신 연구로, DRT 계열 중에서도 단일 유전자로 작동하는 간단하면서도 영리한 메커니즘으로 주목받고 있어.

이 논문은

미국 스탠퍼드 대학(Alex Gao 연구팀)이

세균의 항파지 방어 시스템(DRT3)에서 발견한 특이한 효소에 대한 연구입니다.

Defense-associated reverse transcriptase(DRTs)

anti-phage system

핵심 내용 요약

전통적으로 DNA 합성은

항상 핵산 주형(nucleic acid template, DNA 또는 RNA)을

필요로 한다고 알려져 있었습니다. (중심교리: template-directed polymerization)

그러나

이 연구팀은

세균의 DRT3 (Defense-Associated Reverse Transcriptase 3) 시스템에서

다음과 같은 혁신적인 메커니즘을 발견했습니다:

• Drt3a: 비코딩 RNA(ncRNA) 안에 있는 ACACAC 서열을 주형으로 사용하여 poly(GT) 가닥을 합성합니다. (일반적인 역전사효소 메커니즘)
• Drt3b: 완전히 핵산 주형 없이 DNA를 합성합니다.
• 대신 효소 자신의 활성부위 아미노산 잔기(amino acids)가 주형 역할을 하여 poly(AC) 가닥을 정밀하게 합성합니다.

결과적으로 DRT3 복합체는

교대하는 dinucleotide repeat DNA (poly(GT/AC))를 만들어냅니다.

Cryo-EM 구조 분석(2.6 Å 해상도)에서 D3 대칭의 6:6:6 복합체(Drt3a : Drt3b : ncRNA)가 확인되었으며,

Drt3b의 활성부위 특정 잔기가 정확한 염기 교대를 강제하는 것으로 밝혀졌습니다.

이 발견은

“단백질이 DNA 서열의 청사진(blueprint)으로 작용할 수 있다”는

완전히 새로운 생물학적 원리를 제시하며,

수십 년간의 DNA 합성 원리를 뒤흔드는 결과로 평가받고 있습니다.

Drt3b는

핵산 주형(nucleic acid template) 없이

단백질 자체의 아미노산 잔기가 주형 역할을 하는 완전히 새로운 DNA 합성 메커니즘을 보여주는

핵심 효소입니다.

아래에서 구조·촉매 기전·복합체 상호작용까지 단계적으로 설명하겠습니다.

1. 전체 DRT3 복합체 구조 (Cryo-EM 기반)

• 복합체 조성: Drt3a : Drt3b : ncRNA = 6:6:6 (D3 대칭, D3-symmetric hexamer-of-hexamers)
• 해상도: 2.6 Å (Cryo-EM)
• Drt3a와 Drt3b는 각각 별개의 역전사효소(reverse transcriptase, RT) 도메인을 가지며, ncRNA(비코딩 RNA)가 Drt3a의 주형(template) 역할을 합니다.
• 이 6:6:6 복합체는 세균의 항파지 방어 시스템(antiphage defense)으로 작동하며, 파지 감염 시 교대 dinucleotide repeat DNA [poly(GT/AC)]를 합성합니다.

Drt3a vs Drt3b의 역할 구분

• Drt3a: 전통적 RT 방식 ncRNA 내 보존된 ACACAC 서열을 주형으로 사용하여 poly(GT) 가닥을 합성 (template-directed polymerization).
• Drt3b: 혁신적 protein-templated 방식 (논문의 핵심 발견) 핵산 주형 완전 없이 poly(AC) 가닥을 합성. 대신 Drt3b 자신의 활성부위(active site) 아미노산 잔기들이 “단백질 주형(protein template)”으로 작용.

2. Drt3b의 protein-templated DNA 합성 메커니즘 (핵심)

Drt3b는 protein-primed + protein-templated라는 이중 혁신을 보여줍니다.

(1) Protein-priming (합성 개시)

• 일반 RT나 DNA polymerase는 3'-OH (프라이머)나 RNA/DNA 주형을 필요로 하지만, Drt3b는 단백질 잔기의 하이드록실기(-OH)를 이용해 첫 번째 뉴클레오티드를 공유결합(covalent attachment)으로 붙입니다.
• 이는 일부 바이러스 프라이밍(예: protein-primed DNA polymerase)과 유사하지만, Drt3b에서는 template 자체도 단백질이라는 점이 다릅니다.

(2) Active site 아미노산이 “주형(template)” 역할

• Drt3b 활성부위에는 Drt3b 특이적 보존된 잔기들이 배치되어 있으며, 이들이 정확한 A와 C의 교대(alternation)를 강제합니다.
• 논문에서 명시된 핵심 잔기: Glu26 (글루탐산 26번)이 주요 역할. “In contrast to Drt3a, the Drt3b active site directly encodes the alternating poly(AC) product. Specifically, Glu26 serves as the primary ...” (논문 본문)
• Glu26 외에도 Drt3b 특이적 conserved residues들이 dNTP (deoxynucleotide triphosphate)의 염기 선택을 “코딩(encode)”합니다. → 아미노산 측쇄(side chain)의 화학적 성질(전하, 수소결합, 공간적 배치)이 A와 C를 번갈아 선택하도록 유도하여 poly(AC) 반복 서열을 정밀하게 합성합니다.
• 이는 생물학 중심교리(central dogma)의 확장으로, 단백질 → DNA 정보 전달 가능성을 최초로 입증합니다.

(3) 촉매 사이클 (Proposed mechanism, 단계별)

1. 복합체 조립: Drt3a–Drt3b–ncRNA 6:6:6 hexamer 형성 (D3 대칭).
2. Drt3a 작동: ncRNA의 ACACAC motif를 주형으로 poly(GT) 가닥 합성 (전통 RT).
3. Drt3b 작동 (동시/연속):
   • Drt3b 활성부위로 dATP 또는 dCTP가 들어옴.
   • Glu26 등 잔기가 염기 선택을 “템플릿화” → A 다음에는 반드시 C, C 다음에는 반드시 A가 선택됨.
   • Protein-priming site에서 첫 번째 뉴클레오티드가 단백질 -OH에 공유결합.
   • 연속적인 phosphodiester bond 형성으로 poly(AC) 가닥 연장.
4. 최종 산물: poly(GT) (Drt3a) + poly(AC) (Drt3b) = alternating poly(GT/AC) double-stranded DNA.

이 메커니즘은 정확한 dinucleotide repeat를 보장하며, 무작위 homopolymer나 다른 서열이 아닌 GT/AC 교대만 생산하도록 진화된 것으로 보입니다.

3. 구조적 근거 (Cryo-EM Figure 설명)

• 주요 Figure (본문 Fig. 1~3, Supplementary Fig. S1~S14):
  • 전체 복합체 구조: D3 대칭 6:6:6 hexamer. Drt3a와 Drt3b가 번갈아 배치된 ring-like 구조. ncRNA가 Drt3a subunit에 결합.
  • Drt3b active site close-up: Glu26을 포함한 conserved residues가 활성부위 pocket에 배치된 모습. dNTP가 들어가는 pocket에서 아미노산 측쇄가 염기와 상호작용하는 상세 구조 (2.6 Å 해상도로 원자 수준 확인).
  • Poly(AC) 제품 모델: Drt3b가 합성한 DNA 가닥이 protein template에 의해 어떻게 “코딩”되는지 schematic diagram으로 제시.
  • Mutational analysis: Drt3b 특이적 잔기(Glu26 등) 돌연변이 시 poly(AC) 합성 완전 소실 → protein template 역할 직접 증명.

4. 생물학적·응용적 의미

• 항파지 방어: 파지 감염 시 특이적 dinucleotide repeat DNA를 생산해 파지 DNA를 방해하거나 신호 전달.
• 중심교리 확장: “DNA/RNA가 항상 template”이라는 70년 규칙을 깨뜨림. 단백질이 sequence-specific DNA blueprint 역할을 할 수 있음.
• 합성생물학 응용: sequence-specific, template-independent DNA 합성 도구 개발 가능성 (예: 반복 서열 합성, 새로운 polymerase 엔지니어링).

https://www.nature.com/articles/s41467-025-66997-x

DRT4는

단일 유전자로 구성된 항-파지 방어 시스템으로,

템플릿 독립적(template-independent)으로 무작위 서열의 단일가닥 DNA(ssDNA)를 합성합니다.

파지 감염 시, 파지가 인코딩한 DNA 결합 단백질 ORF55가 DRT4가 만든 ssDNA의 3' 말단을 숙주 exonuclease(외절효소)로부터 보호합니다. 이로 인해 독성 ssDNA가 축적되어 세포 사멸(abortive infection)을 유발하고, 파지 증식을 막습니다.

또한 ORF55는 DRT4의 구조 동족체인 DRT6도 활성화하여, DRT 계열 시스템 간 보존된 활성화 메커니즘을 시사합니다.

이 연구는 DRT4가 protein-primed, template-independent 역전사효소로서 작동하며, 파지-유래 단백질이 방어 시스템을 “활성화 + 보호”하는 새로운 패러다임을 제시합니다.

DRT4의 구조 및 효소 메커니즘

• 올리고머 구조: DRT4는 hexamer (6량체)로 조립되며, 두 개의 trimer가 back-to-back으로 스택된 형태 (약 10° tilt).
• 도메인 구성: 전형적인 역전사효소(RT) 도메인(손가락·손바닥 영역) + αRep 도메인 (엄지손가락 역할을 대신함). 촉매 부위는 보존된 YIDD motif.
• 촉매 활성: Mg²⁺ (또는 Mn²⁺) 이온 의존적. 템플릿 없이 dNTP를 중합하여 평균 60~80 nt 길이의 무작위 ssDNA를 생성.
• 프라이밍 메커니즘: 보존된 티로신 잔기 (Tyr125)가 priming site로 작용. Tyr125의 -OH기가 첫 번째 dNTP와 공유결합(covalent attachment)을 형성하여 DNA 합성을 개시합니다.

Fig. 1 설명 (DRT4의 생화학적 특성) (a) 플라크 어세이: DRT4 야생형(WT)은 파지 T5에 강한 보호 효과를 보이지만, 촉매 불활성화 mutant(D240A/D241A)는 보호 효과 상실. (b) SEC(크기 배제 크로마토그래피)와 SDS-PAGE: 정제된 DRT4의 A260/A280 비율이 0.92로 핵산 결합을 시사. (c) 분석적 원심분리(AUC): WT는 418 kDa hexamer 확인, mutant는 핵산이 적어 390 kDa. (d) 시간 경과 중합 어세이: MgCl₂ 존재 하 WT는 ssDNA 생성, mutant는 없음. (e) Fluorescein-dATP 결합: WT에서 SDS-PAGE 이동 시프트 관찰 (공유결합 증거). (f) 생성된 DNA 길이 분포: Mg²⁺ 조건에서 평균 62 nt, Mn²⁺에서 80 nt 정도의 무작위 서열 ssDNA.

Fig. 2 설명 (DRT4-DNA 복합체 cryo-EM 구조) (a) 전체 cryo-EM 밀도 지도: 서로 다른 색으로 표시된 6개의 서브유닛과 녹색으로 표시된 DNA. (b) 전체 hexameric 구조 모델. (c) 단일 서브유닛 구조: 주황(손가락), 청색(손바닥), 자주(αRep 도메인), 적색(YIDD 촉매 motif). 오른손 모양의 RT 도메인에서 αRep가 엄지 역할을 대신함. hexamer는 두 trimer가 back-to-back으로 약간 기울어진 형태.

Fig. 3 설명 (프라이밍 잔기 Tyr125) (a) cryo-EM 밀도 지도: DNA 5' 끝과 Tyr125 사이의 연결 밀도 명확히 관찰. (b) 플라크 어세이: Y125F mutant는 항-파지 방어 완전 상실. (c) SEC/SDS-PAGE: Y125F mutant에서 A260/A280 비율 급감 (핵산 결합 상실). (d) 중합 활성: Y125F mutant에서 ssDNA 합성 완전 소실 → Tyr125이 priming site임을 직접 증명.

항-파지 방어 메커니즘 (Abortive Infection, Abi)

파지 감염 → DRT4 활성화 → 무작위 ssDNA 대량 합성 → 파지 ORF55가 ssDNA 3' 끝 보호 → 숙주 exonuclease에 의한 분해 방지 → 독성 ssDNA 축적 → 세포 사멸 → 파지 증식 차단.

Fig. 4 설명 (Abi 메커니즘) (a) 성장 곡선: 낮은 MOI(multiplicity of infection)에서는 보호, 높은 MOI에서는 세포 성장 정지. (b) CFU(colony forming unit) 어세이: 고 MOI 시 살아있는 세포 수 급감 (세포 사멸). (c-d) Propidium Iodide(PI) 염색 형광 현미경: MOI=2 조건에서 WT DRT4 발현 세포에서 세포막 파괴(사멸) 관찰.

추가 발견

• ORF55는 DRT4 합성 ssDNA의 3' 끝을 보호할 뿐 아니라, 구조적으로 유사한 DRT6도 활성화.
• 이는 DRT 계열 시스템에서 파지-유래 DNA 결합 단백질이 공통적인 “활성화 + 보호” 역할을 할 가능성을 시사.

연구 의의

• DRT4는 템플릿 독립적 + protein-primed 역전사효소로서, 기존 알려진 DRT 시스템(예: DRT3의 protein-templated)과 다른 새로운 메커니즘을 제시.
• 파지와 숙주 간 “arms race”에서 파지가 자신의 단백질(ORF55)로 숙주의 방어 시스템을 역이용하다가 오히려 세포 사멸을 유발하는 아이러니한 전략.
• 박테리아 면역 연구의 새로운 패러다임 제공. 향후 합성생물학, 항바이러스 도구 개발에 응용 가능성.

이 논문은

이전에 설명드린 DRT3 (Science 2026) 논문과 함께

DRT 계열 항-파지 시스템의 다양성을 잘 보여줍니다.

DRT3가 “protein-templated dinucleotide repeat”라면,

DRT4는 “template-independent random ssDNA + phage protein protection”입니다.

DRT3 vs DRT4 비교 (방어 관련 역전사효소의 두 가지 전략)

세균이 파지(바이러스)를 막기 위해 사용하는 방어 관련 역전사효소(DRT)는 DNA를 만드는 특별한 효소예요. 그런데 DRT3와 DRT4는 DNA를 만드는 방식이 완전히 다릅니다.

항목DRT3 (Science 2026)DRT4 (Nature Communications 2025)

DNA 만드는 방식	단백질 주형 방식 (protein-templated)	주형 없이 무작위 방식 (template-independent)
만들어지는 DNA	정확한 GT/AC 교대 반복 DNA (예: GTGTGTACACA...처럼 규칙적)	완전 무작위 ssDNA (단일가닥 DNA, 길이 60~80개 정도)
주형(template)	Drt3b 효소 자신의 아미노산 잔기가 주형 역할 (단백질이 DNA 서열을 직접 “지시”)	주형 전혀 없음 그냥 dNTP(DNA 재료)를 마구 붙임
특징	매우 정밀하고 규칙적 효소가 “A 다음은 반드시 C, C 다음은 반드시 A”처럼 엄격히 통제	아주 무질서하고 랜덤 효소가 “아무거나 붙여!” 하는 느낌
파지 방어 방식	특이적인 반복 DNA를 만들어 파지를 방해	무작위 ssDNA를 대량 생산 → 독이 쌓여 세포가 스스로 죽음 (Abortive Infection)
추가 도움	ncRNA(비코딩 RNA)와 Drt3a가 함께 협력	파지가 만든 ORF55 단백질이 ssDNA 끝을 보호해 더 독하게 만듦

비유로 더 쉽게 이해하기

• DRT3는 정밀 공장 같아요. 공장에서 “항상 빨강-파랑-빨강-파랑” 색깔로만 제품을 찍어내듯, Drt3b 효소가 자신의 몸 일부(아미노산)를 주형(틀)처럼 사용해서 정확한 반복 패턴(GT/AC)의 DNA만 만듭니다. → “내가 정한 규칙대로만 만들어!” 하는 고급 전략입니다. (과학적으로 아주 놀라운 발견: 단백질이 DNA 서열 정보를 직접 알려줄 수 있다!)
• DRT4는 난장판 공장 같아요. “주형(틀) 없이 그냥 재료(dNTP)를 마구잡이로 붙여!” 합니다. 그래서 만들어지는 DNA는 완전 랜덤이고 규칙이 없어요. 그런데 파지가 들어오면, 파지가 만든 ORF55라는 단백질이 이 무작위 DNA의 한쪽 끝을 보호해줍니다. → DNA가 분해되지 않고 세포 안에 계속 쌓여 독이 됩니다. 결국 세포가 스스로 죽으면서 파지도 같이 죽는 전략(자폭 방어)입니다.

한 줄 요약

• DRT3 = “내 몸(단백질)으로 정확한 반복 DNA를 정성껏 만든다” → 정밀 타격
• DRT4 = “주형 없이 마구잡이 DNA를 뽑아내고, 파지가 도와서 그 DNA를 독으로 만든다” → 대량 자폭

두 시스템 모두 세균이 파지를 막기 위해 진화시킨 똑똑한 방법이지만, DNA를 어떻게 만들고, 어떻게 독으로 쓰느냐에서 완전히 다른 아이디어예요.

1. nature  
2. nature communications  
3. articles  
4. article

Anti-phage defense mechanism involving phage-encoded DNA binding protein and bacterial reverse transcriptase DRT4

Download PDF

• Article
• Open access
• Published: 02 December 2025

Anti-phage defense mechanism involving phage-encoded DNA binding protein and bacterial reverse transcriptase DRT4

• Yong Wang, 
• Hao Wu, 
• Jinyue Li, 
• Qiyin An, 
• Zhenhua Tian & 
• Zengqin Deng 

Nature Communications volume 17, Article number: 289 (2026) Cite this article

•  
•  
•  
•  

Abstract

Prokaryotic defense-associated reverse transcriptase (DRT) systems confer host resistance to viral infection through DNA synthesis; however, the molecular mechanisms underlying their function remain poorly understood. Here, we demonstrate that DRT4, a single-gene anti-phage defense system, synthesizes single-stranded DNA (ssDNA) products of random sequences in a template-independent manner. High-resolution cryo-EM structures of DRT4 in multiple functional states elucidate its oligomeric architecture, catalytic metal ion coordination, and substrate/DNA product binding, offering mechanistic insights into its promiscuous polymerization activity. Structural and biochemical analyses further identify a conserved tyrosine residue that acts as the priming site for the initiation of DNA synthesis. Upon phage infection, a phage-encoded DNA-binding protein, ORF55, protects the 3’ end of the DRT4-synthesized ssDNA from host exonuclease degradation, likely resulting in toxic ssDNA accumulation that leads to cell death. Remarkably, ORF55 also activates DRT6, a structural homolog of DRT4, suggesting a conserved activation mechanism among related DRT systems. These findings provide structural and mechanistic insights into DRT4-mediated antiviral defense, establishing a distinct paradigm for antiviral reverse transcriptase in bacterial immunity.

Similar content being viewed by others

Protein-primed homopolymer synthesis by an antiviral reverse transcriptase

Article 28 May 2025

Structural basis for broad anti-phage immunity by DISARM

Article Open access27 May 2022

A DNA-gated molecular guard controls bacterial Hailong anti-phage defence

Article 30 April 2025

Introduction

Bacteriophages (phages), the most abundant biological entities on Earth, are ubiquitously distributed across virtually all ecological niches1. As the primary cause of bacterial mortality in natural environments, phages have imposed strong selective pressure, driving the evolution of diverse bacterial anti-phage defense systems and mechanisms through long-term co-evolution2,3,4,5. Reverse transcriptases (RTs), enzymes capable of synthesizing DNA from an RNA template, are widely distributed across the tree of life. Prokaryotic RTs, first identified in 1989 as the core component of retrons6,7,8,9,10, are now classified into several groups, including group II intron, diversity-generating retroelements (DGRs), retrons, abortive infection (Abi) RTs, CRISPR-Cas-associated RTs, and the unknown group (UG)11,12,13,14. Further phylogenetic analyses indicated that Abi-RTs cluster with the UG-RTs clade, named UG/Abi RTs15. UG/Abi RTs encompass at least 42 members with considerable sequence diversity and different domain organizations, which can be further classified into three major classes. Class 1 RTs are fused to an alpha-helix HEAT-like repeat-containing domain (also named αRep domain), class 2 RTs are highly diverse and have not fused with any known domains, and class 3 is commonly associated with nitrilase or phosphohydrolase16. Notably, some Abi-RTs (e.g., AbiK, Abi-P2, and AbiA) synthesize long single-stranded DNA (ssDNA) products of random sequences in a template- and nucleic acid primer-independent manner17,18, yet the mechanistic link between this activity and phage defense remains unresolved. The phylogenetic proximity of UG-RTs to Abi-RTs implies that UG-RTs may function analogously to Abi-RTs, conferring host anti-phage ability. Indeed, bioinformatic analyses identified a specialized subset of UG-RTs known as defense-associated RTs (DRTs), comprising nine types (DRT1-9) with experimentally confirmed anti-phage activity16,19.

Although the DRTs play a well-established role in defending against DNA phages, the lack of detailed biochemical characterization of their enzymatic activities and structural information has left the molecular mechanisms underlying their functionality poorly understood. To date, the DRT2 system, composed of an RT and a non-coding RNA (ncRNA), is the best-characterized system, in which phage infection triggers the synthesis of concatemeric cDNA from the ncRNA template, ultimately producing a toxic protein that inhibits phage replication20,21. These advances have enhanced our understanding of DRT systems, yet a comprehensive understanding of how DRT systems recognize invading phages and exert anti-phage function is still lacking.

In this study, we characterize DRT4, a single-gene anti-phage defense system that provides robust protection against phage T5 infection. Our biochemical analyses demonstrate that DRT4 functions as a template-independent polymerase capable of synthesizing ssDNA products of random sequences. High-resolution cryo-EM structures (2.27 to 2.93 Å) of DRT4 in multiple functional states elucidate its oligomeric architecture, catalytic metal ion coordination, and substrate/DNA product binding, providing mechanistic insights into its promiscuous polymerization activity. Structural and mutational analyses identify a conserved tyrosine residue that serves as the priming site for the initiation of DNA synthesis. More importantly, we identify ORF55, a phage-encoded DNA-binding protein, as the trigger for the DRT4 antiviral immunity. Mechanistic studies reveal that ORF55 protects the 3’ end of the DRT4-synthesized ssDNA from host exonuclease degradation, likely leading to toxic ssDNA accumulation, which induces cell death upon phage infection. We further demonstrate that this activation mechanism is conserved in DRT6, a structural homolog of DRT4, indicating a broad paradigm for immune activation among related DRT systems.

Results

Functional characterization of DRT4

The DRT4 system is widely distributed across diverse bacterial taxa and features a minimal genetic architecture comprising only a reverse transcriptase (RT) gene, with its putative catalytic activity intrinsically coupled with immune function16,19. Unlike DRT2, DRT4 does not require ncRNA or additional protein partners to execute its anti-phage function. Expression‎ of the wild-type DRT4 from E. coli 12-c8-a in E. coli BL21(DE3) provided robust protection against phage T5 and a relatively weak defense against phages T3, T6, and T7 (Fig. 1a and Supplementary Fig. 1a). By contrast, mutating the conserved catalytic motif YIDD to YIAA (D240A/D241A) completely lost the protective activity against phage T5 (Fig. 1a). Currently, the enzymatic activity of DRT4 has not been characterized. We expressed and purified DRT4 in E. coli BL21 (DE3) using affinity chromatography followed by size-exclusion chromatography (SEC). Notably, the wild-type DRT4 exhibited an A260/A280 ratio of 0.92 (Fig. 1b), suggesting co-purification with nucleic acids. By contrast, the catalytically inactive mutant reduced the A260/A280 ratio to 0.54 (Supplementary Fig. 1b), implying diminished nucleic acid association. Analytical ultracentrifugation (AUC) revealed that the wild-type DRT4 forms a 418 kDa oligomer, significantly larger than its theoretical monomeric mass (63.8 kDa). The D240A/D241A mutant exhibited a slightly smaller oligomeric mass (~390 kDa), suggesting a loss of ~28 kDa per oligomer, consistent with the absence of bound nucleic acids in the mutant protein (Fig. 1c).

Fig. 1: Purification and characterization of DRT4.

The alternative text for this image may have been generated using AI.

Full size image

a Plaque formation of phage T5 on cells expressing empty vector, DRT4 WT, or D240A/D241A mutant. 10-fold serial dilutions of phage lysate were spotted on plates. b Size-exclusion (Superose 6 Increase 10/300 GL Cytiva) and SDS-PAGE profiles of the purified recombinant DRT4. The results are representative data from at least three independent replicates. c Sedimentation coefficient distribution of the DRT4 and the D240A/D241A mutant. d Time course of template-independent polymerization activity of DRT4 and mutant analyzed with MgCl2. The results are representative data from at least three independent replicates. M: Marker comprises a heterogeneous mixture of ssDNA fragments, sequence information of Marker is provided in Supplementary Data 1. e Analysis of labeled nucleotide (Fluorescein-12-dATP) covalently attached to DRT4 and mutant in the presence of MgCl2. Protein samples were analyzed on an SDS-PAGE gel and visualized using fluorescent readout (right panel), then stained with Coomassie Blue (left panel, CB staining). The results are representative data from at least three independent replicates. f Distribution of read lengths of DRT4 polymerization products in the presence of MgCl2 or MnCl2.

As a member of the UG/Abi RTs family, DRT4 belongs to class 1, which also includes three Abi-RTs (AbiA, AbiK, and Abi-P2) known for their template-independent DNA polymerization activity. We therefore assessed whether DRT4 exhibits similar enzymatic activity. The results showed that DRT4 generates DNA products in a template-independent manner, which is strongly activated by divalent ions Mg2+, Mn2+, and Co2+, whereas the D240A/D241A mutant lost the DNA polymerization activity (Fig. 1d and Supplementary Fig. 1c). Intriguingly, when DRT4 was incubated with dNTP substrates and analyzed by SDS-PAGE, we observed a pronounced mobility shift in the DRT4 protein following the reaction. By contrast, the D240A/D241A mutant displayed no such shift, strongly suggesting that the DNA products are covalently linked to the protein (Fig. 1e). High-throughput sequencing of the DNA products in the presence of Mg2+ or Mn2+revealed an average length of 62 nt and 80 nt with random sequences, respectively, under the in vitro conditions tested (Fig. 1f). Nucleotide composition analysis of the DNA products indicated that all four dNTPs were incorporated into the ssDNA products, with preference for dATP and dGTP in the presence of Mg2+ (Supplementary Fig. 1d and e). Strikingly, DRT4 exhibits constitutive polymerase activity in vitro, generating DNA products with an average length exceeding 60 nt; however, purified DRT4 is only associated with short oligonucleotides. This discrepancy suggests that either host factors suppress DRT4’s processivity in vivo or DRT4 generates long DNA products that are subsequently degraded by host nucleases.

DRT4 assembles as a dimer of trimers

To elucidate the structural basis for the activity of DRT4, we determined the cryo-EM structure of DRT4 at 2.27 Å resolution. The high-quality electron density map enabled the modeling of nearly the entire protein, with the exception of the highly flexible C-terminal region (residues 507-540) (Supplementary Figs. 2 and 3). In each DRT4 monomer, we were able to model a 5-nt oligomer. Additionally, we determined the cryo-EM structure of the D240A/D241A mutant at 2.7 Å resolution, which is essentially identical to the wild-type DRT4 except for the absence of observable electron density corresponding to bound nucleotide oligomer (Supplementary Fig. 2). DRT4 assembles into a hexameric structure comprising two back-to-back stacked trimers, adopting a three-fold symmetric architecture with a ~ 10° interlayer tilt around the central symmetry axis. The N-terminal reverse transcriptase (RT) domain adopts a canonical “right-hand” fold, characteristic of group II intron reverse transcriptases, but lacks the thumb subdomain, which is replaced by an αRep domain (Fig. 2a–c). The RT domain consists of two subdomains. The fingers subdomain (residues 1–108 and 157–207) is composed of seven α-helices (α1–α4 and α6-α8) and a β-hairpin (β1–β2). The palm subdomain (residues 109–156 and 208–280) contains three α-helices (α5, α9, and α10) and an antiparallel β-sheet (β3–β6). The catalytic YIDD motif resides within a hairpin structure formed by the antiparallel β4 and β5 in the palm subdomain. The αRep domain, comprising 12 α-helices (α11–α22), occupies the spatial position typically filled by the thumb subdomain in the conventional RTs, with α11-α13 mimicking the structural role of the thumb subdomain (Fig. 2c).

Fig. 2: Cryo-EM structure of the DRT4-DNA complex.

The alternative text for this image may have been generated using AI.

Full size image

a Cryo-EM density map of the DRT4-DNA complex. Each subunit is in different color, and the bound DNA is color forest green. b Overall structure of the DRT4-DNA complex. Subunits and DNA follow same color scheme as in (a). c Structure of a single subunit. Linear domain organization of DRT4 is shown. Each domain is labeled and in a different color. The finger subdomain, palm subdomain, and αRep domain are colored orange, cyan, and purple, respectively, while the DNA is shown in green. The catalytic motif YIDD is shown as sticks and colored red.

Two distinct interfaces stabilize the oligomeric architecture. Interface 1 mediates the intra-trimer stabilization through extensive interactions between the α5 helix (palm subdomain) and α11 helix (αRep domain) of one subunit and α12, α14, and α16 helices (αRep domain) of an adjacent subunit, burying a total interface area of 1292 Ų. Interfaces 2 governs the inter-trimer interactions, forming between corresponding subunits of the upper and lower trimers, which is characterized by an interface area of 1067 Ų (Supplementary Fig. 4a). Detailed structural analysis reveals that hydrophilic interactions play a crucial role in stabilizing both interfaces 1 and 2. Notably, in interface 1, residue Phe140 inserts into a pocket formed by residues from α14 and 16 helices of the adjacent subunit (Supplementary Fig. 4b). To validate the functional importance of these two interfaces, we generated targeted mutants disrupting each interface. For interface 1, the F140A mutation altered the oligomeric organization of DRT4, as evidenced by comparing the retention volume of the WT and F140A mutant proteins (Supplementary Fig. 4c). For interface 2, the E202A/H204A mutant could not be obtained as a soluble protein, suggesting that this mutation compromises structural integrity. Consistently, both mutants lost anti-phage activity. Furthermore, the F140A mutant lost polymerization activity, supporting the notion that oligomer formation is essential for DRT4’s enzymatic activity and anti-phage function (Supplementary Fig. 4c–e).

Interactions between DRT4 and DNA products

Our biochemical analysis demonstrates that DRT4 exhibits dependence on Mg²⁺, Mn²⁺, Ni²⁺, or Co²⁺ for its polymerase activity, with no detectable catalytic activity in the presence of other tested divalent metal ions, including Ca²⁺, Zn²⁺, and Cu²⁺ (Supplementary Fig. 1c). To elucidate the structural basis of the metal ion and substrate recognition, we successfully determined a 2.93 Å cryo-EM structure of DRT4 in the presence of Mn²⁺ and dNTP substrates. The high-resolution reconstruction reveals two well-ordered Mn²⁺ ions coordinated by the catalytic residues and a bound dNTP in the active site, capturing the enzyme in a pre-catalytic state primed for nucleotide incorporation. The substrate-binding pocket can accommodate all four dNTPs without steric hindrance from surrounding residues (Supplementary Fig. 3b). Conserved residues stabilize the incoming dNTP near the 3’-hydroxyl group of the DNA product. Arg75 collaborates with Asn271 and Lys274 in the palm domain to position the dNTP for nucleophilic attack on its α-phosphate, facilitating 3’-5’ phosphodiester bond formation. Metal ion coordination is mediated primarily by conserved Asp149, Asp240, and Asp241 in the palm domain, with additional stabilization provided by the carboxyl oxygen of Ile150 (Supplementary Fig. 5a). For DNA product interaction, the three most 3’-terminal nucleotides are stacked against Tyr238 and Tyr337, while Asn124 provides additional contacts with bases at positions -1 and -2. The phosphate backbone of nucleotides -2 and -3 is engaged by Lys24 and Ser26, respectively. The 5’-terminal bases are solvent-exposed, with only Arg301 contacting base -5 (Supplementary Fig. 5a and b). The promiscuous binding of dNTP substrates and nonspecific interaction with DNA products provide a structural explanation for DRT4’s ability to synthesize ssDNA of random sequences.

To assess the functional relevance of these interactions, we mutated residues involved in metal ion coordination, dNTP binding, or DNA product binding. Disruption of the metal-coordinating residue (D149A) completely abolished DRT4’s anti-phage activity. Similarly, a single mutation of dNTP or DNA-interacting residues (K24A, R75A, N124A) significantly impaired anti-phage defense, while the double mutant K24A/R75A completely lost the protection ability (Supplementary Fig. 5c). Importantly, the DNA polymerization activity of these mutants correlated directly with their anti-phage efficacy (Supplementary Fig. 5d), reinforcing the notion that DRT4’s antiviral function is intrinsically linked to its enzymatic activity.

DNA synthesis of DRT4 is dependent on protein priming

Our biochemical data indicate that DRT4 synthesizes DNA in a manner independent of a template and nucleic acid primer, with the product covalently linked to the protein (Fig. 1e). We proposed that the DNA polymerization activity of DRT4 is initiated by the protein itself acting as the primer. Previous studies have shown that residues such as serine, threonine, and tyrosine often play key roles in protein-primed DNA synthesis22. Consistent with this, close inspection of the cryo-EM map of DRT4 revealed an unmodeled density connecting the 5′ end of the DNA product to a tyrosine (Tyr125) (Fig. 3a), suggesting that this residue may serve as the priming site. To test this hypothesis, we generated a Y125F mutant, substituting this tyrosine with phenylalanine. To improve the protein yield, an MBP tag was fused to the N-terminus of the mutant protein. Functional assays demonstrated that the Y125F mutant completely lost its ability to defend against phage T5 (Fig. 3b). Furthermore, the SEC profile of the mutant exhibited an A260/A280 ratio of 0.51 (Fig. 3c), indicating the absence of bound nucleic acids. Enzymatic activity assays confirmed that, like the catalytically dead D240A/D241A mutant, Y125F was defective in processive DNA synthesis (Fig. 3d). Evolutionary conservation analysis of 174 DRT4 systems across diverse bacterial species showed absolute conservation of Tyr125 (Supplementary Fig. 6). These complementary biochemical, functional, and evolutionary conservation analyses establish Tyr125 as the essential priming residue for DRT4-mediated DNA polymerization.

Fig. 3: Priming residue of DRT4 polymerization activity.

The alternative text for this image may have been generated using AI.

Full size image

a The unmodeled density connects the 5’ end of the DNA product to residue Y125. The cryo-EM density is shown as a grey surface. DRT4 and DNA are colored as in Fig. 2c. b Plaque formation of phage T5 on cells expressing empty vector, wild-type (WT) DRT4, or DRT4 variants. 10-fold serial dilutions of phage lysate were spotted on plates. c Size-exclusion (Superose 6 Increase 10/300 GL Cytiva) and SDS-PAGE profiles of the purified MBP-DRT4Y125F mutant. The results are representative data from at least three independent replicates. d Template-independent polymerization activity of DRT4 and variants in the presence of 5 mM MgCl2. The results are representative data from at least three independent replicates.

DRT4 exerts anti-phage function via abortive infection by inducing cell death

The bacterial immune system exerts anti-phage functions through various immune strategies23,24. Abortive infection (Abi) refers to a strategy that prevents the propagation of phages within a population by triggering self-destruction or dormancy after viral infection25,26. Abi-RTs were initially thought to function through the Abi strategy by inducing cell death27. However, recent studies have shown that AbiA may not necessarily induce cell death following phage infection17. For the DRT systems, DRT2 has been proven to provide anti-phage defense through the Abi strategy by inducing cell dormancy20,21, while the anti-phage characteristics of DRT4 remain unclear. We performed liquid culture growth assays to test whether the DRT4 system defends against phage infection via abortive infection. The DRT4 system provided robust phage resistance at a low multiplicity of infection (MOI = 0.5), but this defense was compromised at higher viral loads (MOI = 5), consistent with an abortive infection mechanism (Fig. 4a). To determine whether DRT4-expressing cells undergo growth arrest or cell death upon phage infection, we compared the viability of cells with and without DRT4 system following phage T5 challenge at different MOIs (0, 0.5, and 5). Colony-forming unit (CFU) assays conducted at multiple time points revealed that DRT4-expressing cells exhibited significantly higher survival rates than DRT4-lacking cells at an MOI of 0.5. However, the vast majority of cells with the DRT4 system succumbed to phage infection by 120 minutes post-infection at an MOI of 5 (Fig. 4b). This observation supports a model in which the DRT4 system defends against phage infection by triggering host cell suicide upon viral infection. The results of the fluorescence microscopy experiment further corroborated these findings. Using propidium iodide (PI), a membrane-impermeant dye that selectively labels dead cells, we observed that more than 70% DRT4 system-containing cells were stained by PI after phage T5 infection at an MOI of 2 (Fig. 4c and d), confirm‎ing that activation of the DRT4 system induces cell death as a protective strategy.

Fig. 4: DRT4 defend phage via Abi mechanism by inducing cell death.

The alternative text for this image may have been generated using AI.

Full size image

a Bacterial growth curves in the presence of either DRT4 or empty vector upon infection by phage T5 at indicated MOI. The data represent the mean ± SD (shaded areas) of three independent experiments. b The survival status of E. coli BL21 (DE3) transformed with either DRT4 or empty vector was assessed by plating and counting colony-forming units (CFUs) following 120 min of infection with phage T5 at indicated MOI. All values are the mean ± SD of three independent experiments. c Representative images of cells expressing DRT4 either uninfected or infected by phage T5 at an MOI of 2. The bacterial samples were stained with propidium iodide (PI). Scale bars = 5 µm. d Quantification of the percentage of stained cells in c with two replicates.

Phage-encoded DNA-binding protein ORF55 triggers DRT4 immunity

Since DRT4 expression‎ alone does not induce host toxicity, we hypothesized that its immune activation depends on a phage-derived factor. To investigate the mechanism by which the DRT4 system inhibits phage propagation and identify the phage-encoded trigger of DRT4 immunity, we conducted RNA sequencing of DRT4-expressing and D240A/D241A mutant-expressing cells after infection with phage T5. Transcriptomic analysis showed a pronounced downregulation of phage late genes upon DRT4 activation, consistent with impaired phage propagation, while most genes driven by early promoters were upregulated (Supplementary Fig. 7a). We subsequently successfully subcloned 61 genes showing ≥3-fold increased expression‎ into an expression‎ vector and co-transformed them with DRT4 (Supplementary Fig. 7b). Among these, eight genes were found to inhibit host cell growth when co-expressed with DRT4. Notably, seven of these genes exhibited intrinsic toxicity to host cells even when expressed alone (Supplementary Fig. 7c). Neither DRT4 nor ORF55 alone showed toxicity to the host cells, the growth inhibition was specifically observed upon their co-expression‎ (Fig. 5b, c, Supplementary Fig. 7b, c), suggesting that ORF55 activates DRT4-mediated antiviral immunity.

Fig. 5: Phage-encoded protein ORF55 triggers the antiviral immunity of DRT4.

The alternative text for this image may have been generated using AI.

Full size image

a Superimposition of the AlphaFold3-predicted ORF55 structure with that of the PriA NTD-DNA complex. The putative DNA-binding residues of ORF55 are shown as sticks. b The survival status of E. coli cells upon co-expression‎ of DRT4 with ORF55 or the indicated mutants. c Bar graphs from two independent experiments quantifying cell viability in CFUs upon co-expression‎ of ORF55 or the indicated mutant with DRT4 or catalytically inactive mutant. d MST binding assays of ORF55 with four different 30-nt ssDNA homopolymers (poly-dA, poly-dT, poly-dG, and poly-dC). Dissociation constants (Kd ± SD) were calculated from three technical replicates. e SDS-PAGE analysis of DRT4-synthesized ssDNA protected by ORF55. After DNA polymerization reaction, samples were treated with ExoI or nuclease S1 in the presence of increasing concentrations of ORF55. The SDS-PAGE gels were visualized by fluorescence, then stained with Coomassie Blue (CB).

Currently, no functional studies of ORF55 have been reported. The role of ORF55 in the phage T5 life cycle remains unknown. However, the orf55 gene is located within a deletable region of the phage T5 genome, suggesting that it is non-essential for phage T528. Using AlphaFold3-predicted ORF55 structure for a Dali server search, we identified the N-terminal domain (NTD) of Klebsiella pneumoniae PriA helicase (PDB: 6DGD) as the closest structural homolog (Z-score = 9.1, RMSD = 5.8 Å). Notably, PriA-NTD contains a conserved pocket that sequesters the 3’-terminal nucleotide of nascent DNA at replication forks. Based on the PriA NTD-DNA complex structure, we modeled ORF55-DNA interaction and predicted key DNA-binding residues (Tyr16, Tyr18, Tyr62, and Lys63) (Fig. 5a). Alanine substitution of these residues abolished or attenuated the cytotoxicity of ORF55-DRT4 co-expression‎ (Fig. 5c).

To determine whether ORF55 directly interacts with ssDNA, we conducted electrophoretic mobility shift assays (EMSA) by incubating increasing concentrations of ORF55 with four different 30-nt 5’-FAM-labeled ssDNA homopolymers (poly-dA, poly-dT, poly-dG, and poly-dC) (Supplementary Fig. 8a). The results showed that ORF55 directly binds to all four substrates, suggesting a direct interaction between ORF55 and ssDNA. Substitution of the predicted DNA-binding residue Tyr18 with alanine (Y18A) reduced the DNA-binding ability (Supplementary Fig. 8b). Furthermore, we quantitatively assessed the binding affinities of ORF55 to these ssDNA homopolymers using microscale thermophoresis (MST). The results revealed that ORF55 binds to four ssDNA substrates with comparable affinities at tens of nanomolar levels (Fig. 5d). Collectively, these data establish ORF55 as a ssDNA-binding protein that does not require specific nucleotide sequences for interaction.

Although DRT4 exhibits constitutive activity in vitro, purified DRT4 is only covalently linked to short nucleotide oligomers. This suggests that host nucleases may cleave DRT4-generated ssDNA to suppress antiviral immunity under normal conditions. Given ORF55’s ssDNA-binding capability and its role in DRT4 activation, we investigated whether it protects DRT4 ssDNA products from nuclease degradation. We assessed the degradation of DRT4-derived ssDNA by two nucleases: ExoI (a 3′ to 5′ ssDNA-specific exonuclease) and nuclease S1 (an endonuclease targeting ssDNA and RNA). While both nucleases degraded DRT4-synthesized ssDNA, ORF55 inhibited the activity of ExoI in a concentration-dependent manner, whereas the activity of nuclease S1 remained unaffected (Fig. 5e). ORF55 also protected 30-nt ssDNA homopolymers from ExoI degradation (Supplementary Fig. 8c). These findings support a model in which ExoI continuously degrades DRT4-synthesized ssDNA in uninfected cells. Upon phage T5 infection, however, the phage-encoded ORF55 protein competes with ExoI for the binding to the 3′ end of ssDNA, likely leading to ssDNA accumulation and then triggering DRT4-mediated antiviral immunity. Consistent with this model, DRT4 expression‎ in sbcB (encodes ExoI)-knockout cells caused significant growth inhibition, whereas cells expressing a catalytically inactive DRT4 mutant exhibited normal growth (Supplementary Fig. 8d).

DRT4-synthesized ssDNA mediates the interaction between DRT4 and ORF55

Since ORF55 activates the DRT4 system, we asked whether ORF55 can directly bind to the DRT4 protein. To test this, we performed an MBP pull-down assay to examine the potential DRT4-ORF55 interaction. MBP-tagged wild-type DRT4 covalently linked to a short nucleotide oligomer showed weak interaction with ORF55, while wild-type DRT4 extended with dNTPs to generate longer ssDNA exhibited robust binding to ORF55 (Fig. 6a). In contrast, the ORF55 Y18A mutant, which exhibited reduced ssDNA binding ability as determined by EMSA and is not toxic to cells when co-expressed with DRT4, showed diminished association with DRT4. Moreover, the catalytically inactive D240A/D241A mutant, which does not covalently link ssDNA, failed to interact with ORF55. Together, these results indicate that DRT4-synthesized ssDNA mediates the interaction between ORF55 and DRT4, rather than ORF55 directly binding to DRT4 itself, consistent with the DNA-ORF55 interaction feature.

Fig. 6: ssDNA synthesized by DRT4 meditates the DRT4-ORF55 interaction.

The alternative text for this image may have been generated using AI.

Full size image

a In vitro MBP pull-down of ORF55 by MBP-tagged DRT4 variants. MBP-tagged DRT4 proteins (Wild-type and D240A/D241A mutant) preincubated with or without dNTPs were used as baits. ORF55 and its Y18A mutant were used as prey. Two faint bands observed between 70 kDa and 40 kDa likely correspond to degraded DRT4 and the MBP tag. b Western blot (WB) analysis of DRT4 protein levels at different time points after phage infection. E. coli BL21(DE3) cells expressing Flag-tagged DRT4 or its mutant were infected with phage T5 at an MOI of 5, and DRT4 protein levels were monitored over time. Because the DRT4-long nucleic acid complex produced only a weak signal under overall exposure, separately exposed images are presented for DRT4 and the DRT4-long nucleic acid complex. Full exposure results are provided in the source data. Coomassie Brilliant Blue staining under the same conditions is shown below as a loading control.

In vitro protection, MBP pull-down, and cell growth assays indicate that ORF55 binds and protects the DRT4-synthesized ssDNA, thereby likely promoting ssDNA accumulation in vivo. To further validate this, we performed immunoblotting of Flag-tagged DRT4 in cell lysates during phage infection. A high-molecular-weight band appeared after infection and became progressively stronger as the infection advanced (Fig. 6b). This band was not observed in the catalytic-dead DRT4 mutant following phage infection. Notably, the band migrated above 250 kDa, larger than the DRT4-ssDNA covalent complex observed in the in vitro polymerization assay, suggesting that unidentified factors may enhance DRT4 processivity in vivo. Immunoblotting also showed that DRT4 protein levels remained unchanged during phage infection, indicating that the DRT4 system is not regulated at the level of protein expression‎. Collectively, these results indicate that phage infection promotes the accumulation of longer ssDNA species.

Discussion

In this study, we demonstrate that DRT4 assembles as a hexamer to confer anti-phage defense through abortive infection by inducing host cell death, specifically by catalyzing the synthesis of long ssDNA products of random sequences. Moreover, we determined high-resolution cryo-EM structures of DRT4 in multiple functional states, providing mechanistic insights into DRT4-mediated DNA synthesis. More importantly, we identified the phage-encoded DNA-binding protein ORF55 as an activator of DRT4’s immune function. Through comprehensive structural analysis combined with extensive biochemical and functional studies, we propose a mechanistic model for DRT4-mediated anti-phage defense.

In this model, in the absence of phage infection, the reverse transcriptase DRT4 exists as a dimer of trimers architecture that constitutively synthesizes ssDNA through a protein priming mechanism. The newly synthesized ssDNA is subsequently degraded by host 3’-5’ exonucleases such as ExoI, thereby preventing cytotoxicity to the host cells. However, upon phage infection, a phage-encoded DNA-binding protein, ORF55, binds to the 3’ end of the nascent ssDNA, shielding it from exonuclease-mediated degradation and then likely leading to toxic ssDNA accumulation (Fig. 7). Accumulation of ssDNA can act as a critical trigger for bacterial cell death, as exemplified by persistent ssDNA generated during replication stress or thymine deprivation29,30, in which ssDNA accumulation overactivates the SOS response, ultimately leading to cell death. In addition, excess ssDNA can sequester essential single-stranded DNA-binding (SSB) proteins, which are required for both host cell viability and phage replication31. In some cases, ssDNA accumulation may also trigger ion channel activation, leading to abortive infection, as observed in other phage defense systems32. Nevertheless, the precise mechanistic link between ssDNA accumulation and cell death in the DRT4 system requires further investigation.

Fig. 7: Proposed working model of DRT4-mediated anti-phage immunity.

The alternative text for this image may have been generated using AI.

Full size image

Hexameric DRT4 constitutively synthesizes ssDNA of random sequences through protein priming, which is degraded by host 3’−5’ exonucleases, thereby is not toxic to the host cells. Upon phage infection, the phage-encoded ORF55 binds to the 3’ end of the newly synthesized ssDNA, protecting it from host exonuclease degradation. Constitutive cDNA synthesis likely leads to toxic ssDNA accumulation which induces the host cell death.

DRT6 from Sphingopyxis sp, a class 1 UG/Abi RT that shares 23% sequence identity with DRT4, provides robust protection against phage T5 infection16. Liquid culture growth assays showed that, like DRT4, DRT6 mediates phage defense through an abortive infection mechanism (Supplementary Fig. 9a, b). Structural prediction of DRT6 using AlphaFold3 followed by superimposition with DRT4 revealed a high degree of similarity, with an RMSD of 1.9 Å across 326 aligned Cα atoms (Supplementary Fig. 9c). The predicted structure indicates that DRT6 shares the same domain organization as DRT4. To determine whether DRT6 possesses DNA polymerization activity, we expressed and purified MBP-tagged DRT6. SEC analysis revealed that the recombinant protein is eluted at a molecular weight substantially higher than that of monomeric MBP-DRT6, indicating that, like DRT4, DRT6 also adopts an oligomeric state in solution (Supplementary Fig. 9d). Under the same reaction conditions used for DRT4, DRT6 exhibited template-independent DNA synthesis, with the DNA product covalently attached to the protein (Supplementary Fig. 9e, f). Given the structural, enzymatic activity, and antiviral immunity similarities between DRT4 and DRT6, we hypothesized that they share a common activation mechanism in vivo. Indeed, co-expression‎ of DRT6 and ORF55 inhibited cell growth, mirroring the toxicity observed with DRT4. This growth inhibition was abolished or attenuated when either the catalytic motif of DRT6 or the predicted DNA-binding residues of ORF55 were mutated (Supplementary Fig. 9g and h). Furthermore, like DRT4, DRT6 expression‎ was toxic in sbcB-knockout cells (Supplementary Fig. 8d). Collectively, these findings suggest that DRT4 and DRT6 operate through a conserved immune pathway, wherein ORF55 likely triggers toxic ssDNA accumulation, ultimately leading to abortive infection and phage resistance.

Structural comparison demonstrates that DRT4 adopts an overall architecture similar to three previously characterized Abi-RTs despite sharing only 12.4-18.0% sequence identity (Supplementary Fig. 10a). Despite the similar domain composition and overall organization between DRT4 and Abi-RTs, notable structural differences exist. For instance, the RMSD between the DRT4 and AbiK protomers is 3.1 Å (over 303 pairs of Cα atoms), while the RMSD of their RT domains is 1.5 Å (over 163 pairs of Cα atoms). Larger differences are observed in the αRep domain, with an RMSD of 5.0 Å (over 141 pairs of Cα atoms) (Supplementary Fig. 10b). Their dimer of trimers organization also differs. In the DRT4 hexamer, the tilt angle between the upper and lower trimers is smaller ( ~ 10°) compared with ~60° in AbiK. Finally, while AbiK requires nucleic acids to stabilize its hexameric assembly and predominantly exists as a monomer in the absence of nucleic acids18, the DRT4 hexamer is intrinsically more stable and can form a stable hexamer even without ssDNA, as demonstrated by the catalytic-dead mutant, which assembles into a hexamer essentially identical to wild-type DRT4 except for the absence of ssDNA. Furthermore, DRT4 and Abi-RTs display striking variability in the positioning of their priming residues. AbiK and Abi-P2 utilize Tyr44 and Tyr61, respectively, located in the loop of the fingers region. AbiA employs either Tyr298 or Tyr303 within a flexible linker between its RT and αRep domains, while DRT4 uniquely positions Tyr125 in its palm subdomain (Supplementary Fig. 10). Intriguingly, despite these structural differences, all four RTs demonstrate constitutive in vitro activity and can synthesize ssDNA products of random sequences. This observation raises an important question of whether phages might encode functional analogs of ORF55-like DNA-binding proteins to activate these Abi-RTs during infection. Further investigation is warranted to elucidate this.

Recently, two other independent groups and we elucidated the molecular mechanism of DRT9, revealing a conserved noncoding RNA to function as both a structural scaffold and reverse transcription template that orchestrates hexameric complex assembly and mediates RNA-templated DNA homopolymer synthesis31,33,34. Phage infection induces the accumulation of poly-dA single-stranded cDNA in the host cells, which may exert its antiviral immunity through sequestration of the essential phage SSB protein, ultimately disrupting phage replication and establishing abortive infection-mediated population immunity. Notably, one study further identified that ORF55 expression‎ alone is sufficient to trigger cellular toxicity in DRT9-expressing strains33. These findings and our study of DRT4 and DRT6 support a model in which DRT4, DRT6, and DRT9, despite their distinct enzymatic properties, utilize a conserved activation mechanism to confer host immunity.

In summary, our study elucidates the molecular mechanism underlying DRT4 system function, uncovering a distinct pathway for DRT4-mediated antiviral immunity activation. These findings significantly expand our understanding of reverse transcriptase-dependent bacterial defense systems and provide fundamental insights into the ongoing evolutionary arms race between bacteria and phages. Notably, the DRT4 system is widely distributed among clinically relevant pathogens, including Neisseria meningitidis, Campylobacter concisus, and Helicobacter pylori16. Targeted activation of the DRT4 pathway may offer a promising therapeutic strategy for novel antimicrobial development, potentially addressing the growing challenge of antibiotic resistance.

Methods

Strains and phages

The bacterial strains used in this study are listed in Supplementary Data 1. Unless otherwise specified, bacteria were cultured in LB medium (10 g/L Tryptone, 5 g/L Yeast Extract, 10 g/L NaCl, supplemented with appropriate antibiotics) under aerobic conditions (220 rpm/min) or on solid LB medium (supplemented with 1.5% agar). The phage strains, along with their sources, are provided in Supplementary Data 1.

Plasmids

The genes encoding DRT4 proteins from E. coli 12-c8-a (Accession Number: WP_031606642.1) and DRT6 proteins (Accession Number: PHR10625.1), each containing their respective native promoter sequences, were synthesized by Sangon Biotech, which were subsequently cloned into the pET28a or pBAD vector using a homologous recombination method. Detailed information can be found in Supplementary Data 1. All mutants were obtained using site-directed mutagenesis.

The ORF55 gene was amplified via PCR from the T5 phage genome and then cloned into the pET28a expression‎ vector using homologous recombination. The E. coli Exo1 gene was amplified from the E. coli genome and constructed into the pET28a expression‎ vector. The primers used for cloning are listed in Supplementary Data 1.

Protein expression‎ and purification

For the expression‎ and purification of the DRT4. The DNA fragment containing the native promoter and a codon-optimized DRT4 gene was subcloned into the pET28a expression‎ vector containing an N-terminal His-tag and transformed into E. coli BL21(DE3) cells. DRT4 was expressed under its native promoter, eliminating the need for an inducer. The cells were cultured for 6 h at 37 °C with shaking at 220 rpm and then harvested via centrifugation at 4500 × g for 10 min at 4 °C. The collected cells were then resuspended in lysis buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 1 mM PMSF). Cells were lysed by high-pressure homogenization (Union-Biotech) at 600 bars, followed by clarification via centrifugation at 30,000 × g for 60 min at 4 °C. The supernatant was collected, loaded onto Ni-charged Resin FF (GenScript), and washed with 50 mL of wash buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 100 mM imidazole). The target protein was eluted with the elution buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 500 mM imidazole) and further purified on a Superose 6 Increase 10/300 GL size exclusion chromatography column (SEC, Cytiva) pre-equilibrated with buffer containing 20 mM Tris-HCl, pH 8.0, and 150 mM NaCl. DRT4-containing fractions were collected, concentrated, flash-frozen in liquid nitrogen, and stored at −80 °C for future use. All mutants except for R75A and Y125F were generated by site-directed mutagenesis and purified using the procedure described above.

We were unable to obtain soluble protein of mutants R75A and Y125F following the procedure described above. Therefore, wild-type DRT4 and two mutant genes were cloned into the pET28a vector with an N-terminal His-MBP fusion tag. The construct was transformed into BL21(DE3) cells, followed by culture at 37 °C until OD600 reached 0.6-0.8. Protein expression‎ was induced with 0.2 mM isopropyl β-D-thiogalactopyranoside (IPTG) at 16 °C overnight. Cells were harvested by centrifugation at 4500 × g for 10 min at 4 °C, resuspended in lysis buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 1 mM PMSF), and then lysed using high-pressure homogenization. The lysate was clarified by centrifugation at 30,000 × g for 60 min at 4 °C, and the supernatant was collected. The supernatant was loaded onto the Amylose Resin column (NEB) and washed with 50 mL of wash buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl). The target protein was eluted with the elution buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 20 mM Maltose). For the DRT4 R75A variant, the eluate containing the target protein was treated with HRV-3C protease overnight. For the DRT4 Y125F variant, the MBP tag was retained to enhance protein solubility. Meanwhile, the wild-type DRT4 retains the MBP tag as a control to eval‎uate the effect of MBP tag to the protein’s enzymatic activity. Following further purification by SEC, the proteins were concentrated, flash-frozen, and stored at −80 °C for further use.

For the expression‎ and purification of DRT6. The gene was cloned into the pET28a expression‎ vector with an N-terminal His-MBP tag. The plasmid was transformed into BL21(DE3). Cells were grown at 37 °C with shaking at 220 rpm until the OD600 reached 0.6–0.8. Protein expression‎ was then induced by adding IPTG to a final concentration of 0.2 mM. Cells were further cultured overnight at 16 °C. Cells were harvested by centrifugation and resuspended in lysis buffer (20 mM Tris-HCl, pH 8.0, 500 mM NaCl, and 1 mM PMSF). After cell disruption by high-pressure homogenization, the lysate was clarified by centrifugation at 30,000 × g for 60 min at 4 °C. The supernatant was loaded onto the Amylose Resin column, washed with 50 mL of wash buffer (20 mM Tris-HCl, pH 8.0, 500 mM NaCl), and the target protein was eluted using elution buffer (20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 20 mM Maltose). Finally, the protein was further purified by SEC. The DRT6 protein was concentrated, flash-frozen, and stored at −80 °C for subsequent use. The RT catalytically inactive mutant protein (D223A/D224A) was obtained using the same method.

For the expression‎ and purification of ORF55, the DNA sequence encoding ORF55 was amplified from phage T5 and cloned into a pET28a vector containing a His tag at the N-terminus of ORF55 and transformed into E. coli BL21(DE3) cells. Protein expression‎ was induced by the addition of IPTG to a final concentration of 0.2 mM at 16 °C when the OD600 reached 0.6−0.8, and the cells were cultured for another 16 hr. Cells were harvested via centrifugation at 5000 × g for 10 min at 4 °C. The collected cells were resuspended in lysis buffer (20 mM Tris-HCl, pH 8.0, 1 M NaCl, 3 M urea) with 1 mM PMSF and 5 μg/mL DNase1. Lysate was centrifuged at 30,000 × g for 60 min at 4 °C after high-pressure crushing. To remove nucleic acids bound to proteins during the purification process, a partial denaturation-renaturation purification method was employed. In brief, the supernatant was collected and loaded onto Ni-charged Resin FF (GenScript), then a wash buffer with gradually reduced urea concentration (starting from 3 M) and simultaneously gradually increased imidazole concentration was used for protein renaturation and removal of non-target proteins. The target protein was eluted with the elution buffer containing 20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 500 mM imidazole. SEC was employed to further purify the protein. The fractions containing the target protein were pooled, concentrated, flash-frozen, and stored at −80 °C for further use. The ORF55 protein with a C-terminal GFP-His tag was purified following the same protocol and subsequently used for MicroScale Thermophoresis (MST) experiments.

To express and purify E. coli exonuclease ExoI, the gene encoding ExoI was cloned into the pET-28a vector to generate an N-terminal His-tagged fusion protein. The plasmid was transformed into BL21(DE3) cells, which were grown at 37 °C with shaking at 220 rpm until the OD₆₀₀ reached 0.6-0.8. Protein expression‎ was induced by adding IPTG to a final concentration of 0.2 mM, followed by overnight culture at 16 °C. For protein purification, harvested cells were resuspended in a lysis buffer (20 mM Tris-HCl, pH 8.0, 150 m M NaCl), 1 mM PMSF, and 5 μg/mL DNaseI) and disrupted by high-pressure homogenization. Following centrifugation, the supernatant was collected and loaded onto Ni-charged Resin FF (GenScript). The column was washed with 50 mL of wash buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 100 mM imidazole), and then the target protein was eluted with 10 mL of elution buffer (20 mM Tris-HCl, pH 8.0, 500 mM NaCl, and 100 mM imidazole). The protein was further purified by SEC, and the fractions containing the target protein were collected, concentrated, flash-frozen, and stored at -80 °C for future use.

Analytical ultracentrifugation

The sedimentation velocity measurements were carried out using a Beckman Optima Analytical Ultracentrifuge (Beckman Coulter, Brea, CA, USA) with an An60Ti rotor (Beckman Coulter) at 40,000 rpm at 4 °C. Purified proteins were diluted to ~1 mg/mL in 390 μL TN buffer (20 mM Tris-HCl, pH 8.0, and 150 mM NaCl). TN Buffer was used as a reference at a volume of 400 µL. Absorbance at 280 nm was measured every 60 seconds. The sedimentation coefficient was analyzed by the SEDFIT35 and SEDPHAT36 programs.

Cryo-EM sample preparation and imaging

An aliquot of 3.5 μl purified DRT4 at 0.8 mg/ml was applied to glow-discharged Cu R1.2/1.3 holey carbon grid (200 mesh, Quantifoil). After a 20 s incubation, grids were blotted with a force of 0 for 2 s at 4 °C and 100% humidity and plunge-frozen into liquid ethane using Vitrobot Mark IV (FEI Thermo Fisher). To obtain the structure of substrate-bound DRT4 complex, 100 μM of dNTP was added to DRT4 (0.8 mg/ml) in a buffer containing 20 mM Tris-HCl, pH8.0, 150 mM NaCl, and 5 mM MnCl₂. The mixture was incubated at room temperature for 2 min, and then 3.5 μl of the mixture was applied for cryo-EM sample preparation.

All grids were imaged using a CRYO ARM 300 electron microscope (JEOL, Japan) operating at 300 kV equipped with a K3 direct electron detector (Gatan, USA). Cryo-EM images were acquired automatically using Serial-EM software37 with a super-resolution pixel size of 0.475 Å/pixel at defocus values ranging from −0.5 to −2.5 μm at a calibrated magnification of ×50,000. Data were collected at a frame rate of 40 frames per second. The total electron dose was 40 e-/Å2.

Cryo-EM data processing

All cryo-EM data were processed with cryoSPARC38. Recorded movies were subjected to patch motion correction and followed by contrast transfer function (CTF) estimation. Particles were picked using Blob picking and subjected to 2D classification.

For the wild-type DRT4, 653,913 particles from good 2D classification were selected for heterogeneous refinement, requesting four classes. One good class containing 306,468 particles was subjected to further NU-refinement with D3 symmetry, yielding a 2.27 Å reconstruction. For the D240A/D241A mutant, 628,673 particles from good 2D classification were selected for heterogeneous refinement, requesting four classes. One good class was subjected to further NU-refinement, following a 3D classification with three classes. The best class containing 126,192 particles was NU-refined, yielding a 2.7 Å reconstruction. For the substrate-bound DRT4, 478,121 particles from good 2D classification were selected for heterogeneous refinement, requesting four classes. One good class containing 277,716 particles was subjected to further NU-refinement with D3 symmetry, yielding a 2.93 Å cryo-EM map.

Model building and refinement

The initial model of DRT4 was acquired from AlphaFold3 prediction39. The protein model was manually fitted into the density map using UCSF Chimera. The initial model was then manually inspected and rebuilt in Coot40 and refined in Phenix using phenix.real_space_refine41. The nucleotides and DNA were manually built to fit the densities, followed by iteratively refined with phenix.real_space_refine to generate the final model. The quality of all models was validated with MolProbity42. Refinement statistics are summarized in Supplementary Table S1.

Phage propagation

All phages were amplified in liquid culture. The phage host strain E. coli MG1655 was cultured overnight and then inoculated in a volume ratio of 1:100 into a fresh LB medium. When the OD600 reached ~ 0.5, phage stock solutions were added at a volume ratio of 1:10. Cultures were incubated at 37 °C with shaking at 220 rpm for 5 h. The lysate was clarified by centrifugation at 5000 × g for 10 min. The supernatant was filtered through a 0.22 μm filter to remove cellular debris. Phage titers were determined using the double agar overlay plaque assay.

Small-drop plaque assays

Small-drop plaque assays were performed using a double agar overlay protocol43. In brief, for the DRT4 system, the pET28a empty vector, or plasmids containing the DRT4 or mutants with native promoter sequence were transformed into E. coli BL21(DE3). Transformants were cultured overnight on LB agar plates containing 50 μg/mL kanamycin. A single colony was picked and grown overnight in LB medium at 37 °C. Next, 1 mL of the E. coli culture was mixed with 9 mL of top agar (0.5% agar) containing 50 μg/mL kanamycin. The mixture was poured onto LB agar plates containing 50 μg/mL kanamycin. For the DRT6 system, E. coli BL21(DE3) transformed with the pBAD empty vector or plasmids containing wild-type DRT6 or mutant were cultured overnight in LB medium containing 100 μg/mL ampicillin. Subsequently, 1 mL of bacterial culture was mixed with 9 mL of 0.5% top agar-LB (containing 100 μg/mL ampicillin and 0.2% L-arabinose). This mixture was poured onto LB agar plates containing 100 μg/mL ampicillin and 0.2% L-arabinose. After incubating the plates at 37 °C for 1 h, they were dried in a biosafety cabinet for another 10 min. The phage stock solution was subjected to serial ten-fold dilutions in phosphate-buffered saline (PBS). Then, 3 μL of each diluted phage solution was spotted onto the top layer agar. Plates were incubated overnight at 37 °C, phage plaque formation was imaged and analyzed by the BIO-RAD GelDoc Go Gel imaging system.

Phage-infection dynamics in liquid medium

E. coli BL21(DE3) transformed with pBAD vector or recombinant plasmids (DRT4 or mutant) were cultured overnight in LB medium supplemented with 100 μg/mL ampicillin. The bacterial culture was diluted 1:50 in fresh LB medium containing 100 μg/mL ampicillin and 0.2% L-arabinose and grown until OD600 reached ~ 0.2. Then, 180 μL aliquots of the culture were transferred to a 96-well plate containing 20 μL phage T5 solutions at varying concentrations to achieve the specified multiplicity of infection (MOI). Plates were incubated at 37 °C with continuous shaking in BioTek Synergy H1 Plate Reader. The OD600 values were measured every 15 min for 8 h. The experiment was performed in three independent replicates. Liquid infection growth curve assays for DRT6 were performed using the same protocol.

In vitro DNA polymerase activity assays

Under standard conditions, template-independent DNA polymerase activity assays for DRT4 and the mutants was performed at a protein concentration of 3 μM, in the presence of 100 μM dNTP or 100 μM dNTP supplemented with 5 μM fluorescein-dATP (AAT Bioquest, 17036). Reactions were carried out in buffer containing 20 mM Tris-HCl (pH 8.0), 150 mM NaCl, and either 5 mM MgCl₂ or 5 mM MnCl₂, and incubated at 37 °C for 20 min or the indicated times. Reactions were terminated by adding EDTA to a final concentration of 50 mM, followed by protein digestion with Proteinase K at 37 °C for 30 min. Samples were then mixed with 5× DNA loading buffer, resolved on 15% denaturing urea-polyacrylamide gels. We used ssDNA molecules of different lengths as markers. The specific sequences are listed in Supplementary Data 1. Gel was visualized either by Stains-All staining or directly using an Amersham Typhoon biomolecular imager (Cytiva). For in vitro assays of DRT6, reactions were performed with 6 μM protein and incubated for 60 min, with all other conditions identical to those described above.

Visualization of covalent protein-oligonucleotide conjugation

To demonstrate that the synthesized nucleic acid products are covalently conjugated to DRT4/DRT6, both proteins were subjected to reactions following the in vitro polymerase activity assay protocol. The reactions were terminated by adding 5× protein loading buffer. Samples were denatured by heating at 98 °C for 5 min and then separated via 10% SDS-PAGE electrophoresis. The gel was first scanned using a fluorescence imaging system, followed by Coomassie Brilliant Blue staining for subsequent analysis.

DRT4 DNA product sequencing

DRT4 protein and dNTPs were mixed at a 1:200 molar ratio in reaction buffer and incubated for 20 min. Proteinase K was then added, followed by digestion at 37 °C for 30 min. Nucleic acids were subsequently extracted via phenol-chloroform extraction and ethanol precipitation. The purified nucleic acid products were verified by agarose gel electrophoresis and subsequently sent to Igenebook (Wuhan, China) for sequencing library preparation and sequencing. Nucleic acid products were prepared into Illumina-compatible libraries using the VAHTS ssDNA Library Prep Kit (Vazyme, ND620) and sequenced on a NovaSeq 6000 platform (Illumina, San Diego, CA) with the PE150 mode. Trimmomatic (v0.38) was used to filter out low-quality reads44, and Cutadapt (v1.2.1) was applied to remove adapter sequences and demultiplex the raw reads45. The remaining reads were subjected to length distribution and base composition analyses.

Post-infection bacterial survival assay

Overnight cultures of E. coli BL21(DE3) containing the pBAD empty vector or DRT4 system were diluted 1:100 into fresh medium containing 100 µg/mL ampicillin and 0.2% L-arabinose. The cultures were incubated at 37 °C with shaking at 220 rpm until they reached an OD600 ~ 0.5. Phage T5 was then added to the cultures at multiplicities of infection (MOI) of 0, 0.5, and 5. Samples were taken at 0, 15-, 30-, 60-, and 120-min post-infection. Samples from each time point were subjected to serial 10-fold dilutions. From each dilution, 5 µL was spotted onto LB agar plates containing 100 µg/mL kanamycin and 0.2% L-arabinose. After overnight incubation of the plates, colony-forming units (CFUs) were counted. The experiment was performed with three independent replicates.

Live single-cell static fluorescence microscopy

E. coli BL21(DE3) containing DRT4 system was cultured overnight at 37 °C. The cultures were sub-cultured at a 1:100 ratio into 10 mL of fresh medium containing 100 µg/mL antibiotic and 0.2% L-arabinose, then incubated until reaching an OD600 ~ 0.5. The culture was then aliquoted, and phage T5 was added to one portion at an MOI of 2. Samples were collected from both the infected and uninfected groups at 0-, 30-, 60- and 120-min post-infection. All samples were centrifuged at 3000 × g for 5 min, and the resulting pellets were washed twice with ice-cold PBS buffer. The resuspended bacterial samples were stained with propidium iodide (PI) at a final concentration of 50 μg/mL for 30 min at room temperature. Subsequently, 1 μL of the stained bacteria was transferred onto a 0.5% solid agarose pad and imaged using a Dragonfly 200 confocal microscope equipped with a 100× oil immersion objective under consistent optical settings to ensure reproducibility. Image analysis was performed with Imaris Viewer software (v10.1.0).

RNA sequencing and data analysis

E. coli MG1655 harboring DRT4 or its mutants were cultured until the OD600 reached 0.2. Phage T5 was added at an MOI of 0.5. Following incubation at 37 °C for 60 min, the cells were harvested by centrifugation at 5000 × g for 5 min. Total RNA was then extracted from the cell pellets using TRIzol Reagent (ABclonal) according to the manufacturer’s instructions. The RNA concentration and quality were measured on a bioanalyzer using Bioanalyzer RNA chips (Agilent Technologies, Santa Clara, CA, USA). An RNA-Seq library was prepared from approximately 1 μg of total RNA using the VAHTS Universal V8 RNA-seq Library Prep Kit for Illumina according to the manufacturer’s instructions. Following reverse transcription, the synthesized cDNA was fragmented, ligated to adaptors, PCR-amplified, and size-selected using magnetic beads. Sequencing was performed on an Illumina Novaseq platform (Illumina, San Diego, CA, USA). Fastp (v0.23.2) was used for adapter trimming and quality filtering of raw data, producing clean data for subsequent analysis46. Quality control results from all data were summarized using MultiQC (v1.19)47. SortMeRNA (v4.3.3)was utilized to remove rRNA sequences from the samples using default rRNA databases48. Quality-controlled reads were aligned to the reference genome using STAR (v2.7.11a), followed by per-sample alignment quality assessment with Qualimap 2 (v2.3.1)49. Read counts mapped to all annotated genes in the reference genome were quantified using STAR50. Gene expression‎ levels were then represented as Transcripts Per Million (TPM), derived from count normalization. Finally, differential expression‎ analysis between comparison groups was performed using DESeq2(v1.48.1)51, with differentially expressed genes (DEGs) identified under the thresholds of |Fold Change | ≥ 3 and adjusted p-value ≤ 0.05. The analysis results can be found in Supplementary Data 1. All analyses incorporated three biological replicates per experimental group.

Bacterial growth inhibition assays

To identify phage proteins capable of activating the DRT4 system, upregulated phage T5 genes identified in transcriptome sequencing (listed in Table S2) were cloned into the pET28a vector. Subsequently, the recombinant plasmids and the pBAD-DRT4/DRT6 plasmid were co-transformed into E. coli BL21-AI. A single colony was inoculated and cultured overnight in LB medium supplemented with 50 μg/mL kanamycin and 100 μg/mL ampicillin. After serial 10-fold dilution of the bacterial cultures, 5 μL of each dilution was spotted onto LB agar plates containing ampicillin and kanamycin or dual-antibiotic plates supplemented with 0.2% L-arabinose. All plates were incubated overnight at 37 °C. Bacterial growth was imaged and analyzed by BIO-RAD GelDoc Go. For phage proteins that cause bacterial growth inhibition, plasmids were co-transformed with the pBAD empty vector to determine whether the expression‎ of phage protein alone is toxic to the host cells. For the ORF55 protein, the wild-type and mutant were co-transformed with either DRT4/DRT6 or their corresponding mutants into E. coli BL21 AI cells. After performing the experimental procedures as described above, Colony Forming Units (CFU) were counted. The experiment was performed with two replicates.

Electrophoretic mobility shift assay (EMSA)

Four 5’-FAM-labeled 30-nt ssDNA homopolymers (Sangon Biotech, Supplementary Data 1) were incubated with varying concentrations of ORF55 at a final DNA concentration of 0.1 μM in the reaction buffer containing 20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 5 mM MgCl2, and 5 mM DTT at room temperature for 20 min. Loading buffer containing 50% glycerol was then added, and the samples were subjected to electrophoresis on 5% native polyacrylamide gels and then visualized by Amersham Typhoon (Cytiva) biomolecular imager. The experiment was repeated at least three times.

Microscale thermophoresis (MST) analysis

The MST technique was performed to quantify the interaction between C-terminally GFP-tagged ORF55 protein and ssDNA. Reactions were set up by mixing 10 μl of 200 nM ORF55-GFP in a buffer containing 20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 5 mM MgCl₂, 5 mM DTT, 0.05% (vol/vol) Tween-20, and 1 mg/mL bovine serum albumin (BSA) with 10 μl of 30-nt ssDNA solution (Sangon Biotech, Supplementary Data 1) at different concentrations. After incubation at room temperature for 10 min, the mixtures were transferred to the capillaries (NanoTemper Technologies), and thermophoresis was measured at 25 °C by using 20% LED power and medium MST power. The dissociation constant value (Kd) was calculated using MO. Affinity Analysis Software (V2.3) with the Kd fit function, based on three technical replicates.

ORF55-mediated protection of the 3’-end of nucleic acids

For assays with the four ssDNA homopolymers, 0.1 μM ssDNA was incubated in reaction buffer (20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 5 mM MgCl₂, 5 mM DTT) with the indicated concentrations of ORF55 at room temperature for 20 min. 20 nM ExoI was then added, and digestion proceeded for 10 min at room temperature. Reactions were terminated by adding 50 mM EDTA followed by Proteinase K treatment at 37 °C for 20 min. After addition of DNA loading buffer, samples were heated at 98 °C for 2 min and resolved on 15% denaturing urea-PAGE. Gels were visualized using an Amersham Typhoon biomolecular imager (Cytiva).

For the DRT4-synthesized nucleic acid products, DNA products were first synthesized under the in vitro enzymatic activity condition, followed by buffer exchange to remove excess substrates. Subsequently, 1.5 µM of the DRT4-DNA complex in buffer (20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 5 mM MgCl2, and 5 mM DTT) was incubated with different concentrations of ORF55 (0.2-3.2 µM) for 20 min. Each reaction mixture was then split equally into two aliquots. One aliquot received 20 nM ExoI, while the other received 0.5 U of nuclease S1 (Thermo Scientific™) along with its working buffer. Both digestions proceeded at room temperature for 10 min before being stopped by adding protein loading buffer. Samples were heated at 98 °C for 5 min and analyzed by 12% SDS-PAGE electrophoresis. The gel was first scanned using a fluorescence imager and subsequently stained with Coomassie Brilliant Blue.

ΔsbcB E. coli cell growth assay

The pBAD vector expressing DRT4, DRT6, or their corresponding mutants was transformed into E. coli BW25113 and the ΔsbcB strains. After an overnight culture at 37 °C, the bacterial suspension was subjected to a tenfold serial dilution. Subsequently, 5 µL of each dilution was spotted onto LB-agar plates containing ampicillin, either supplemented with or without 0.2% L-arabinose. The plates were incubated overnight, and E. coli growth results were analyzed by the BIO-RAD GelDoc Go Gel imaging system.

MBP pull-down assay

DRT4 and the D240A/D241A mutant were diluted to a final concentration of 6 μM and equally divided into two aliquots. One aliquot was subjected to ssDNA synthesis following the in vitro enzymatic activity assay method, while an equal volume of enzyme reaction buffer was added to the other aliquot. Subsequently, ORF55 or the ORF55 Y18A mutant protein was added to the above mixtures at a final concentration of 6 μM, and the reaction was incubated in reaction buffer (20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 5 mM MgCl2, 5 mM DTT) at room temperature for 20 minutes. Then, 20 μL of Amylose resin (NEB) was added and incubated at room temperature for 30 minutes, followed by centrifugation at 10,000 × g and 4 °C for 10 minutes to remove the supernatant. The Amylose resin was washed three times with pre-chilled TBST buffer. Finally, the bound proteins were eluted with 100 μL of reaction buffer containing 100 mM maltose, then analyzed by 12% SDS-PAGE followed by Coomassie brilliant blue staining.

Western blot

E. coli BL21(DE3) harboring a plasmid with DRT4 or the D240A/D241A mutant containing a N-terminal Flag tag under the control of native promoter were grown overnight and then subcultured at a 1:100 ratio into fresh LB medium (50 μg/mL kanamycin). When the cell density reached an OD600 of ~0.5, phage T5 was added at an MOI of 5. Aliquots of 500 μL were collected at 0-, 2-, 5-, 10-, 15-, 30-, and 60-minutes post-infection. Cells were pelleted by centrifugation at 3000 × g for 5 minutes at 4 °C, and the supernatant was discarded. The pellets were washed twice with pre-chilled PBS, and all samples were normalized to the same OD600 value with PBS. Equal volumes were then centrifuged to harvest the bacterial cells. The resulting pellets were subjected to SDS-PAGE and subsequently transferred to a PVDF membrane. After blocking with skim milk, the membrane was incubated overnight with a mouse anti-Flag primary antibody (Abclonal). Following three washes with TBST, a goat anti-mouse secondary antibody (Abclonal) was applied and incubated for 1 hour at room temperature. After additional washes with TBST to remove nonspecifically bound antibodies, chemiluminescent signals were detected using a QinXiang imaging system.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The atomic coordinates have been deposited in the Protein Data Bank under accession codes 9VDP (DRT4), 9VDV (D240A/D241A mutant), and 9VDO (substrate-bound DRT4). Cryo-EM maps have been deposited in the Electron Microscopy Data Bank under corresponding accession codes EMD-64991, EMD-64992, and EMD-64990. The RNA-seq data used to identify ORF55 has been deposited in the SRA database under corresponding accession code PRJNA1280264. The DRT4 DNA product sequencing data has been deposited in the SRA database under corresponding accession code PRJNA1330739. Source data are provided with this paper.

References

1.  
2.  
3.  
4.  
5.